Iter #50: [tensor([-0.0322,  0.1150, -0.0130,  0.0485, -0.0477,  0.1136, -0.0170, -0.0176,
        -0.0451, -0.0081,  0.0234,  0.0145,  0.0268, -0.0242, -0.0035, -0.0512,
         0.0043,  0.0354, -0.0546, -0.0062,  0.0348,  0.0065,  0.1249, -0.1318,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0676, -0.2682,  0.0056, -0.0250, -0.0169, -0.1586,  0.0343, -0.1199,
        -0.1888,  0.0023, -0.1128,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0401, -0.1508, -0.3307, -0.0533,  0.0130,  0.0268, -0.0195,  0.0139,
        -0.0220, -0.0522,  0.0121, -0.0326, -0.0072,  0.0481,  0.0097, -0.0026,
        -0.0080,  0.0575, -0.0201, -0.0135,  0.0664,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2671,  0.0990, -0.0124,  0.0683, -0.1958, -0.0028,  0.0610, -0.0462,
        -0.0311,  0.0077, -0.0455,  0.0843,  0.0503,  0.0285,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0446,  0.0637, -0.0499,  0.0474, -0.0102,  0.0628,  0.0248,  0.0095,
        -0.0449, -0.0244,  0.0857, -0.1697,  0.0534,  0.0342, -0.0224,  0.0366,
         0.0218, -0.0140,  0.0069, -0.0965,  0.0769,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1890, -0.0007, -0.0117, -0.0779,  0.0671,  0.0440,  0.0608, -0.0157,
         0.0489,  0.0539,  0.1512, -0.2525,  0.0266,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0030, -0.0020, -0.0135,  0.0051, -0.0173, -0.0401, -0.0242, -0.0115,
         0.1032, -0.0042,  0.0022, -0.0004, -0.0741, -0.0522, -0.0171,  0.0130,
         0.0239, -0.0287, -0.0262,  0.0345, -0.0269,  0.0314,  0.0316, -0.0310,
         0.0419,  0.0196, -0.0064,  0.0123, -0.0047,  0.0420, -0.0157,  0.0027,
         0.0878,  0.0013,  0.0058, -0.0225, -0.0089,  0.0105,  0.0132, -0.0079,
        -0.0211, -0.0008, -0.0045, -0.0152, -0.0378], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.5850e-01,  2.0174e-02, -2.9112e-02, -6.6960e-03, -1.7170e-03,
         1.6559e-02, -6.3131e-03,  1.4961e-02,  1.1525e-02,  3.3931e-03,
        -3.5555e-02, -1.2404e-02,  6.5062e-03,  6.4869e-05,  3.8454e-02,
         6.0043e-02, -3.4429e-02,  1.6259e-02, -2.3045e-02, -2.1090e-04,
        -9.3362e-03, -3.9065e-03, -9.3961e-03,  4.6073e-02, -3.8928e-03,
         1.1971e-01,  7.7480e-03, -3.1738e-02, -1.6674e-02,  4.4546e-02,
        -3.4553e-02,  4.3107e-02,  1.2642e-01,  6.9807e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0131, -0.2087, -0.0507, -0.0065, -0.0076, -0.0965, -0.0176, -0.0039,
        -0.0295, -0.0124,  0.0065,  0.0313, -0.0125,  0.0173,  0.0228,  0.0499,
        -0.0320, -0.0022, -0.0333, -0.0103, -0.0197, -0.0500, -0.0585, -0.0157,
         0.0881, -0.1038,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1003, -0.0033,  0.0450,  0.0344, -0.1952,  0.0691, -0.0161,  0.0361,
         0.0078,  0.0152,  0.0181,  0.0405,  0.0549,  0.0836,  0.0277, -0.0069,
        -0.0010, -0.0491,  0.0036,  0.0401,  0.0085,  0.1382,  0.0056,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0405, -0.0336, -0.0581, -0.0002,  0.0026, -0.0071,  0.0841, -0.0250,
        -0.0284, -0.0498, -0.0182,  0.0133, -0.0612, -0.0122, -0.0288, -0.0603,
         0.0395,  0.0078, -0.0095, -0.0121, -0.0007, -0.0206,  0.0081, -0.0123,
         0.0116,  0.0326, -0.0119, -0.0239,  0.0160,  0.0308,  0.0096,  0.0331,
         0.0153,  0.0433, -0.0169,  0.0053,  0.0054, -0.0279, -0.0057, -0.0766,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2313,  0.0034,  0.0082,  0.0106, -0.0133,  0.0029, -0.0889, -0.0264,
        -0.0125, -0.0485,  0.0171, -0.0111, -0.0128, -0.0279, -0.0144, -0.0289,
        -0.0243, -0.0236,  0.0253,  0.0012,  0.0186, -0.0046,  0.0209, -0.0143,
        -0.0207, -0.0477, -0.0006,  0.0157, -0.0061, -0.0807,  0.0141, -0.0653,
         0.0131,  0.0451,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0012,  0.0792, -0.0356, -0.0207, -0.0835, -0.0213, -0.0541, -0.0113,
        -0.0202, -0.0022,  0.0010, -0.0018,  0.0522, -0.0097, -0.0335, -0.0245,
        -0.0229,  0.0033, -0.0092, -0.0083, -0.0412,  0.0008,  0.0239, -0.0063,
        -0.0186, -0.0228, -0.0055, -0.0079, -0.0637, -0.0214, -0.0060, -0.0193,
        -0.0019, -0.0110, -0.0056, -0.0180, -0.0197,  0.0175,  0.0060,  0.0069,
         0.0063, -0.0199,  0.0002, -0.0030, -0.0005, -0.0147, -0.0102,  0.0101,
         0.0186, -0.0070,  0.0456,  0.0442], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0897, -0.0178, -0.0085, -0.1061,  0.0125,  0.0171,  0.0071, -0.0123,
        -0.0218, -0.0441, -0.0033, -0.0029,  0.0309,  0.0227, -0.0050, -0.0025,
         0.0052,  0.0401,  0.0026, -0.0809, -0.0362, -0.0407, -0.0196,  0.0024,
        -0.0105,  0.0037, -0.0617,  0.0301,  0.1137,  0.0272,  0.1215,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0874,  0.0922, -0.1885, -0.0224, -0.0108,  0.0144,  0.0069, -0.0049,
         0.0090, -0.0266, -0.0124,  0.0012,  0.0461, -0.0327,  0.0039, -0.0020,
         0.0668,  0.0273,  0.0027,  0.0215, -0.0233, -0.0046, -0.0517, -0.0794,
        -0.1614,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2059,  0.1113, -0.0600, -0.0774,  0.1674,  0.0050,  0.1160, -0.0343,
        -0.0763,  0.0305,  0.1063,  0.0095,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0647, -0.0024,  0.0267, -0.0239,  0.0347,  0.0193, -0.0153,  0.0247,
         0.0417, -0.0625, -0.0873, -0.0068, -0.0135,  0.0174,  0.1116, -0.0011,
         0.0343,  0.0006, -0.0189,  0.0337, -0.0246, -0.0614,  0.0050,  0.0941,
         0.0322,  0.0350, -0.0245, -0.0379, -0.0441,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1661, -0.0072, -0.0711, -0.0092,  0.0383, -0.0554, -0.0401,  0.3514,
        -0.0370,  0.0230, -0.0610, -0.1401,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1030, -0.0990, -0.0693, -0.0288, -0.0229, -0.0216, -0.0211, -0.1702,
         0.0749, -0.0291, -0.0562,  0.3040,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.7033e-02,  1.9336e-02, -6.2592e-02,  2.9298e-02,  2.2045e-02,
         3.0325e-01, -5.8288e-02,  3.9445e-02, -4.0891e-02, -2.3326e-02,
         9.8520e-03,  5.7652e-02, -3.2897e-02, -3.1695e-03, -1.6351e-05,
        -1.9753e-03, -1.0349e-01, -3.4926e-02, -3.5092e-02,  6.5421e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0363, -0.0048,  0.0147,  0.0084,  0.0127, -0.0530,  0.0706,  0.0731,
         0.0182, -0.0057, -0.0104, -0.0076, -0.0247, -0.0154,  0.0231, -0.0720,
         0.0408, -0.0321, -0.0288, -0.0063, -0.0010, -0.0213,  0.0315,  0.0844,
         0.0253, -0.0328, -0.0092, -0.0017, -0.0106,  0.0367,  0.0058,  0.0038,
        -0.0229, -0.0122, -0.0077,  0.0581,  0.0764,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1115, -0.0148,  0.0972,  0.3016,  0.0342, -0.0721, -0.0936, -0.1775,
        -0.0974,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0072,  0.1184,  0.2884,  0.0516,  0.0013, -0.0470,  0.0309,  0.0077,
        -0.0228, -0.0141, -0.0425,  0.0016,  0.0263, -0.0275,  0.0257,  0.0212,
         0.0138, -0.0036,  0.0141,  0.0290, -0.0599,  0.0118, -0.0018, -0.0284,
         0.0240,  0.0442,  0.0351,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0072,  0.0051, -0.0537, -0.0451,  0.0480,  0.0032, -0.0314,  0.0213,
        -0.0008,  0.0370, -0.1158, -0.0143,  0.0009, -0.0192, -0.0226,  0.0479,
        -0.0125, -0.0106, -0.0551, -0.0180, -0.0039, -0.0365,  0.0039, -0.0857,
        -0.0221,  0.0229,  0.0092,  0.0112, -0.0004, -0.0168, -0.0450,  0.0191,
        -0.0809,  0.0726,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0619, -0.0415, -0.0248, -0.0275,  0.0171, -0.0312, -0.0101,  0.0307,
         0.0030, -0.0100,  0.0756, -0.0248, -0.0144,  0.0108, -0.0026,  0.0128,
         0.0198,  0.0366,  0.0044,  0.0081,  0.0100,  0.0036, -0.0171,  0.0036,
         0.0259, -0.0107,  0.0231, -0.0303, -0.0222, -0.0811, -0.0122,  0.0004,
        -0.0101, -0.0294,  0.0133, -0.0176, -0.0122, -0.0099,  0.0166, -0.0312,
         0.0027, -0.0036,  0.0055,  0.0260, -0.1143,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1124, -0.0592, -0.0662, -0.0736,  0.0512,  0.0108,  0.0403,  0.0053,
         0.0480,  0.0348,  0.0761,  0.0264,  0.1031, -0.0451,  0.1406,  0.1070,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1583, -0.1621, -0.1348, -0.0003,  0.0159,  0.0519, -0.0196, -0.0184,
         0.0117, -0.0094, -0.0097,  0.0141,  0.0273, -0.0048, -0.0150,  0.0162,
        -0.0176, -0.0021, -0.0154,  0.0140, -0.0030, -0.0044, -0.0501,  0.0496,
        -0.0542,  0.0575,  0.0627,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0190,  0.3019,  0.0236, -0.0088,  0.0108, -0.0129, -0.0135, -0.0007,
        -0.0062,  0.0023,  0.0022, -0.0069,  0.0049, -0.0066, -0.0092,  0.0104,
         0.0046, -0.0059, -0.0118, -0.0185,  0.0233, -0.0061,  0.0223,  0.0003,
        -0.0208, -0.0117, -0.0183,  0.0124,  0.0179, -0.0054,  0.0308, -0.0046,
         0.0111,  0.0669,  0.0500,  0.0892, -0.0010,  0.0142,  0.0011,  0.0317,
        -0.0100,  0.0035, -0.0112, -0.0192,  0.0162,  0.0206], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0620,  0.1197,  0.0082, -0.0009, -0.0106, -0.0276, -0.0213, -0.0521,
         0.0212,  0.0215,  0.0129, -0.0657,  0.0056, -0.0482,  0.0377,  0.0345,
        -0.0129,  0.0177,  0.0216, -0.0309,  0.0318,  0.0048, -0.0027, -0.0160,
        -0.0086,  0.0181, -0.0695,  0.2051,  0.0073,  0.0032,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0024,  0.0271, -0.1834, -0.1382, -0.0080,  0.0592,  0.0023,  0.0130,
         0.2176, -0.0142, -0.3346,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1940, -0.0247,  0.0630, -0.0084, -0.0009,  0.0263, -0.0141,  0.0436,
         0.0095, -0.0292, -0.0500, -0.1018, -0.0392,  0.0045, -0.0617, -0.0977,
        -0.0323, -0.0469,  0.0084, -0.1438,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0648, -0.0916,  0.0287,  0.0489,  0.0788, -0.0240, -0.0659, -0.0911,
        -0.0681, -0.1741, -0.2639,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0448, -0.0452, -0.0421, -0.0017,  0.0813,  0.0100, -0.0568, -0.0927,
        -0.0429, -0.1480, -0.0420, -0.0318, -0.0972, -0.0432, -0.0179,  0.0108,
         0.0246, -0.0179, -0.0121, -0.0206, -0.1165,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0231,  0.0078,  0.0767,  0.0542, -0.0282, -0.0247, -0.1611, -0.1871,
        -0.0215, -0.0331, -0.0715,  0.0905,  0.0148,  0.0138,  0.0208,  0.0931,
         0.0339,  0.0441,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0058,  0.0469,  0.0428,  0.0231,  0.0174,  0.0368, -0.0999, -0.0251,
         0.0296, -0.0121,  0.0113, -0.0178,  0.0444,  0.0164, -0.0091, -0.0111,
        -0.0235,  0.0051,  0.0056,  0.0211,  0.0227, -0.0093,  0.0303, -0.0172,
        -0.0005,  0.0611,  0.0171,  0.0166, -0.0136,  0.0223, -0.0002, -0.0052,
         0.0120, -0.0274, -0.0129,  0.0142, -0.0258,  0.1573,  0.0295,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0132,  0.0082,  0.0490,  0.0227,  0.0207,  0.0159, -0.0485,  0.0708,
         0.0348,  0.0456, -0.0303,  0.0274,  0.0512,  0.0324, -0.0263,  0.0206,
         0.0043, -0.0255, -0.0010,  0.0378, -0.0937, -0.0213, -0.0007,  0.0106,
         0.0220,  0.0086, -0.0134, -0.0591,  0.0141,  0.0083, -0.0312,  0.0002,
        -0.0953,  0.0355,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.2492, -0.2134,  0.0499,  0.0149,  0.1362, -0.1116,  0.1842,  0.0242,
         0.0163,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1769,  0.0006,  0.0397, -0.0280, -0.0275, -0.0275, -0.0084, -0.0092,
         0.0382, -0.0355, -0.0079,  0.0014,  0.0235, -0.0136,  0.0573, -0.0017,
        -0.0106,  0.0012, -0.0169, -0.0072,  0.0152, -0.0037, -0.0415,  0.0551,
         0.0390, -0.0184, -0.0601,  0.0129,  0.0236,  0.0081, -0.0169, -0.0098,
        -0.0259, -0.0370,  0.0029,  0.0145,  0.0048, -0.0056,  0.0435, -0.0287,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1405,  0.0630, -0.0053, -0.0207, -0.0018, -0.0043, -0.0279,  0.0284,
        -0.0043,  0.0086,  0.0137,  0.0047, -0.0061,  0.0145,  0.0254, -0.0002,
        -0.0278,  0.0074,  0.0110,  0.0017, -0.0202, -0.0135,  0.0162, -0.0295,
         0.0032, -0.0217, -0.0250, -0.0071,  0.0260,  0.0259,  0.0993, -0.0158,
         0.0299, -0.0129, -0.0067, -0.0117, -0.0086,  0.0043,  0.0049,  0.0201,
        -0.0011, -0.0045,  0.0219,  0.0002, -0.0107,  0.0052, -0.0276,  0.0013,
         0.0095,  0.0110, -0.0551, -0.0318,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0832, -0.0003,  0.0175, -0.0083,  0.0493, -0.0407,  0.0227, -0.0393,
        -0.0097,  0.0413,  0.1955, -0.0169, -0.0497, -0.0857, -0.1007, -0.0792,
        -0.0119, -0.0026,  0.0072, -0.0308, -0.0725,  0.0351,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0011, -0.0680,  0.0088, -0.0004, -0.1013, -0.0588,  0.0944,  0.0626,
        -0.0537, -0.0880, -0.1531,  0.0615,  0.0341,  0.0349,  0.0247,  0.0028,
         0.0196, -0.0245,  0.0265,  0.0619, -0.0193,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0293, -0.0095,  0.0077, -0.0270, -0.0225, -0.0057, -0.0549,  0.0322,
        -0.0093, -0.0127,  0.0004, -0.0059,  0.0249, -0.0097, -0.0085,  0.0054,
         0.0389,  0.0139,  0.0085,  0.0265,  0.0107,  0.0413,  0.0009,  0.0571,
        -0.0010,  0.0183,  0.0111,  0.0079,  0.0127,  0.0112,  0.0278,  0.0015,
        -0.0053,  0.0007, -0.0050,  0.0194,  0.0161, -0.0117,  0.0042, -0.0190,
         0.0301,  0.0020, -0.0700, -0.0448, -0.0149, -0.0122,  0.0099,  0.0011,
         0.0205, -0.0227, -0.0006, -0.0115,  0.0311, -0.0767, -0.0156],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0433, -0.0665, -0.0240, -0.0110, -0.0403, -0.0046, -0.0094, -0.0020,
         0.0159,  0.0279,  0.0232,  0.0307, -0.0196, -0.0123,  0.0030,  0.0218,
         0.0021,  0.0226,  0.0176, -0.0019, -0.0306, -0.0208, -0.0056,  0.0160,
        -0.0139,  0.0051, -0.0378, -0.0509,  0.0108, -0.0103, -0.1267, -0.0473,
        -0.0013, -0.0389,  0.0336, -0.0063, -0.0047,  0.0197, -0.0148, -0.0039,
        -0.0066, -0.0181,  0.0117, -0.0147, -0.0501,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1156,  0.0277,  0.0045,  0.0511,  0.0596,  0.0625, -0.0397,  0.0190,
         0.0785,  0.0016,  0.0274, -0.0257, -0.0115,  0.0739,  0.0130, -0.0119,
        -0.0059,  0.0167,  0.0361,  0.0018, -0.0093,  0.0397,  0.2008,  0.0664,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0931,  0.0336, -0.0222, -0.0396, -0.0527,  0.0358, -0.0640,  0.0787,
        -0.0390, -0.0317,  0.0196, -0.0198,  0.0439, -0.0432, -0.0049, -0.0026,
         0.0658,  0.0343, -0.0054,  0.0108, -0.0191,  0.0070, -0.0066, -0.0003,
         0.0124,  0.0033,  0.0200, -0.0135,  0.0032,  0.0064, -0.0360, -0.1212,
         0.0105,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0533, -0.0250, -0.0220,  0.0477, -0.0105,  0.1943,  0.0242,  0.0144,
        -0.0418,  0.0647, -0.0963,  0.0813, -0.0682,  0.0148,  0.0021, -0.0775,
         0.0888,  0.0227, -0.0504,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0645,  0.2079, -0.1172, -0.0366,  0.0445, -0.0303,  0.0197,  0.0161,
        -0.0351,  0.0485,  0.0158, -0.0080,  0.0523, -0.0161,  0.0086,  0.0124,
         0.0604, -0.0019,  0.0066,  0.0151,  0.0104,  0.0233,  0.0453, -0.0512,
        -0.0524,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0493, -0.0761, -0.0151, -0.0072, -0.0762,  0.0310, -0.0643, -0.2103,
         0.0434, -0.0251, -0.0043,  0.0710, -0.0063,  0.0160,  0.0209,  0.0694,
        -0.2139,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.0046, -0.0658,  0.0221, -0.0877,  0.0506, -0.0155, -0.0385,  0.0600,
        -0.0158, -0.0386,  0.0200, -0.0156, -0.0104, -0.0755, -0.0205,  0.0089,
         0.0033, -0.0090,  0.0300,  0.0193,  0.0184,  0.0129, -0.0062,  0.0120,
         0.0540,  0.0198,  0.0473, -0.0220,  0.0574,  0.0903,  0.0082, -0.0395,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2278, -0.0120,  0.0266, -0.0296, -0.0019,  0.0054, -0.0129, -0.0032,
        -0.0015,  0.0249, -0.0405, -0.0020,  0.0553,  0.0323,  0.0493,  0.1167,
         0.0123, -0.0445,  0.0404,  0.1066, -0.0026, -0.0173, -0.0096, -0.0230,
         0.0029, -0.0170,  0.0317,  0.0502,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0009,  0.0432,  0.0868,  0.1567, -0.0783,  0.0020,  0.0399,  0.0027,
         0.0049,  0.0209, -0.0248,  0.0157, -0.0082, -0.0092, -0.0112, -0.0130,
        -0.0245,  0.0016, -0.0358,  0.0737, -0.0034, -0.0109, -0.0099, -0.0719,
         0.0075,  0.0176, -0.1048, -0.0523, -0.0680,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0398, -0.0432,  0.0312,  0.0212, -0.0067,  0.0182,  0.1381, -0.0034,
         0.0082, -0.0160, -0.0244,  0.0090, -0.0189, -0.0231, -0.1529, -0.0375,
        -0.0037, -0.0028,  0.0082,  0.0515,  0.0573, -0.0030,  0.0196,  0.0197,
         0.0043, -0.0172, -0.0278, -0.0147,  0.0304,  0.0440,  0.0180, -0.0356,
        -0.0504,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1546, -0.0309,  0.0143,  0.0170, -0.0425, -0.0389, -0.0749, -0.0219,
         0.0074, -0.0211,  0.0165, -0.0137,  0.0151,  0.0016,  0.0167, -0.0131,
        -0.0363, -0.0312,  0.0065, -0.1540, -0.0007,  0.1597, -0.0478, -0.0636,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0611, -0.0988, -0.0571, -0.0386, -0.1208,  0.0445, -0.1042, -0.0047,
         0.0814, -0.0546,  0.0192, -0.0095,  0.0196, -0.0115, -0.0594,  0.0126,
         0.0127, -0.0132,  0.0031, -0.0343, -0.0314, -0.1004, -0.0072,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.3774e-01, -2.4540e-02,  1.1458e-02, -1.4363e-02, -2.0535e-02,
        -1.7342e-02, -1.3689e-01,  8.8428e-04, -1.8072e-02, -6.0069e-02,
         5.1188e-02, -2.1045e-02, -6.6361e-05, -5.1665e-03,  1.7871e-02,
        -2.0881e-02, -5.2565e-04, -7.3608e-03, -1.0087e-02,  3.6104e-02,
        -2.8466e-03, -5.6088e-02,  1.9634e-02, -8.0187e-03, -2.5764e-02,
         8.0983e-03,  1.8366e-02, -2.7976e-03,  1.0913e-02, -1.6128e-03,
        -2.8963e-02,  1.8558e-02,  8.5563e-03,  4.5718e-02,  6.6035e-03,
        -1.6711e-02, -4.4985e-03, -8.2747e-02, -2.1318e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0262, -0.0587, -0.0014, -0.0049,  0.0117, -0.0301,  0.0256,  0.0307,
        -0.0167,  0.0044, -0.0048, -0.0034,  0.0063, -0.0383,  0.0124,  0.0130,
         0.0333, -0.0022,  0.0104, -0.0073, -0.0220,  0.0260, -0.0136, -0.1129,
         0.0389, -0.0128,  0.0406,  0.0411, -0.0162,  0.0030,  0.0229,  0.0557,
         0.0316,  0.0672, -0.0566, -0.0032,  0.0031,  0.0007,  0.0016,  0.0062,
        -0.0039,  0.0376, -0.0408,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0163,  0.0058,  0.0026,  0.0114,  0.0186, -0.0021, -0.0186, -0.0080,
         0.0111,  0.0432,  0.0201,  0.0022, -0.0166, -0.0099, -0.0018, -0.0133,
        -0.0042,  0.0124,  0.0068, -0.0197, -0.0061, -0.0125, -0.0201, -0.0118,
         0.0116, -0.0105, -0.0037, -0.0013, -0.0190, -0.0145, -0.0091,  0.0236,
        -0.0049,  0.0159, -0.0363, -0.0419, -0.1369, -0.0238, -0.0083,  0.0293,
         0.0117,  0.0154, -0.0049,  0.0553, -0.0046,  0.0201,  0.0003,  0.0033,
         0.0003,  0.0095,  0.0132,  0.0183, -0.0067,  0.0049,  0.0240, -0.0573,
        -0.0642], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.4166,  0.0636, -0.0690, -0.0833, -0.0158, -0.0202, -0.0071, -0.1477,
        -0.0214,  0.0492, -0.0490,  0.0049,  0.0329, -0.0192,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.9834e-02,  4.1150e-02, -5.1966e-02,  9.5777e-02,  1.7622e-02,
        -4.8257e-04, -7.6405e-02,  4.8361e-03, -1.7490e-02, -1.0397e-02,
        -4.1752e-06,  1.5395e-02,  1.6572e-02,  2.3366e-02, -3.9373e-02,
        -1.2123e-01, -7.6080e-02,  1.1170e-02,  6.4959e-02, -2.2190e-03,
         2.3532e-02,  8.0272e-03, -1.6633e-02,  7.8016e-03, -9.1388e-02,
        -1.6343e-02, -1.5728e-02, -7.6961e-03, -2.5286e-02, -3.1240e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1468,  0.3157,  0.1228, -0.1455, -0.0770,  0.0362,  0.1311, -0.0250,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([-0.0137,  0.0293, -0.0141,  0.0227,  0.0164, -0.0292,  0.0072, -0.0115,
         0.0203,  0.0469, -0.0198, -0.0504,  0.0066, -0.0004, -0.0009, -0.0247,
         0.0372, -0.0053,  0.0454, -0.0146,  0.0175,  0.0162, -0.0317,  0.0310,
        -0.0053, -0.0063, -0.0457,  0.0441, -0.0022, -0.0087, -0.1004, -0.1987,
        -0.0758,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.7304e-02,  2.3727e-02,  1.6393e-02,  3.5158e-04, -1.3794e-02,
        -2.7122e-03,  9.7201e-03, -5.1054e-03,  2.4595e-02,  1.3710e-02,
         6.4358e-03,  1.0340e-02,  1.7226e-02,  7.2682e-03, -4.6579e-03,
         1.3540e-02,  1.5523e-03,  1.1956e-02,  1.5418e-02,  1.2241e-02,
         6.5279e-03,  2.3109e-02, -3.0559e-02,  6.6739e-03, -3.3692e-03,
         3.0744e-02, -6.1514e-05, -1.2295e-02, -9.5164e-03,  2.9943e-03,
        -4.9112e-03,  4.1301e-03,  8.0272e-04, -1.2528e-02, -5.2439e-03,
         1.7585e-02,  9.8034e-03,  1.4643e-02,  1.1608e-02, -6.3475e-03,
        -6.7614e-03, -1.9644e-02, -2.0649e-02,  3.5821e-02,  5.7034e-02,
         4.8225e-02,  1.7275e-02,  2.8308e-02,  1.0599e-02, -1.1840e-02,
         5.4325e-03,  2.9998e-02,  6.8517e-03,  9.5003e-03,  1.2664e-02,
         6.7360e-03,  1.7738e-02, -1.7054e-03,  5.9948e-03, -1.7867e-02,
         6.9390e-03, -2.4103e-03, -2.9582e-02,  2.8982e-02,  1.6399e-03,
        -4.2069e-03, -6.3304e-03,  1.1648e-02,  1.7163e-02,  6.5595e-02,
         3.3588e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0491, -0.0218, -0.0280, -0.0057, -0.0551, -0.0199, -0.0229, -0.0115,
        -0.0121,  0.0286,  0.0051, -0.0025, -0.0281,  0.0480,  0.0478,  0.0220,
         0.0112,  0.0067, -0.0219, -0.0195, -0.0523,  0.0202,  0.0121, -0.0350,
        -0.0005, -0.0163, -0.0554,  0.0077,  0.0059,  0.0718,  0.0023, -0.0083,
         0.0080,  0.0092, -0.0202, -0.0261, -0.0007, -0.0185,  0.0125, -0.0057,
         0.0263,  0.1178,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0483, -0.0639, -0.0343, -0.0626, -0.0548, -0.0066, -0.0593, -0.0366,
         0.0008, -0.0076,  0.0011, -0.0013,  0.0176, -0.0043, -0.0005,  0.0163,
         0.0147, -0.0090,  0.0120, -0.0359, -0.0038, -0.0442,  0.0037, -0.0318,
        -0.0196,  0.0135, -0.0626, -0.0119,  0.0244, -0.0603, -0.0412, -0.0333,
         0.0033,  0.0060,  0.0748, -0.0210, -0.0571,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0259,  0.0373,  0.0900,  0.0775,  0.0452,  0.0837,  0.0747,  0.0751,
        -0.1569,  0.0196,  0.0776, -0.0232, -0.0294,  0.1675, -0.0162,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2149,  0.0551, -0.0343,  0.0253,  0.0186,  0.0011, -0.0015,  0.0147,
         0.0025, -0.0155, -0.0387, -0.0825,  0.0223, -0.0234, -0.0185,  0.0675,
         0.0422,  0.1262,  0.0211,  0.1741,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0096, -0.0243, -0.0473, -0.0512, -0.0026, -0.0222, -0.0167, -0.0103,
        -0.0138,  0.0033, -0.0022, -0.0135,  0.0308,  0.0036,  0.0238,  0.0429,
         0.0337,  0.0105, -0.0567,  0.0353, -0.0242,  0.0025,  0.0403,  0.0184,
         0.0080, -0.0194, -0.0021, -0.0212, -0.0248,  0.0597, -0.0119, -0.0497,
         0.0039, -0.0040,  0.0272,  0.0071,  0.0633, -0.0044, -0.0300, -0.0070,
         0.0393, -0.0524,  0.0246,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0608,  0.0485, -0.0731, -0.0694, -0.0269, -0.0378, -0.0075, -0.0487,
        -0.0283,  0.0294, -0.0390, -0.0134,  0.0122,  0.0085,  0.0207,  0.0249,
        -0.0138, -0.0132,  0.0063, -0.0091,  0.0093,  0.0188,  0.0159, -0.0032,
        -0.0168,  0.0212,  0.0166,  0.0508, -0.0078,  0.0454, -0.0072,  0.0183,
         0.0451,  0.0016,  0.0106,  0.0180, -0.0287,  0.0734,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1395,  0.0246, -0.0234,  0.0642, -0.0335,  0.0024,  0.1329,  0.0013,
         0.0137,  0.0682,  0.0048, -0.0531,  0.0588,  0.0022, -0.0005,  0.0193,
        -0.0535,  0.0111, -0.0995,  0.0264, -0.1592,  0.0079,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1375, -0.0125, -0.0107, -0.0458, -0.0282,  0.0798,  0.0066, -0.0401,
        -0.1076,  0.0123, -0.0066, -0.0124,  0.0048, -0.0066, -0.0007, -0.0002,
         0.0158, -0.0285, -0.0174, -0.0043,  0.0185,  0.0028,  0.0295,  0.0139,
        -0.0282,  0.0145, -0.0175,  0.0090, -0.0062, -0.0217, -0.0196, -0.0432,
        -0.0212, -0.0067, -0.0097,  0.0279,  0.0419, -0.0132, -0.0557, -0.0206,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0208,  0.0011,  0.1037,  0.2066, -0.0342,  0.0285, -0.0016,  0.0132,
         0.0254, -0.0554, -0.0189,  0.0140,  0.0288, -0.0191, -0.0057, -0.0603,
         0.0028,  0.0770, -0.0182,  0.0163,  0.0005, -0.0060, -0.0521,  0.0089,
         0.0350,  0.0019, -0.0253,  0.0118,  0.0166, -0.0068,  0.0092,  0.0743,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.5736e-02, -6.4385e-02, -1.0242e-02,  1.9198e-02, -9.3812e-05,
        -9.2147e-03, -8.4534e-02, -1.3456e-02, -1.4486e-02,  2.2917e-02,
        -2.0633e-03,  1.2333e-02,  3.5481e-02,  6.0767e-03,  5.7461e-02,
         1.5197e-02, -9.1667e-03,  3.4138e-02, -2.9631e-03,  5.3349e-05,
         1.5635e-02,  3.2924e-02,  3.2699e-02,  2.0481e-02,  7.3932e-02,
         1.1144e-01,  1.5369e-02, -3.4186e-03, -2.3440e-03, -1.2699e-02,
         7.2658e-03,  1.9194e-04,  7.8022e-04, -2.0229e-01, -1.9339e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0540,  0.0085,  0.1366,  0.1155,  0.1089, -0.0378,  0.0226,  0.0690,
        -0.0153,  0.0327,  0.0140, -0.0020,  0.0892,  0.0009, -0.0559, -0.0789,
         0.0056, -0.0111,  0.0049,  0.0233, -0.1131,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1536, -0.0734, -0.0635, -0.0362, -0.0122, -0.1052,  0.0067, -0.0540,
         0.0275, -0.0032, -0.0462, -0.0264,  0.0467,  0.1449, -0.0178, -0.0426,
         0.0264, -0.0888, -0.0246,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0332,  0.2343, -0.0760, -0.0102, -0.0038, -0.0056,  0.0283, -0.1064,
         0.0059,  0.1671,  0.0267,  0.0147,  0.0013,  0.0687, -0.0101,  0.0034,
         0.0744,  0.0126,  0.0711, -0.0462,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0206, -0.1451, -0.0865,  0.0362, -0.0544, -0.0152, -0.0063,  0.0591,
         0.0417, -0.0283,  0.0029, -0.0186, -0.0561, -0.0220, -0.0431,  0.0252,
        -0.0210,  0.0507,  0.0430,  0.0206, -0.0183,  0.1666, -0.0188,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0774, -0.0006,  0.0084, -0.0368,  0.0331,  0.0164, -0.0072, -0.0085,
         0.0489, -0.0422, -0.0220,  0.0410,  0.0526,  0.0040,  0.0687, -0.0311,
         0.0010, -0.0110, -0.0524, -0.1456,  0.0045, -0.0491,  0.0984, -0.1017,
        -0.0376,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0008,  0.0447,  0.0172,  0.0298,  0.0023,  0.0063,  0.0241, -0.0452,
        -0.0196,  0.1714,  0.0163, -0.0052,  0.0138, -0.0208, -0.0486, -0.0204,
        -0.0141,  0.0609, -0.0349,  0.0191, -0.0249, -0.0152,  0.0177,  0.0186,
         0.0409, -0.0170, -0.0037, -0.0327,  0.0071,  0.0604,  0.0227, -0.0512,
        -0.0091, -0.0584,  0.0049,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1387, -0.0165,  0.1297,  0.0437,  0.0522, -0.0050,  0.0301,  0.0027,
         0.0152,  0.0647,  0.0514,  0.0169,  0.0745,  0.0277, -0.0704,  0.0172,
         0.0153, -0.0230,  0.0713,  0.0570, -0.0402, -0.0365,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1248,  0.0594,  0.1615,  0.0104, -0.0176, -0.0389,  0.0315, -0.0054,
         0.0284,  0.0087,  0.0635,  0.0098,  0.0782, -0.0611, -0.0882,  0.0131,
        -0.0635,  0.0185, -0.0025, -0.0005,  0.1145,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0568, -0.1443, -0.0055, -0.0390,  0.0077, -0.0588, -0.0192,  0.0130,
        -0.1252, -0.0177, -0.1082, -0.0017,  0.0497, -0.0550, -0.1500,  0.0114,
        -0.0356, -0.0573, -0.0440,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.4027e-02, -1.3422e-01,  2.9670e-02,  3.1861e-02, -6.0482e-03,
         1.4271e-02,  1.3151e-03, -1.9150e-02, -1.7671e-02, -9.3508e-03,
        -1.6682e-02,  9.8369e-03,  5.0986e-03,  1.2002e-02,  9.7671e-03,
        -1.1302e-02, -1.0268e-02,  7.0266e-03, -2.7468e-03,  2.4181e-02,
        -2.3304e-02,  4.0192e-02,  1.4364e-02, -2.0078e-02,  2.6354e-02,
         1.0289e-02,  1.4954e-01,  1.4745e-02,  1.8177e-02,  7.6245e-03,
         1.1511e-04, -1.2440e-02, -3.0497e-03, -4.7930e-03,  5.2457e-04,
         3.3069e-03, -5.4052e-02,  4.0213e-04, -1.1967e-02,  5.6352e-03,
         1.5976e-02, -1.2807e-02,  1.2591e-02,  2.4260e-02,  1.0848e-04,
        -1.2409e-02, -4.4400e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1205,  0.0079, -0.0332, -0.0038,  0.0255,  0.0037, -0.0045,  0.0878,
         0.0010,  0.0067, -0.0184,  0.0074, -0.0152, -0.0555,  0.0070,  0.0444,
        -0.0006,  0.0008, -0.0464, -0.0160, -0.0367, -0.0197,  0.0256, -0.0276,
        -0.0075,  0.0650, -0.0207, -0.0025,  0.0251,  0.0653,  0.0110, -0.0412,
        -0.0685, -0.0772,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1047,  0.0005,  0.0458, -0.0206, -0.0060,  0.0167,  0.0232, -0.0680,
         0.0132,  0.0236,  0.0611,  0.0048, -0.0140, -0.0173,  0.0125, -0.0502,
        -0.0581, -0.0157,  0.0856,  0.0123,  0.0312,  0.0016, -0.0183, -0.0434,
        -0.0648,  0.0037, -0.0205, -0.0351, -0.0018, -0.0262, -0.0289, -0.0212,
         0.0349,  0.0144,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.1079, -0.0449,  0.1966,  0.0008, -0.0026,  0.0088, -0.0613, -0.0568,
        -0.0235, -0.0294, -0.0748, -0.0329,  0.0053, -0.1180, -0.1866, -0.0141,
         0.0179,  0.0176,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0513, -0.0171, -0.0588, -0.0477, -0.0032,  0.0292,  0.0188, -0.0407,
        -0.0042,  0.0065, -0.0126,  0.0956,  0.0613, -0.0634,  0.0708,  0.1563,
        -0.0478, -0.0089, -0.0262,  0.0232,  0.0065,  0.0651,  0.0031, -0.0095,
         0.0102,  0.0499,  0.0120,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1773, -0.0136, -0.0245,  0.0097, -0.0349,  0.0673, -0.0771, -0.0672,
        -0.0316, -0.0317, -0.0090,  0.0085, -0.0228, -0.1028,  0.0064, -0.0872,
        -0.0276, -0.0190, -0.0359,  0.0241, -0.0267,  0.0061, -0.0155,  0.0190,
         0.0142, -0.0401,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0851,  0.0357,  0.0040,  0.0225, -0.0093,  0.0756,  0.0397, -0.0415,
        -0.0443, -0.0036, -0.0983,  0.0039,  0.0967,  0.0339, -0.0480, -0.0238,
         0.0360, -0.0203, -0.0641,  0.0997,  0.1141,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2534,  0.0537, -0.0210, -0.0046,  0.0173, -0.0378,  0.0789, -0.0159,
         0.0590, -0.0787, -0.1046, -0.0871, -0.0991, -0.0448, -0.0442,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0652,  0.0141,  0.0342,  0.0520, -0.1662, -0.0836, -0.1006,  0.0417,
        -0.0528,  0.0142, -0.0769, -0.0131, -0.0333, -0.0416, -0.0611, -0.0655,
        -0.0093,  0.0748,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0378,  0.0220, -0.0099,  0.0251,  0.0808,  0.2164,  0.0690, -0.0199,
         0.0200, -0.0276, -0.0069,  0.0371,  0.0338, -0.0573, -0.0526,  0.0195,
         0.0183,  0.1728,  0.0735,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.2423e-01, -2.7501e-02,  9.0503e-03, -6.3512e-03,  1.2809e-03,
        -1.3923e-02, -5.0714e-02,  2.2053e-02, -9.9247e-03,  7.4203e-04,
         7.8238e-03,  5.6964e-02, -7.2241e-05, -1.0158e-02, -8.1444e-03,
        -5.9639e-02,  1.5884e-02, -3.8319e-03,  4.8086e-02, -1.3723e-03,
        -1.7721e-02,  3.2694e-02, -1.3789e-02,  1.2038e-02, -1.6766e-02,
        -2.5445e-02, -3.8851e-02, -2.5341e-02,  6.1517e-03, -1.9989e-03,
        -4.3945e-03, -3.2696e-02, -2.3583e-02, -1.4741e-02, -1.7692e-02,
         6.0578e-02, -3.1356e-03,  4.2529e-02, -1.7785e-02,  1.8886e-03,
        -2.6060e-02,  1.7881e-02,  6.5234e-02, -3.2604e-03], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0148, -0.0073, -0.0291, -0.0940,  0.0141,  0.0575,  0.0231,  0.0630,
        -0.0179,  0.0275, -0.0053, -0.0499,  0.0304, -0.0081, -0.0149, -0.0347,
         0.0674, -0.0489, -0.0191, -0.0030, -0.0328, -0.0018, -0.0215,  0.0134,
        -0.0244, -0.0173, -0.0638, -0.0201,  0.0133,  0.0177, -0.0096,  0.0087,
        -0.0306,  0.0038, -0.0915,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0422, -0.0984, -0.0618, -0.0014,  0.0810,  0.1038, -0.0393,  0.0203,
        -0.0262, -0.0056,  0.0270, -0.0564, -0.0443, -0.0269,  0.0579,  0.0012,
        -0.0362, -0.0098,  0.0272,  0.0061, -0.0381,  0.0109, -0.0694, -0.1086,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0299, -0.0565,  0.0132, -0.0321,  0.0042, -0.0011,  0.0045, -0.0351,
        -0.0016,  0.0034, -0.0411, -0.0361, -0.0144,  0.0058, -0.0491,  0.0053,
        -0.0180,  0.0297, -0.0080, -0.0925, -0.0082, -0.0598, -0.0179, -0.0421,
         0.0425, -0.0383,  0.0231, -0.0021,  0.0392,  0.0089,  0.0484,  0.0006,
        -0.0170,  0.1704,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0151, -0.2334, -0.1248, -0.0877, -0.0110,  0.0047, -0.0277,  0.0515,
        -0.0149, -0.0435, -0.0024,  0.0139, -0.0014,  0.0139, -0.0036, -0.0113,
        -0.0033, -0.0025, -0.0741, -0.0475, -0.0084, -0.0077,  0.0083,  0.0093,
        -0.0095,  0.0463,  0.0109, -0.0179, -0.0588, -0.0346,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0266, -0.3408, -0.0941, -0.0222, -0.0378, -0.0302, -0.0106, -0.0111,
        -0.0091, -0.1432, -0.1793,  0.0349, -0.0600,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0059,  0.1056, -0.0806, -0.1611,  0.0800,  0.0077, -0.0571, -0.0237,
        -0.0114, -0.0036,  0.0079,  0.0220, -0.0576, -0.0400,  0.0234,  0.0735,
         0.1303,  0.1086,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2707, -0.0160, -0.1908, -0.1181,  0.0145,  0.2036,  0.1473, -0.0390,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0663, -0.0064, -0.0137,  0.0173, -0.0225,  0.0058, -0.0019, -0.0285,
         0.0184, -0.0105, -0.0318,  0.0064, -0.0023, -0.0409, -0.0049, -0.0139,
         0.0120, -0.0303,  0.0045,  0.0043, -0.0771,  0.0127,  0.0224, -0.0379,
         0.0074, -0.0102, -0.0112,  0.0218,  0.0157, -0.0092,  0.0422,  0.0143,
         0.0148,  0.0193,  0.0219,  0.0228, -0.0297,  0.0258,  0.0370, -0.0052,
         0.0044,  0.0014, -0.0098, -0.0161, -0.0766, -0.0277, -0.0627],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2920, -0.0110,  0.1768,  0.0287, -0.1244,  0.0283, -0.0131,  0.0481,
        -0.1098, -0.0682,  0.0995,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2005,  0.0389, -0.0091, -0.0350,  0.0350,  0.0207, -0.0196, -0.0973,
         0.0188, -0.0089, -0.0008,  0.0385, -0.0601, -0.0074,  0.1257, -0.0587,
        -0.1529,  0.0720,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2327,  0.0609, -0.0203, -0.0761, -0.0475, -0.0594,  0.0032,  0.0282,
        -0.0430,  0.0111, -0.0215,  0.0223, -0.1047,  0.0343, -0.2348,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1398,  0.0388, -0.0353, -0.0715, -0.0800,  0.0356, -0.0006, -0.2293,
        -0.0866, -0.0266, -0.0475,  0.0344,  0.0201, -0.1134, -0.0405,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0227, -0.0813,  0.0009, -0.0174, -0.0881,  0.0496,  0.0438,  0.0432,
        -0.0132,  0.0488, -0.0341,  0.0143, -0.0112, -0.0573, -0.0122, -0.0031,
        -0.0266,  0.0662, -0.0539,  0.0037,  0.0291, -0.0591, -0.0166, -0.1202,
         0.0145,  0.0690,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0661,  0.0244, -0.1549, -0.1822,  0.0538, -0.0434,  0.1396, -0.0222,
        -0.0267,  0.0012,  0.0359, -0.0337,  0.0169,  0.1937,  0.0054,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1183, -0.0154,  0.0017,  0.0145, -0.0173, -0.0717,  0.0258, -0.0248,
        -0.0029,  0.0105, -0.0647, -0.0152,  0.0015, -0.0065,  0.0849,  0.0070,
        -0.0406, -0.0313,  0.0156, -0.0564, -0.0137, -0.0579,  0.0347, -0.0090,
        -0.0947,  0.1636,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1083, -0.1069, -0.0138,  0.0860, -0.0093,  0.0071, -0.0228,  0.0517,
         0.0068,  0.0103,  0.0772, -0.0115, -0.0014, -0.0356, -0.0201, -0.0013,
        -0.1176,  0.0253,  0.0064,  0.0303, -0.0404,  0.0196, -0.0158,  0.0188,
        -0.0144,  0.1414,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0322, -0.0158, -0.0341, -0.0418, -0.0279,  0.0066, -0.0249, -0.0118,
        -0.0076, -0.0263, -0.0667, -0.0096, -0.0100,  0.0543, -0.0150,  0.0232,
        -0.0466,  0.0701, -0.0036, -0.0314,  0.0242, -0.0323, -0.0368, -0.0210,
        -0.0224, -0.0354,  0.0036, -0.0186, -0.0133, -0.0094, -0.0354,  0.0298,
         0.0077, -0.0061, -0.0680,  0.0764,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0342, -0.0604, -0.0534, -0.0503,  0.0107, -0.0097,  0.0073,  0.0066,
        -0.0049, -0.0024, -0.0107,  0.0131,  0.0128, -0.0235,  0.0094, -0.0203,
        -0.0272,  0.0084,  0.0209, -0.0707, -0.0791,  0.0005,  0.0338, -0.0223,
        -0.0124, -0.0016, -0.0145,  0.0054, -0.0028,  0.0338, -0.0067,  0.0109,
        -0.0285,  0.0509, -0.2060,  0.0337,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0363,  0.0023,  0.0477,  0.0092, -0.1457,  0.1149,  0.0706,  0.0113,
        -0.1196,  0.0591,  0.0337,  0.0031,  0.0273, -0.0094, -0.0282,  0.0292,
         0.0264,  0.1317, -0.0021, -0.0924,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0454, -0.0646,  0.0049,  0.0118, -0.0276, -0.0310,  0.0158, -0.0115,
         0.0092, -0.0137, -0.0208, -0.0136,  0.0092, -0.0208,  0.0006, -0.0229,
        -0.0072, -0.0192,  0.0255,  0.0112,  0.0064,  0.0184, -0.0004, -0.0195,
        -0.0674, -0.0309, -0.0283,  0.0552, -0.0035,  0.0398, -0.0722, -0.0097,
        -0.0456, -0.0071,  0.0352, -0.0531, -0.1121, -0.0089,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.9761e-02,  1.5039e-02,  2.3663e-02,  1.2104e-02,  1.7413e-03,
         1.9044e-02, -2.1656e-02, -1.3383e-02,  7.1365e-05, -4.6492e-02,
         1.2267e-02, -1.7148e-02, -1.0368e-03, -4.4728e-03,  5.6407e-02,
        -1.1345e-02,  2.0304e-02, -9.4340e-05, -1.1770e-02, -2.6906e-02,
        -3.7134e-02, -1.0028e-02,  1.1637e-02, -6.5752e-02, -2.9422e-02,
        -3.2230e-02, -1.8609e-02,  6.4686e-03,  5.6063e-04,  4.9237e-02,
         2.8706e-03, -2.9959e-02, -4.6544e-03, -2.0386e-02,  4.3589e-03,
         4.0779e-02,  1.7210e-02,  7.2479e-02, -1.6264e-02,  4.1369e-02,
         8.3887e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1499, -0.1455, -0.0341, -0.0547,  0.0736,  0.0205,  0.0015,  0.0032,
        -0.0160,  0.0040,  0.0073,  0.0107,  0.0161, -0.1179,  0.0603, -0.0232,
         0.0073,  0.0029,  0.0195,  0.0009,  0.0597, -0.1714,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0379,  0.0246,  0.0167, -0.0176,  0.0116, -0.0130, -0.0319, -0.0379,
         0.0010, -0.0108, -0.0718, -0.0018,  0.0170,  0.0011,  0.0336,  0.0086,
        -0.0024, -0.0273,  0.0273,  0.0236,  0.0370, -0.0099,  0.0243,  0.0069,
         0.0270, -0.0217,  0.0015, -0.0271, -0.0632,  0.0166,  0.0129,  0.0295,
        -0.0205,  0.1982, -0.0862,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0571, -0.0454, -0.0595, -0.1127, -0.0017,  0.0813,  0.0550,  0.0636,
         0.0208, -0.0513, -0.0183, -0.0009, -0.0386,  0.0030,  0.0245, -0.0603,
        -0.0547,  0.0306, -0.0107,  0.0233, -0.0077, -0.0205, -0.0312,  0.0223,
         0.0025, -0.0334, -0.0041,  0.0330, -0.0323,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1522,  0.0244,  0.0384,  0.0303,  0.0062,  0.0006,  0.0140, -0.0158,
        -0.0269,  0.1009, -0.0312, -0.0270, -0.0064, -0.0059,  0.0356,  0.0061,
        -0.0198, -0.0228, -0.0534, -0.0111,  0.0238, -0.1220,  0.0499, -0.1192,
         0.0559,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1983,  0.0510, -0.0266, -0.1068,  0.0170,  0.0350,  0.0082,  0.0021,
        -0.0105,  0.0475,  0.0298,  0.0181,  0.0087,  0.0909,  0.0156,  0.0119,
        -0.0017, -0.0206,  0.0092,  0.0711, -0.0545, -0.0851, -0.0799,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1561,  0.0256,  0.0009, -0.0328,  0.0005,  0.0817, -0.0579,  0.0233,
         0.0071,  0.0087,  0.0235,  0.0070,  0.0035, -0.0044, -0.0014, -0.0353,
         0.0060, -0.0046,  0.0022,  0.0009,  0.0167, -0.0267,  0.0371,  0.0726,
         0.0335,  0.0170,  0.0176,  0.0004,  0.0116, -0.0011,  0.0054, -0.0056,
        -0.0298, -0.0212, -0.0485, -0.0586, -0.0267, -0.0373, -0.0493,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.3506e-01, -3.7124e-01, -1.8243e-01, -7.3169e-02,  9.7775e-03,
        -5.6125e-02,  3.7192e-03, -1.2104e-02,  1.1331e-01, -1.0574e-02,
         3.2423e-02, -6.7650e-05,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.1459, -0.0314, -0.0147,  0.0208,  0.0088, -0.0426, -0.0175, -0.0125,
        -0.0290,  0.0008, -0.0197,  0.0041, -0.0046, -0.0092, -0.0016, -0.0079,
         0.0094, -0.0097, -0.0038,  0.0022,  0.0045,  0.0008,  0.0081,  0.0125,
         0.0211, -0.0365,  0.0858, -0.0977, -0.0058,  0.0386, -0.0300, -0.0276,
        -0.0193, -0.0101, -0.0366, -0.0613,  0.0011,  0.0010,  0.0240,  0.0034,
        -0.0013, -0.0119, -0.0267,  0.0085,  0.0297,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2067,  0.0164,  0.0005,  0.0069,  0.2660, -0.0904, -0.1075,  0.0628,
        -0.0044,  0.1827,  0.0186, -0.0371,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.4135, -0.0514,  0.1083, -0.0970,  0.0981, -0.0405, -0.1611, -0.0303,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0024,  0.0293, -0.0095,  0.0249,  0.0123,  0.0004, -0.0072, -0.0141,
         0.0042, -0.0029,  0.0131,  0.0773, -0.0147, -0.0109, -0.0055,  0.0189,
        -0.0103, -0.0424, -0.0073,  0.0051,  0.0023,  0.0248,  0.0074, -0.0357,
        -0.0059,  0.0081,  0.0045, -0.0037,  0.0009,  0.0120,  0.0284,  0.0222,
         0.0135,  0.0219,  0.0008, -0.0107,  0.0228, -0.0033,  0.0067,  0.0154,
         0.0003,  0.0059,  0.0160, -0.0212, -0.0030, -0.0930, -0.0626, -0.0162,
        -0.0275, -0.0178, -0.0011,  0.0235,  0.0035,  0.0156,  0.0067,  0.0083,
        -0.0090, -0.0080,  0.0084, -0.0178, -0.0710], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0993, -0.0419, -0.0909, -0.0434,  0.0338,  0.0212, -0.0318, -0.0338,
        -0.0188, -0.0253, -0.0236, -0.0127, -0.0126, -0.0213,  0.0059, -0.0150,
         0.0107, -0.0272, -0.0133, -0.0143,  0.0357,  0.0181,  0.0103, -0.0731,
        -0.0195,  0.0007,  0.0131, -0.0140, -0.0225,  0.0318,  0.0578,  0.0768,
         0.0297,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1870, -0.0271,  0.0011, -0.0202, -0.0396,  0.0318, -0.0135,  0.0008,
         0.0232, -0.0713,  0.0165, -0.0010, -0.0426, -0.0167,  0.0129,  0.1240,
        -0.0066, -0.0016, -0.0142, -0.0268, -0.1232,  0.0502,  0.0176, -0.1304,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0606, -0.0689, -0.0825,  0.0484, -0.0147, -0.0205, -0.0355,  0.0298,
         0.0039, -0.0207,  0.0424,  0.0283, -0.0542, -0.0213,  0.0075,  0.0016,
         0.0086,  0.0454,  0.0211,  0.0259,  0.0765,  0.1057, -0.0414,  0.1346,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2031,  0.0105, -0.0671, -0.0224, -0.0202,  0.0134, -0.1303, -0.0227,
        -0.0181, -0.0473,  0.0122, -0.0064, -0.0082, -0.0058, -0.0018,  0.0079,
        -0.0193, -0.0298, -0.0381, -0.0297, -0.0153,  0.0013, -0.0165,  0.0128,
         0.0177,  0.0291, -0.0166,  0.0103, -0.0150, -0.0041,  0.0052, -0.0496,
        -0.0803,  0.0119,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0136, -0.1036, -0.0612,  0.0340, -0.0469, -0.0803, -0.0414,  0.0686,
         0.0287,  0.0039, -0.0013, -0.0123,  0.0312,  0.0841, -0.0424, -0.0497,
        -0.0069, -0.0127,  0.0074, -0.0133, -0.0345,  0.0069, -0.0721,  0.0006,
         0.1424,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0530,  0.1212, -0.0032,  0.0181,  0.0243, -0.0133, -0.0404,  0.0215,
         0.0468, -0.0233, -0.2751,  0.0168, -0.0275, -0.0484,  0.1811, -0.0860,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1688,  0.0544, -0.0018,  0.0867, -0.0084,  0.0228, -0.0201, -0.0088,
        -0.0947, -0.0273,  0.0104,  0.0215,  0.0339,  0.0410,  0.0713,  0.0363,
         0.0360, -0.0148,  0.0008, -0.0127, -0.0421, -0.0237,  0.1617,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0626, -0.3582, -0.0934, -0.0015, -0.0385, -0.0538, -0.0032, -0.0322,
        -0.0886,  0.0299,  0.0394, -0.0384, -0.0419, -0.1184,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-2.6694e-01,  8.8169e-04, -2.2788e-02, -4.9600e-03, -1.4619e-02,
         7.9232e-02, -1.4147e-04,  2.4019e-02, -4.5156e-03, -1.5208e-02,
        -3.5025e-02, -2.9002e-04,  1.1186e-02, -1.7679e-02,  3.0108e-02,
        -8.0641e-02, -4.2726e-02,  1.5601e-02,  8.3521e-02, -8.3559e-02,
        -5.6651e-02, -2.5316e-02,  3.4324e-02,  5.0069e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2420,  0.2033,  0.0174, -0.1065,  0.0930, -0.1326, -0.0055, -0.0421,
        -0.0682, -0.0288,  0.0605,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0989, -0.3113,  0.0422,  0.0595, -0.0532, -0.0269, -0.0453, -0.0323,
        -0.0274, -0.0206, -0.0076, -0.0263, -0.0295, -0.0707, -0.0033, -0.0118,
        -0.0235, -0.0272, -0.0067, -0.0156, -0.0604,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0029, -0.0303, -0.1814,  0.0429,  0.0124, -0.0941, -0.0862, -0.1034,
         0.0250,  0.0211, -0.0476, -0.1809, -0.0641,  0.1077,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0729,  0.0303, -0.0345, -0.0806,  0.0897,  0.0458,  0.0189, -0.1039,
        -0.0042,  0.0697, -0.0075, -0.0904,  0.0542, -0.0180, -0.0039,  0.0191,
         0.0105, -0.0681, -0.0472,  0.0142, -0.1161,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1932, -0.0179, -0.0127, -0.0036,  0.0209, -0.0340,  0.0580, -0.0295,
        -0.2757, -0.0614,  0.0117, -0.1891, -0.0923,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1473,  0.0148, -0.0150,  0.0275,  0.0195,  0.0080, -0.0052,  0.0021,
         0.0980, -0.0286,  0.0012,  0.0228,  0.0049,  0.0102,  0.0073, -0.0171,
        -0.0107,  0.0197, -0.0317,  0.0069, -0.0031,  0.0471, -0.0660, -0.0508,
        -0.0259, -0.0135, -0.0174, -0.0096, -0.0298, -0.0440, -0.0015, -0.0069,
        -0.0221,  0.0014,  0.0057,  0.0052, -0.0098, -0.0036,  0.0006,  0.0148,
        -0.0063,  0.0002,  0.0031,  0.0530, -0.0599], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0531,  0.0117, -0.0198, -0.0188, -0.0181, -0.0057, -0.0135,  0.0107,
        -0.0251, -0.0187, -0.0333, -0.0023, -0.0321, -0.0004, -0.0487, -0.0030,
         0.0235,  0.0075, -0.0074,  0.0104, -0.0454, -0.0326,  0.0102, -0.0384,
        -0.0067, -0.0377, -0.0667,  0.0337, -0.0568, -0.0687,  0.0681, -0.0211,
         0.0906, -0.0597,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0605, -0.0958, -0.0274, -0.0529, -0.0135, -0.0404, -0.0202, -0.0313,
        -0.0224, -0.0359, -0.0233,  0.0070,  0.0473, -0.0921,  0.0146, -0.0016,
        -0.0669,  0.0307,  0.0182, -0.0031, -0.0847, -0.0442, -0.0183,  0.0078,
         0.0225,  0.1174,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0335, -0.1378,  0.0890,  0.0171, -0.1948, -0.0404, -0.1189,  0.0218,
         0.0176, -0.0167,  0.0078,  0.0060, -0.0153, -0.0584, -0.0560, -0.0127,
         0.0506, -0.0345,  0.0094, -0.0262, -0.0037, -0.0200,  0.0118,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.8609e-01, -4.6407e-02, -1.7081e-02,  3.7591e-02, -1.2796e-02,
         3.7466e-03, -8.4674e-03, -3.8438e-03, -1.3840e-02,  4.9973e-03,
        -1.7144e-02, -1.8082e-02,  1.7600e-02, -7.1701e-03, -7.0722e-03,
        -6.8939e-03, -2.5032e-02,  1.5448e-04,  7.0538e-04,  1.4888e-04,
         1.4208e-02,  9.1543e-03, -2.2070e-02, -6.4080e-02,  2.8430e-02,
        -4.5540e-02,  1.3849e-02, -4.2250e-04,  4.7007e-02, -2.6959e-02,
         2.6318e-02, -4.0242e-03,  7.0471e-02,  2.2558e-02,  2.7379e-02,
         6.0773e-02, -5.5696e-03,  6.2157e-02,  7.1096e-04, -1.3460e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0486, -0.0112,  0.0291, -0.0163, -0.0317,  0.0126, -0.0958,  0.0019,
        -0.0566, -0.0138,  0.0079,  0.0030, -0.0430, -0.0114, -0.0146, -0.0155,
        -0.0318,  0.0546,  0.0578,  0.0225,  0.0477, -0.0031,  0.0004, -0.0222,
         0.0019, -0.0294,  0.0246, -0.0060, -0.0272,  0.0525,  0.0665,  0.1110,
         0.0265, -0.0015,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 7.4865e-02,  2.3778e-02, -3.3591e-02, -1.7412e-02, -3.6459e-02,
        -3.6406e-02,  6.2351e-03, -1.6602e-02, -2.0752e-02,  2.8060e-04,
         4.6112e-03, -2.4151e-03, -1.9767e-02,  3.3645e-02, -6.1575e-02,
        -1.0764e-01, -2.2325e-03,  1.4324e-02,  4.0024e-03,  1.8319e-02,
        -3.4165e-02,  9.3308e-03, -1.0678e-01, -1.8929e-02,  3.5069e-05,
        -3.1005e-02, -5.5207e-04,  2.5836e-03,  7.9431e-04,  1.2480e-03,
         3.9155e-02,  6.5647e-03, -1.2603e-02, -1.1077e-02, -1.7079e-02,
        -3.1806e-03,  1.9052e-03, -3.1888e-03,  1.6769e-02, -3.5883e-04,
        -1.7798e-03,  1.6376e-02,  6.4566e-03, -2.8013e-02, -8.4722e-03,
         3.2153e-03, -2.1753e-02,  5.4786e-03, -5.7811e-03, -4.0673e-02,
        -6.5996e-03, -3.1843e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0264, -0.0485, -0.0192, -0.0358, -0.0022,  0.0233,  0.0259, -0.0400,
         0.0135, -0.0244,  0.0069,  0.0443, -0.0146,  0.0035, -0.0454,  0.0101,
        -0.0824, -0.0059, -0.0147,  0.0301,  0.0096,  0.0306,  0.0169, -0.0292,
        -0.0617, -0.1113, -0.0564,  0.0144,  0.0771, -0.0743, -0.0015,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0294,  0.0482, -0.0235, -0.0040,  0.0044,  0.0577,  0.0198,  0.0812,
        -0.0149,  0.0490, -0.0304, -0.0338,  0.1130, -0.0986,  0.0307,  0.0025,
         0.1034,  0.0060,  0.0199,  0.0161,  0.0436, -0.0395,  0.0274, -0.0897,
         0.0135,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0438,  0.1170, -0.0537, -0.0095, -0.0562,  0.0253,  0.0387, -0.2026,
         0.0771,  0.1696, -0.1276,  0.0788,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0202,  0.0442, -0.1026,  0.0365, -0.0127, -0.0402,  0.0302,  0.0153,
        -0.0247,  0.1295,  0.0178,  0.0198,  0.0263, -0.0180,  0.0536,  0.0105,
        -0.0524, -0.0153, -0.0350, -0.0366, -0.0141,  0.0970,  0.0026, -0.0251,
        -0.0480,  0.0041, -0.0079,  0.0173, -0.0425,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1865, -0.0154,  0.0042, -0.1757,  0.0757, -0.0960,  0.0056,  0.0458,
         0.0890,  0.0875, -0.1816,  0.0370,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0224, -0.1616, -0.1602,  0.0189, -0.1629, -0.0453,  0.0188, -0.0977,
         0.0128, -0.0384,  0.0610,  0.2000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1235,  0.0322, -0.1717, -0.0722, -0.0100,  0.1230, -0.0049,  0.0500,
         0.0211, -0.0531, -0.0336, -0.0046,  0.0061,  0.0268, -0.0102, -0.0235,
         0.0366, -0.0496,  0.0089,  0.1381,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0468, -0.0168, -0.0214, -0.0104, -0.0039, -0.0202,  0.0493,  0.0230,
        -0.0064,  0.0513,  0.0231,  0.0087,  0.0088,  0.0563,  0.0863,  0.0301,
         0.0046,  0.0593,  0.0570, -0.0201,  0.0279,  0.0140,  0.0432,  0.0044,
        -0.0221,  0.0082, -0.0443, -0.0146,  0.0139,  0.0490,  0.0241, -0.0068,
        -0.0209, -0.0082,  0.0624,  0.0077,  0.0243,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2025, -0.2230, -0.0709, -0.0873, -0.0161,  0.0706,  0.1894,  0.0497,
         0.0903,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0270,  0.1491,  0.0954,  0.0081,  0.0135, -0.0842,  0.0205,  0.0239,
        -0.0035,  0.0231, -0.0129, -0.0050, -0.0078,  0.0081,  0.0795,  0.0158,
         0.0048, -0.0624,  0.0409, -0.0212, -0.0721, -0.0028, -0.0578,  0.0063,
        -0.0008, -0.0887,  0.0649,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0484,  0.0220,  0.0223, -0.0008,  0.0114, -0.0221,  0.0108, -0.0042,
         0.0056,  0.0957,  0.0051, -0.0092,  0.0082,  0.0606,  0.0008, -0.0238,
        -0.0022,  0.0134, -0.0692, -0.0434,  0.0975,  0.0112, -0.0194, -0.0204,
         0.0171,  0.0803,  0.0045,  0.0009, -0.0725, -0.0079, -0.0049, -0.0257,
        -0.0791, -0.0793,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0617,  0.0126,  0.0254, -0.0282, -0.0191,  0.0052, -0.0060, -0.0235,
        -0.0133,  0.0297, -0.0366,  0.0091,  0.0151,  0.0030,  0.0139,  0.0426,
        -0.0100, -0.0121,  0.0013,  0.0017,  0.0246,  0.0100, -0.0144, -0.0151,
        -0.0242,  0.0270, -0.0107,  0.0723,  0.0121,  0.0103,  0.0259,  0.0624,
        -0.0145,  0.0187, -0.0348, -0.0055, -0.0145, -0.0211,  0.0210,  0.0289,
        -0.0077,  0.0054,  0.0330,  0.0166, -0.0987,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.1432e-02, -7.2665e-03, -1.9786e-01, -4.7967e-02,  2.1442e-02,
         5.6823e-03,  1.6258e-01, -2.2481e-02,  2.7545e-01,  4.4694e-02,
        -9.4345e-02,  2.3991e-02,  5.2186e-03,  9.7191e-06, -6.0099e-02,
        -9.4776e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5913e-01, -2.4654e-02,  1.0717e-02, -9.1818e-02, -1.1663e-01,
        -2.9686e-03,  7.5199e-03,  5.2202e-02,  1.9793e-02, -5.9270e-03,
        -2.3830e-04, -7.7854e-03, -1.3763e-02, -1.1909e-03,  4.4316e-03,
         5.3032e-04, -1.4615e-02,  2.3724e-02, -2.5098e-02, -2.3763e-03,
         1.1700e-02,  1.6980e-03, -1.3139e-02, -1.8491e-02, -2.4166e-01,
         5.5096e-02,  7.3098e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0063,  0.0601, -0.0105,  0.0125, -0.0135, -0.0258,  0.0076, -0.0080,
        -0.0081, -0.0585, -0.0023, -0.0105,  0.0167, -0.0027, -0.0298,  0.0180,
         0.0307, -0.0754, -0.0001, -0.1121, -0.0002,  0.0184, -0.0302,  0.0493,
         0.0005,  0.0079, -0.0181, -0.0005,  0.0015, -0.0317, -0.0551,  0.0050,
        -0.0279,  0.0569,  0.0349, -0.0197,  0.0029, -0.0121,  0.0107, -0.0022,
         0.0031, -0.0241, -0.0157,  0.0118,  0.0170, -0.0335], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0343,  0.1148,  0.0096,  0.0023,  0.0038,  0.0386,  0.0013,  0.0176,
        -0.0024,  0.0673,  0.0044, -0.0111,  0.0226, -0.0080,  0.0711, -0.0215,
        -0.0078,  0.0210,  0.0430, -0.0959,  0.0007,  0.0060,  0.0035,  0.0041,
         0.0069, -0.0090, -0.0132, -0.0526,  0.2368,  0.0689,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1907, -0.0315, -0.1823, -0.0282, -0.0925,  0.0003, -0.0075, -0.0921,
        -0.2291, -0.1388,  0.0070,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0741,  0.0561, -0.0348, -0.0670,  0.0775,  0.0267,  0.0328, -0.0056,
        -0.0032,  0.0192, -0.0115, -0.0210, -0.1109, -0.0554,  0.0700,  0.0740,
         0.0165, -0.0129,  0.0781,  0.1529,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1189, -0.0160, -0.0602,  0.0429, -0.1524, -0.1390, -0.0401,  0.2182,
        -0.0229, -0.1065, -0.0829,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1411, -0.0359, -0.0830, -0.0141, -0.0352, -0.0654, -0.0342,  0.0442,
        -0.0537, -0.0746, -0.0046, -0.0486, -0.0156, -0.0242,  0.0130, -0.0268,
        -0.0603,  0.0058, -0.0432, -0.0828,  0.0936,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0622, -0.1116, -0.0415, -0.0944, -0.0132,  0.0499, -0.0108,  0.0500,
         0.1512, -0.0664, -0.0025,  0.0915,  0.0526, -0.0124,  0.0295, -0.0468,
         0.0098, -0.1037,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.4977e-03, -1.2913e-02, -2.0357e-02,  1.0517e-04,  2.6991e-02,
         3.5393e-02,  1.8441e-02, -4.7204e-02, -7.4925e-03, -1.8132e-02,
         2.5010e-02,  6.7264e-02, -2.9298e-03, -1.2653e-02, -2.4762e-02,
         9.2149e-02, -1.5480e-02, -1.6767e-02,  2.5854e-03, -2.8471e-03,
         4.1238e-03, -7.4543e-03, -5.7434e-03,  2.5455e-02,  1.1534e-02,
        -1.0943e-03, -3.6239e-02, -1.3761e-02,  2.9499e-03, -1.3731e-02,
         7.2601e-02, -4.5041e-03, -4.8774e-02, -1.8951e-02,  2.3511e-02,
        -4.1521e-03, -7.3934e-03,  1.7880e-01, -6.4259e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.6971e-02, -8.1773e-02,  1.2094e-02, -7.2876e-03,  7.5449e-03,
         3.0172e-02, -1.6771e-02, -2.0751e-02,  1.9618e-02, -7.3466e-03,
        -9.2869e-03, -2.7934e-02,  1.3365e-02,  1.8694e-02,  1.7010e-02,
         1.1271e-02, -7.2502e-03, -1.1331e-04, -2.6012e-02,  1.2207e-02,
        -5.1183e-02, -3.9848e-02, -2.0978e-02, -3.3727e-03, -3.0850e-02,
        -2.2571e-02,  2.4649e-03, -3.1992e-02,  6.1748e-03,  4.8129e-02,
        -4.6418e-02, -8.2112e-03, -1.6532e-01, -1.1901e-01,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.5634,  0.0229, -0.0386, -0.0240, -0.0208, -0.1535, -0.0595, -0.0497,
         0.0675,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0250, -0.0039,  0.0434, -0.0086, -0.0450,  0.0066, -0.0052, -0.0144,
         0.0309, -0.0416, -0.0054,  0.0126, -0.0127,  0.0013,  0.0680, -0.0087,
         0.0014, -0.0133, -0.0285, -0.0147, -0.0226,  0.0014,  0.0085, -0.0086,
        -0.0171, -0.0554, -0.0330,  0.0418, -0.0050,  0.0106,  0.0288, -0.0275,
         0.0057, -0.0543,  0.0860, -0.0210, -0.0262, -0.0144,  0.1072,  0.0337,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0577,  0.0168,  0.0060,  0.0107,  0.0195, -0.0151, -0.0045,  0.0608,
         0.0114, -0.0326, -0.0073,  0.0041,  0.0002,  0.0134,  0.0338,  0.0225,
         0.0133,  0.0720, -0.0017, -0.0027, -0.0006, -0.0478, -0.0514,  0.0455,
        -0.0478,  0.0096, -0.0096,  0.0150, -0.0132, -0.0303,  0.0008, -0.0088,
         0.0065,  0.0064,  0.0040, -0.0473, -0.0035, -0.0144,  0.0064,  0.0017,
        -0.0233, -0.0138,  0.0067,  0.0128,  0.0055, -0.0095, -0.0045,  0.0013,
        -0.0094,  0.0427, -0.0342, -0.0596,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0431,  0.0538,  0.1718,  0.0607,  0.0566,  0.0099,  0.0243, -0.0988,
        -0.0205, -0.0351,  0.0670, -0.0046,  0.0062,  0.0522, -0.0055,  0.0378,
         0.0375, -0.0264, -0.0126,  0.0424, -0.0560, -0.0771,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2392, -0.0939, -0.0257, -0.0307, -0.0999,  0.0238,  0.0241,  0.0048,
         0.0425, -0.0180,  0.0270, -0.0060,  0.0245,  0.0105,  0.0828,  0.0121,
        -0.0513, -0.0329,  0.0767, -0.0370, -0.0365,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0365, -0.0271,  0.0055,  0.0317,  0.0531,  0.0294,  0.0435,  0.0154,
         0.0201,  0.0229, -0.0033,  0.0033, -0.0266, -0.0184, -0.0003,  0.0119,
        -0.0104,  0.0052, -0.0062, -0.0069, -0.0098, -0.0325,  0.0104, -0.0197,
         0.0042,  0.0008,  0.0712,  0.0174,  0.0222,  0.0334, -0.0055, -0.0043,
         0.0055, -0.0028, -0.0084, -0.0081,  0.0035, -0.0128, -0.0310,  0.0126,
        -0.0090,  0.0250,  0.0475, -0.0325,  0.0022,  0.0066,  0.0231,  0.0115,
         0.0129, -0.0006,  0.0131,  0.0043, -0.0394,  0.0320,  0.0466],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.9780e-02, -3.6216e-02, -1.5001e-02,  3.1016e-02, -1.6040e-03,
         4.0183e-05,  2.9686e-02,  3.7586e-02, -1.4310e-02, -4.1273e-04,
         5.3366e-02, -3.1392e-03,  2.6440e-02,  1.1065e-03,  2.8891e-02,
        -1.0147e-02, -2.1591e-02,  4.5967e-03,  1.0774e-02,  6.0584e-03,
        -1.4133e-02, -2.0640e-02, -3.0719e-02,  3.9741e-03, -1.8964e-02,
        -3.4586e-02, -7.4004e-02, -2.0318e-02, -4.8478e-02,  1.0019e-02,
        -7.2167e-02, -1.6555e-03,  7.4334e-03,  1.9480e-03, -3.4073e-02,
        -8.1584e-03,  2.0617e-03,  1.6247e-02, -1.8575e-02,  1.5905e-02,
        -4.1885e-02,  1.3170e-02,  1.3176e-02,  6.8913e-03,  8.9057e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1526,  0.0040,  0.0653,  0.0155, -0.0353,  0.0079, -0.0167, -0.0174,
        -0.0573, -0.1147, -0.1064, -0.0272,  0.0270, -0.0310, -0.0012, -0.0105,
        -0.0409, -0.0010, -0.0774,  0.0191, -0.0136, -0.0238,  0.0352,  0.0993,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0374, -0.0836, -0.0245,  0.0443,  0.0433, -0.0084,  0.0342, -0.1149,
        -0.0064,  0.0016, -0.0694,  0.0352,  0.0014,  0.0727,  0.0413,  0.0146,
         0.0095,  0.0477,  0.0216,  0.0189, -0.0041,  0.0229,  0.0215,  0.0074,
         0.0383, -0.0023, -0.0031, -0.0003, -0.0019,  0.0016,  0.0133,  0.0999,
         0.0527,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0155, -0.0178,  0.1063, -0.0842,  0.0125,  0.0518,  0.0159, -0.1179,
        -0.0465,  0.0295,  0.0111,  0.0636,  0.0718, -0.0345, -0.1003, -0.0572,
         0.0002, -0.1361,  0.0274,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0723, -0.0876, -0.1007, -0.0908, -0.0208,  0.0068,  0.0376, -0.0049,
        -0.0682, -0.0681, -0.0372, -0.0061,  0.0018, -0.0057, -0.0290,  0.0082,
         0.0233, -0.0931,  0.0003, -0.0202,  0.0317, -0.0071, -0.0456, -0.0068,
        -0.1263,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.3230,  0.0430, -0.0166, -0.0155,  0.0005,  0.0028, -0.0464,  0.0505,
        -0.0174, -0.0117, -0.0522,  0.1074, -0.0567, -0.0493,  0.0524, -0.0243,
        -0.1304,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0464, -0.0055,  0.0627, -0.0055, -0.0067, -0.0343,  0.0186, -0.0917,
         0.0238,  0.0066, -0.0209,  0.0313,  0.0189,  0.0229,  0.0188, -0.0290,
        -0.0075, -0.0308, -0.0187,  0.0084,  0.0420, -0.0067, -0.0443, -0.0020,
        -0.0465,  0.0113,  0.0514,  0.0353,  0.0812, -0.0200,  0.0959, -0.0542,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2929, -0.0620,  0.0061, -0.0332, -0.0045,  0.0243, -0.0022,  0.0276,
         0.0140,  0.0087, -0.0249, -0.0156, -0.0369,  0.0016, -0.0006, -0.0061,
        -0.0256, -0.0146,  0.0160,  0.0123, -0.0128, -0.0249,  0.0365,  0.0152,
         0.0603, -0.0067,  0.0795,  0.1347,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0545,  0.0083,  0.0578, -0.1242, -0.0132, -0.0369, -0.0051, -0.0536,
        -0.0246, -0.0636, -0.0289,  0.0062, -0.0100,  0.0352,  0.0026,  0.0085,
        -0.0223,  0.0095,  0.0112,  0.0692,  0.0043, -0.0042, -0.0088, -0.0233,
        -0.0268,  0.0838,  0.0120, -0.0715, -0.1198,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0719,  0.0454,  0.0097, -0.0428, -0.0666, -0.0270,  0.1148, -0.0020,
         0.0293, -0.0807,  0.0251,  0.0126,  0.0101,  0.0004,  0.0493, -0.0075,
        -0.0114, -0.0180,  0.0062,  0.0215, -0.0170,  0.0128, -0.0268,  0.0121,
         0.0070, -0.0094, -0.0037, -0.0061,  0.0345,  0.0262,  0.0197, -0.1040,
        -0.0684,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0210, -0.0195,  0.0154,  0.1106,  0.0054,  0.0723,  0.1149,  0.0045,
         0.0262,  0.0201, -0.0397,  0.1003, -0.0397,  0.0175,  0.0153,  0.0350,
         0.0056, -0.0048,  0.1301,  0.0323, -0.0100, -0.0797, -0.0667, -0.0133,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0392,  0.0346,  0.0396,  0.0183, -0.0224,  0.0932,  0.1773,  0.0405,
        -0.0295,  0.0126, -0.0057, -0.0142, -0.0226,  0.0568, -0.0872, -0.0043,
         0.0935,  0.0009,  0.0220, -0.0004,  0.0530, -0.1247, -0.0077,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0692,  0.0487,  0.0305, -0.0330,  0.0837,  0.0304,  0.0164, -0.0168,
         0.0233,  0.0443,  0.0286, -0.0153, -0.0041, -0.0534,  0.0055, -0.0124,
         0.0049, -0.0417,  0.0082,  0.0147,  0.0162,  0.0050, -0.0097,  0.0228,
        -0.0114,  0.0191, -0.0124, -0.0030, -0.0005, -0.0129,  0.0175,  0.0252,
         0.0087, -0.0945,  0.0391,  0.0141, -0.0561,  0.0284,  0.0187,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0649, -0.0259, -0.0563, -0.0143, -0.0075,  0.0276, -0.0010, -0.0084,
        -0.0279,  0.0162, -0.0106, -0.0382,  0.0123, -0.0295, -0.0181, -0.0184,
         0.0058, -0.0300,  0.0070, -0.0408, -0.0150, -0.0340, -0.0385, -0.0103,
         0.0222,  0.0143,  0.0076,  0.0098,  0.0044, -0.0010, -0.0286, -0.0312,
        -0.0076, -0.0373, -0.0356, -0.0575,  0.0263, -0.0162, -0.0364, -0.0310,
         0.0210,  0.0512, -0.0024,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0806, -0.0388,  0.0130, -0.0096, -0.0152, -0.0054, -0.0225, -0.0202,
        -0.0141,  0.0006, -0.0161, -0.0074, -0.0219,  0.0077, -0.0265, -0.0307,
        -0.0009, -0.0008, -0.0122, -0.0142,  0.0011, -0.0065,  0.0086, -0.0032,
         0.0020, -0.0116,  0.0209,  0.0076, -0.0045, -0.0180, -0.0054, -0.0025,
        -0.0183, -0.0316,  0.0307,  0.0020, -0.0253,  0.0190, -0.0190,  0.0072,
        -0.0306, -0.0038,  0.0111,  0.0321,  0.0269, -0.0177,  0.0057, -0.0064,
        -0.0022,  0.0129,  0.0436,  0.0149, -0.0631, -0.0201,  0.0170,  0.0354,
        -0.0531], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0709, -0.1736,  0.1801, -0.0399, -0.0791, -0.0978,  0.0085, -0.0006,
         0.0349,  0.1953, -0.0242,  0.0082, -0.0739,  0.0130,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1056, -0.0750,  0.0359,  0.0314, -0.0075,  0.0305, -0.0232, -0.0069,
        -0.0158,  0.0220, -0.0122, -0.0522, -0.0183, -0.1150,  0.0056, -0.0107,
        -0.0015, -0.0137,  0.0221, -0.0109, -0.0322, -0.1061, -0.0374, -0.0150,
         0.0003, -0.0245,  0.0083,  0.0722,  0.0558,  0.0321,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0603,  0.1111, -0.1652,  0.1223, -0.0258, -0.1032, -0.2463, -0.1658,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([-0.1079, -0.0322,  0.0289,  0.0498,  0.0055, -0.0028,  0.0314,  0.0028,
         0.0249,  0.0055, -0.0724, -0.0048, -0.0354, -0.0238, -0.0134,  0.0036,
        -0.0120, -0.0073, -0.0105, -0.0241, -0.0362, -0.0423,  0.0113,  0.0540,
         0.0081,  0.1538, -0.0065, -0.0221, -0.0130,  0.0120, -0.0121,  0.0693,
        -0.0604,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0206, -0.0008,  0.0014, -0.0433, -0.0239,  0.0168, -0.0100, -0.0273,
         0.0067,  0.0081,  0.0060,  0.0025, -0.0002,  0.0005, -0.0051,  0.0094,
         0.0035, -0.0162,  0.0048,  0.0066, -0.0154, -0.0015,  0.0058, -0.0055,
         0.0019, -0.0103,  0.0130, -0.0054,  0.0008,  0.0031,  0.0031,  0.0017,
         0.0075, -0.0024,  0.0095,  0.0141,  0.0020, -0.0086, -0.0003,  0.0039,
        -0.0066, -0.0144,  0.0009,  0.0125, -0.0199, -0.0503, -0.0010,  0.0052,
        -0.0268,  0.0084,  0.0284, -0.0027,  0.0165,  0.0073,  0.0369, -0.0040,
         0.1023, -0.0117,  0.0464,  0.0242, -0.0038,  0.0148, -0.0149, -0.0511,
        -0.0104, -0.0193, -0.0063, -0.0058, -0.0197,  0.0671,  0.0310],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0059,  0.0340,  0.0104,  0.0281, -0.0148,  0.0059,  0.0337,  0.0061,
         0.0213,  0.0255,  0.0447, -0.0013, -0.0197,  0.0159, -0.0066,  0.0167,
        -0.0092,  0.0138, -0.0142, -0.0048, -0.0036,  0.0115,  0.0936, -0.0125,
         0.0422,  0.0342, -0.0197,  0.0415, -0.0020,  0.0539,  0.0042,  0.0237,
         0.0341, -0.0256, -0.0451,  0.0338,  0.0076,  0.0112, -0.0279,  0.0348,
        -0.0778,  0.0271,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0121, -0.0252, -0.0208, -0.0412, -0.0621,  0.0227, -0.0246,  0.0319,
        -0.0550, -0.0431, -0.0210, -0.0081, -0.0081, -0.0099,  0.0188,  0.0160,
         0.0061, -0.0198, -0.0123, -0.0758, -0.0273, -0.0070, -0.0005,  0.0072,
         0.0127,  0.0045,  0.0127,  0.0111,  0.0064, -0.0956,  0.0303, -0.0024,
         0.0751,  0.0439, -0.0145, -0.0916,  0.0225,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1550,  0.1011,  0.0753,  0.0189,  0.0649, -0.1911,  0.0601, -0.0254,
         0.0347, -0.0241,  0.0493,  0.0429, -0.0491, -0.0952,  0.0127,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0321, -0.0349,  0.0449, -0.0363,  0.0567, -0.0378, -0.0044,  0.0011,
        -0.0069,  0.0184,  0.0346,  0.0525, -0.0414, -0.0800, -0.0413,  0.0179,
        -0.1641, -0.1267, -0.0266,  0.1414,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0251,  0.0333,  0.0146, -0.0096, -0.0174, -0.0427, -0.0035,  0.0081,
        -0.0352, -0.0167, -0.0133, -0.0119, -0.0003, -0.0111,  0.0120,  0.0313,
         0.0552, -0.0138, -0.0165,  0.0434,  0.0312,  0.0110, -0.0023, -0.0491,
         0.0837, -0.1248,  0.0029,  0.0141,  0.0036,  0.0141, -0.0040, -0.0098,
         0.0156, -0.0083,  0.0028, -0.0324, -0.0054, -0.0107, -0.0063,  0.0146,
        -0.0233,  0.0016, -0.1136,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0010,  0.0149,  0.0205, -0.0147, -0.0352,  0.0225,  0.0282, -0.0136,
        -0.0029,  0.0075, -0.0637, -0.0265, -0.0033,  0.0102, -0.0377, -0.0411,
         0.0132, -0.0313, -0.0043,  0.0395,  0.0612,  0.0100,  0.0168,  0.0414,
         0.0692,  0.0608,  0.0219, -0.0188, -0.0269, -0.0031, -0.0183,  0.0205,
        -0.0246, -0.0274, -0.0056, -0.0744,  0.0284,  0.0391,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0396, -0.0294, -0.1104,  0.0054, -0.0656,  0.0284,  0.0148, -0.1397,
        -0.0298, -0.0455,  0.0115, -0.0091,  0.0275, -0.0161,  0.0080,  0.0149,
        -0.0638, -0.0086, -0.0014,  0.0250, -0.1794,  0.1261,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0604, -0.0103,  0.0204, -0.0651,  0.0008, -0.0331,  0.0703,  0.0395,
         0.0392,  0.0196, -0.0076, -0.0131,  0.0150,  0.0063, -0.0020, -0.0035,
         0.0018, -0.0164,  0.0033, -0.0026, -0.0069, -0.0348,  0.0271, -0.0243,
        -0.0171,  0.0152, -0.0486, -0.0419, -0.0160,  0.0432, -0.0669, -0.0200,
         0.0243, -0.0212, -0.0131, -0.0012,  0.0149,  0.0034,  0.0233,  0.1060,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0245, -0.0713, -0.0484,  0.0313, -0.0545,  0.0428,  0.0222,  0.0113,
        -0.0657, -0.0978,  0.0410, -0.0296, -0.0099,  0.0006, -0.0394, -0.0175,
        -0.0320, -0.0260, -0.0422,  0.0090, -0.0172, -0.0086,  0.0532, -0.0804,
        -0.0188,  0.0327, -0.0017, -0.0008, -0.0214,  0.0117,  0.0105, -0.0258,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1114, -0.0175,  0.0034,  0.0462,  0.0114,  0.0238, -0.0133, -0.0145,
         0.0009,  0.0109,  0.0170,  0.0012, -0.1139,  0.0182, -0.0142, -0.0477,
         0.0281, -0.0268, -0.0014, -0.0244, -0.0065, -0.0320,  0.0650, -0.0063,
         0.0034,  0.0413, -0.0126, -0.0006,  0.0186, -0.0208,  0.0249, -0.0044,
         0.0372,  0.0665, -0.1136,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.1486,  0.0783, -0.0002, -0.0420,  0.1037,  0.0903, -0.1472, -0.0416,
         0.0269, -0.0156,  0.0398, -0.0151, -0.0542, -0.0179,  0.0208, -0.0101,
        -0.0049, -0.0099, -0.0106, -0.0448,  0.0774,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1392, -0.1332,  0.0283,  0.1027,  0.0276, -0.0417, -0.0341,  0.0060,
        -0.0245,  0.0003, -0.0289,  0.0893, -0.0330,  0.0581,  0.0202, -0.0220,
        -0.0450, -0.0829,  0.0828,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0764, -0.1418,  0.0282,  0.0798, -0.0166,  0.0372,  0.0924,  0.1313,
         0.0325, -0.0100, -0.0558, -0.0368, -0.0056, -0.0267,  0.0130, -0.0273,
         0.0176, -0.0134, -0.0754,  0.0823,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1935, -0.2266, -0.0741,  0.0144,  0.0028, -0.0383,  0.1081, -0.0025,
         0.0139, -0.0246, -0.0104, -0.0081, -0.0076, -0.0508, -0.0355, -0.0179,
        -0.0313, -0.0128,  0.0089, -0.0085, -0.0176, -0.0630,  0.0289,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0931, -0.0104, -0.0461, -0.0234, -0.0459,  0.0203, -0.0171,  0.0007,
        -0.0417, -0.0337,  0.0328,  0.0310,  0.0013, -0.0579,  0.1084,  0.0539,
        -0.0400, -0.0250, -0.0206,  0.0605,  0.0066, -0.0653, -0.0075, -0.1441,
         0.0129,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0112,  0.0525,  0.0224,  0.0091, -0.0145,  0.0253,  0.0306,  0.0236,
        -0.0669,  0.1242,  0.0385, -0.0177, -0.0072,  0.1011,  0.0077,  0.0415,
        -0.0157,  0.0377, -0.0250, -0.0023,  0.0419,  0.0025,  0.0084,  0.0470,
         0.0052,  0.0225,  0.0241, -0.0182,  0.0355, -0.0150,  0.0188,  0.0049,
        -0.0086, -0.0534,  0.0196,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0514, -0.0392,  0.0178, -0.0556, -0.0099,  0.0514,  0.0181, -0.0377,
        -0.0165,  0.0361,  0.0324, -0.0328, -0.0610, -0.0577, -0.0662, -0.0238,
        -0.0476, -0.0194, -0.0729, -0.1326,  0.1040,  0.0159,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2649,  0.0591,  0.0391,  0.0339, -0.0064, -0.0199,  0.0250, -0.0126,
        -0.0743, -0.0269,  0.0267,  0.0217,  0.0289, -0.0214, -0.0257, -0.0131,
         0.0643,  0.0091, -0.0621, -0.0984,  0.0666,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1542, -0.0167, -0.0142, -0.0097,  0.0420, -0.0435, -0.0257, -0.0248,
         0.0742,  0.0065, -0.0050, -0.0020, -0.0451, -0.0016, -0.1354, -0.0651,
         0.0824, -0.1229, -0.1291,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0930,  0.0802, -0.0799, -0.0105,  0.0034, -0.0105, -0.0176, -0.0095,
         0.0085,  0.0044, -0.0139, -0.0205, -0.0189, -0.0097,  0.0051, -0.0558,
        -0.0671, -0.0119,  0.0198, -0.0402, -0.0390, -0.0685, -0.0498, -0.0046,
        -0.0312,  0.0009, -0.0034, -0.0093,  0.0121, -0.0041, -0.0102, -0.0183,
        -0.0012,  0.0025,  0.0146, -0.0119,  0.0382, -0.0230, -0.0099, -0.0022,
         0.0019,  0.0184, -0.0089,  0.0022,  0.0080,  0.0126, -0.0124],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1986, -0.0307, -0.0184, -0.0042, -0.0060, -0.0251, -0.0079,  0.0406,
        -0.0159, -0.0031, -0.0108, -0.0246,  0.0209, -0.0220,  0.0051, -0.0051,
        -0.0047, -0.0144, -0.0100,  0.0230,  0.0716, -0.0344, -0.0441,  0.0333,
        -0.0057, -0.0211,  0.0277,  0.0055,  0.0079, -0.0078, -0.0027, -0.0372,
        -0.1224, -0.0877,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0342, -0.0388,  0.0205,  0.0018,  0.0175, -0.0035,  0.0025, -0.0012,
        -0.0188,  0.0117,  0.0628, -0.0326,  0.0009,  0.0091,  0.0664,  0.0905,
         0.0579,  0.0506, -0.0439,  0.0187, -0.0234, -0.0464,  0.0038, -0.0322,
        -0.0408, -0.0066, -0.0310, -0.0231,  0.0268, -0.0252,  0.0014, -0.0052,
         0.1304, -0.0198,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.1900,  0.0219,  0.0415,  0.0830,  0.0015,  0.0034, -0.0060, -0.0726,
        -0.1333, -0.0093, -0.1177, -0.0473, -0.0257, -0.0234, -0.0297, -0.0009,
         0.0955,  0.0974,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0996,  0.0960,  0.0396, -0.0130,  0.0228, -0.0249,  0.0308, -0.0153,
         0.0162, -0.0196,  0.0121, -0.0291,  0.0409,  0.0139,  0.0046, -0.0072,
         0.0174, -0.0888, -0.0080, -0.0559, -0.0341,  0.1065, -0.0441,  0.0388,
        -0.0290, -0.0462,  0.0459,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0306, -0.1794, -0.0688,  0.0040, -0.0094,  0.0047, -0.0657, -0.0759,
        -0.0435, -0.0473, -0.0291,  0.0454, -0.0344, -0.0014, -0.0025,  0.0411,
        -0.0105, -0.0016,  0.0037, -0.0185, -0.0173, -0.0211, -0.0099, -0.0372,
        -0.0935, -0.1033,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2811,  0.0119, -0.0063, -0.0206, -0.0887, -0.0583,  0.0137, -0.0018,
         0.0010, -0.0882, -0.0288,  0.0197, -0.0209, -0.0389, -0.0629,  0.0044,
         0.0617, -0.0313, -0.0684,  0.0276, -0.0637,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0682,  0.0448, -0.1025,  0.0472,  0.0078, -0.0212,  0.0845, -0.0034,
        -0.0479,  0.0624,  0.0064,  0.0234,  0.0459, -0.1262, -0.3082,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0864,  0.1324,  0.0227,  0.0993, -0.1288, -0.0124,  0.0235, -0.0375,
         0.0142, -0.0646,  0.0467,  0.1244, -0.0370,  0.0103, -0.0078, -0.0319,
        -0.0561,  0.0641,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0602, -0.0270,  0.0305, -0.0489, -0.0429, -0.2196,  0.0604, -0.0493,
         0.0732, -0.0509,  0.0352, -0.0339, -0.0703,  0.0255,  0.0320, -0.0014,
         0.0682, -0.0375,  0.0331,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0438, -0.0109,  0.0121,  0.0054, -0.0060,  0.0056, -0.0021, -0.0305,
         0.0051,  0.0180, -0.0012,  0.0323,  0.0198, -0.0228,  0.0014,  0.0164,
         0.0214,  0.0242, -0.0013, -0.1475, -0.0678,  0.1094,  0.0036, -0.0113,
         0.0121,  0.0134, -0.0139, -0.0303, -0.0059,  0.0043,  0.0102,  0.0385,
        -0.0050, -0.0227,  0.0335,  0.0281, -0.0035, -0.0038,  0.0070,  0.0246,
         0.0019,  0.0058, -0.0561,  0.0593], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0689, -0.0412, -0.0404, -0.0186,  0.0084,  0.0425,  0.0135, -0.0120,
        -0.0121,  0.0096, -0.0239, -0.0604, -0.0138, -0.0147,  0.0097, -0.0387,
         0.0457, -0.0356,  0.0466,  0.0233, -0.0294,  0.0032, -0.0429,  0.0106,
         0.0210, -0.0391,  0.0225, -0.0357,  0.0005, -0.0075, -0.0313,  0.0214,
         0.0087,  0.0269, -0.1197,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.7471e-01, -5.3668e-02,  4.3259e-02,  1.8160e-02,  2.6775e-02,
         7.1039e-02,  8.8501e-03,  2.3479e-02,  5.2279e-02,  3.1168e-02,
         2.9764e-02,  1.8831e-02, -3.7820e-05,  4.5585e-02,  1.1887e-02,
         9.9998e-02,  4.2798e-02,  1.4360e-02,  6.8875e-02, -2.2349e-02,
         1.9907e-02,  1.5538e-03, -5.5185e-02, -6.5484e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 9.6707e-02,  1.1617e-02, -7.5318e-03, -4.7560e-02,  4.4562e-02,
        -2.8705e-02, -1.9526e-02, -5.0986e-02, -2.5491e-02,  3.2645e-02,
        -7.9698e-02,  3.2605e-02, -7.6895e-03,  1.7515e-02, -1.5287e-02,
        -1.4381e-02, -2.4417e-02,  1.5833e-02, -3.1116e-02, -4.9362e-02,
        -3.2087e-02, -1.3328e-02, -2.9771e-02, -4.8390e-02, -3.1570e-03,
        -2.0959e-02, -1.2710e-02, -1.5587e-02, -1.5956e-02, -2.7285e-02,
         6.8305e-05, -2.9291e-03, -1.2058e-01,  3.9609e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.0565e-01, -1.5350e-01, -9.2899e-02, -3.6127e-02,  3.1599e-02,
         2.0947e-02,  2.0007e-02, -4.2564e-02, -1.1326e-02,  2.1783e-02,
         3.5774e-02,  1.3050e-02, -1.6205e-02, -1.8643e-02, -1.1348e-02,
        -6.8875e-02,  1.3697e-02, -1.4725e-02,  6.6951e-03,  6.0585e-03,
         5.3803e-03,  1.6888e-02, -4.8569e-02,  7.5754e-03, -1.3955e-02,
        -1.4461e-04,  3.3550e-03,  3.7263e-02,  2.6600e-02, -9.8800e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 3.7792e-01, -5.8915e-02, -6.5191e-02, -1.4253e-01,  4.5877e-02,
        -4.2725e-05,  1.9826e-02, -3.6653e-02, -1.0837e-02, -1.6492e-02,
        -7.1900e-03, -1.8953e-01,  2.8991e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2232,  0.0427,  0.0862, -0.0066, -0.0150,  0.0173, -0.0597, -0.0390,
        -0.0998, -0.0319, -0.0262,  0.0239,  0.0055,  0.0185,  0.0010, -0.0101,
        -0.2158,  0.0777,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0030, -0.1282, -0.0147, -0.2795, -0.1695,  0.1941, -0.0760, -0.1349,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.1448e-02, -1.7096e-02, -1.8309e-02, -2.2255e-02,  5.3269e-03,
         1.9285e-02, -8.8645e-03, -1.5863e-02,  3.0284e-02,  1.0391e-02,
        -2.8249e-04,  2.1454e-02, -1.3741e-02,  7.0168e-03, -1.3050e-03,
        -1.4007e-02, -1.1585e-02,  1.7782e-02, -1.9541e-02, -1.2601e-02,
        -6.7071e-02,  2.3152e-02,  3.9934e-02,  2.4151e-02,  2.6307e-02,
         2.8078e-03, -4.3927e-03,  3.0402e-02, -4.7730e-02, -2.2052e-02,
        -1.6386e-02, -9.1917e-05, -1.0361e-02, -2.5843e-02, -2.9920e-02,
        -1.0449e-02, -2.3770e-02,  5.8496e-03,  4.4078e-02, -3.3538e-02,
        -3.3159e-05, -1.8794e-02, -3.0235e-02,  2.5384e-02, -4.3031e-02,
         1.6009e-02,  7.9793e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.3805,  0.0675,  0.2454, -0.0452, -0.1330,  0.0335,  0.0051, -0.0242,
        -0.0209, -0.0047, -0.0400,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1363, -0.0704,  0.0669,  0.0174, -0.0376, -0.0240, -0.0154, -0.0892,
         0.0275,  0.0328,  0.0401,  0.0788, -0.0254,  0.0097, -0.1153,  0.0466,
        -0.0608,  0.1060,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0509, -0.1367, -0.1208,  0.0040, -0.0719, -0.0024,  0.0417,  0.0174,
        -0.0379, -0.0256, -0.0142,  0.1375, -0.0005, -0.0716, -0.2669,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0908,  0.0555,  0.0284, -0.0295, -0.1380, -0.0919,  0.0663, -0.0530,
        -0.0639, -0.0490,  0.0065, -0.0658,  0.0858,  0.1338, -0.0417,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1794, -0.0617, -0.0850, -0.0234,  0.0598,  0.0270,  0.0116,  0.0183,
        -0.0696, -0.0009, -0.0014, -0.0074, -0.0248,  0.0032, -0.0292, -0.0434,
        -0.0797, -0.0375, -0.0435,  0.0540,  0.0065, -0.0441, -0.0111,  0.0410,
        -0.0134,  0.0234,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1076, -0.2389, -0.1992,  0.0636,  0.0353,  0.0653,  0.0684, -0.0403,
        -0.0262, -0.0410, -0.0389, -0.0069, -0.0041,  0.0431,  0.0212,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2333, -0.0118, -0.0084, -0.0144,  0.0541,  0.0461,  0.0262,  0.0171,
         0.0372,  0.0818, -0.0449, -0.0018,  0.0322,  0.0369,  0.1475, -0.0146,
         0.0024,  0.0126,  0.0073,  0.0062, -0.0258,  0.0320, -0.0439, -0.0141,
        -0.0391,  0.0083,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0317, -0.0073, -0.0216, -0.0523, -0.0447, -0.0231,  0.0004,  0.0128,
         0.0119, -0.0153,  0.0161,  0.0084, -0.0035, -0.0355,  0.0498, -0.0705,
        -0.1454, -0.0051,  0.0549,  0.0321, -0.0229, -0.0362,  0.0147, -0.0003,
         0.2223,  0.0612,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0142, -0.0648, -0.0655, -0.0262, -0.0689, -0.0142, -0.0121, -0.0312,
         0.0010, -0.0269, -0.0011, -0.0133, -0.0556,  0.0620, -0.0194, -0.0544,
        -0.0170, -0.0201, -0.0015, -0.0627,  0.0181, -0.0050,  0.0279, -0.0167,
        -0.0200, -0.0293, -0.0005,  0.0351,  0.0318,  0.0239, -0.0079, -0.0173,
         0.0056,  0.0171,  0.0383, -0.0731,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0599,  0.0138, -0.0026, -0.0495, -0.0131,  0.0533,  0.0038,  0.0037,
        -0.0021,  0.0266,  0.0092,  0.0097,  0.0493,  0.0418, -0.0117, -0.0076,
        -0.0156,  0.0113, -0.0006, -0.0972,  0.0408,  0.0942,  0.0544,  0.1258,
        -0.0113, -0.0050, -0.0036, -0.0189,  0.0018,  0.0130,  0.0017, -0.0004,
         0.0108,  0.0611, -0.0236, -0.0514,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0810, -0.0561, -0.0393,  0.0474, -0.0973,  0.0614,  0.1244, -0.0179,
        -0.0282,  0.0311,  0.0261, -0.0332,  0.0240, -0.0337, -0.0366,  0.0037,
        -0.1041,  0.0333,  0.0792,  0.0420,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2179,  0.0131,  0.0018,  0.0172, -0.0458,  0.0026,  0.0229, -0.0119,
         0.0066,  0.0005,  0.0105, -0.0406,  0.0123,  0.0057, -0.0015, -0.0077,
        -0.0194, -0.0047,  0.0217,  0.0167, -0.0595,  0.0081,  0.0220, -0.0306,
        -0.0147,  0.0074,  0.0180,  0.0627, -0.0454,  0.0100,  0.0185, -0.0123,
         0.0512, -0.0171, -0.0223, -0.0738,  0.0274,  0.0183,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1090, -0.0178, -0.0103, -0.0038, -0.0080, -0.0229, -0.0116,  0.0153,
        -0.0059,  0.0836,  0.0149, -0.0374, -0.0232,  0.0080,  0.0874, -0.0014,
        -0.0300, -0.0009,  0.0408, -0.0126, -0.0328,  0.0015,  0.0631, -0.0253,
        -0.0031,  0.0103,  0.0019,  0.0062, -0.0340, -0.0044, -0.0024,  0.0211,
         0.0148,  0.0208, -0.0216,  0.0298, -0.0069,  0.0561, -0.0331, -0.0279,
        -0.0379], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0437, -0.1956, -0.0267, -0.0142, -0.0157, -0.0557, -0.0599, -0.0172,
        -0.0094,  0.0086,  0.0328,  0.0080,  0.0323,  0.0358, -0.0253, -0.0475,
         0.0005,  0.0083, -0.0103, -0.1067, -0.0981, -0.1478,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0837,  0.0503, -0.0098,  0.0196,  0.0621, -0.0186, -0.0053, -0.0141,
         0.0122, -0.0516, -0.1200,  0.0180,  0.0013,  0.0140,  0.0013,  0.0127,
        -0.0010, -0.0291, -0.0168,  0.0035,  0.0191, -0.0094,  0.0143,  0.0154,
        -0.0003, -0.0278,  0.0136,  0.0096,  0.0087, -0.0120,  0.0194, -0.0157,
        -0.0773,  0.0797,  0.1326,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1219, -0.0200,  0.0446,  0.0063,  0.0539, -0.1179,  0.0187, -0.0405,
        -0.0074,  0.0168,  0.0110,  0.0284,  0.0421,  0.0043, -0.0009,  0.1215,
         0.0427,  0.0370,  0.0120, -0.0251,  0.0240,  0.0139,  0.0095,  0.0303,
        -0.0081,  0.0574, -0.0249, -0.0030,  0.0559,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0904, -0.0207, -0.0354, -0.1003,  0.0309,  0.0606, -0.0822, -0.1398,
         0.0372,  0.0179,  0.0176,  0.0176, -0.0482, -0.0039,  0.0078,  0.0421,
        -0.0114, -0.0326, -0.0154, -0.0147,  0.0332,  0.0411,  0.0507, -0.0320,
         0.0162,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0564,  0.0835, -0.0175, -0.0476, -0.0344,  0.0047, -0.0986, -0.0111,
         0.0198,  0.0922,  0.0177,  0.0482, -0.0041,  0.1218, -0.0178, -0.0198,
         0.0408, -0.0201,  0.0076,  0.0825, -0.0274,  0.1259, -0.0007,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1419,  0.0122,  0.0133, -0.0047,  0.0081,  0.0561, -0.0188, -0.0067,
         0.0065,  0.0229,  0.0194,  0.0131,  0.0183,  0.0093,  0.0018,  0.0240,
        -0.0181,  0.0136, -0.0112,  0.0026,  0.0008,  0.0170, -0.0087, -0.0004,
         0.0053,  0.0073,  0.0092, -0.0105, -0.0268, -0.0868, -0.0044, -0.0033,
        -0.0504, -0.0625, -0.0712, -0.0045,  0.1031,  0.0036, -0.1015,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1053, -0.2335, -0.0345, -0.0612, -0.1008, -0.0567,  0.0734, -0.0525,
        -0.0902, -0.0527,  0.0340,  0.1052,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 1.2711e-01, -4.9023e-02, -1.2934e-01,  8.8224e-03,  1.8915e-03,
        -7.0034e-02, -3.9815e-03,  2.1853e-02, -3.0758e-03,  7.3581e-04,
         6.2850e-03, -5.1842e-03, -2.5905e-02, -4.6833e-02, -2.0705e-02,
        -2.1324e-03,  1.9182e-02, -8.4653e-03, -3.5002e-03, -2.3582e-03,
        -1.5489e-02, -1.4260e-02, -1.2520e-02, -2.1970e-03, -1.8355e-02,
        -1.0267e-02, -1.3210e-02, -4.8504e-02,  8.6306e-03, -6.9306e-02,
        -2.5793e-02, -1.3959e-02, -1.6296e-02, -1.4208e-02, -5.9179e-03,
         6.9066e-03, -3.4123e-02,  3.2353e-03,  9.2095e-05,  3.9498e-03,
         9.3406e-03,  2.5031e-02, -1.4295e-02,  3.1857e-02, -2.5841e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0600, -0.0326, -0.1125, -0.0778, -0.1619,  0.0992, -0.0732, -0.1315,
        -0.0322,  0.1125, -0.0285,  0.0782,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0930, -0.3087,  0.1032, -0.1778, -0.0054, -0.1155, -0.1434,  0.0530,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0992, -0.0486, -0.0341,  0.0135, -0.0215, -0.0061, -0.0058, -0.0005,
        -0.0051, -0.0507, -0.0171,  0.0080, -0.0053, -0.0019, -0.0043,  0.0296,
         0.0059, -0.0108, -0.0017,  0.0063, -0.0066, -0.0015, -0.0008, -0.0337,
        -0.0113, -0.0123,  0.0007,  0.0009, -0.0007,  0.0278,  0.0430, -0.0004,
         0.0028,  0.0135,  0.0048, -0.0094,  0.0269,  0.0079, -0.0086,  0.0114,
         0.0188,  0.0032, -0.0033, -0.0123, -0.0344, -0.0225, -0.0464, -0.0252,
        -0.0325, -0.0351, -0.0053,  0.0074, -0.0006,  0.0101,  0.0139,  0.0113,
         0.0059,  0.0064,  0.0419, -0.0367, -0.0257], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0833,  0.0425,  0.0301, -0.0112, -0.0047, -0.0110, -0.0336,  0.0252,
         0.0306, -0.1171, -0.0445,  0.0389,  0.0161, -0.0182,  0.0104,  0.0014,
         0.0318, -0.0324, -0.0008,  0.0314, -0.0203,  0.0015, -0.0669,  0.0271,
        -0.0099, -0.0296,  0.0454,  0.0278, -0.0175, -0.0371, -0.0666,  0.0265,
        -0.0085,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0403,  0.0243, -0.0309,  0.0056, -0.0421, -0.0126, -0.0385, -0.0177,
         0.0038, -0.0656, -0.0206, -0.0088, -0.0079, -0.0029, -0.0216, -0.0653,
        -0.1874,  0.1010, -0.0622,  0.0063, -0.0233, -0.0661, -0.0508,  0.0946,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0567, -0.0727, -0.0539, -0.0635, -0.0489, -0.0066, -0.0038, -0.0948,
         0.0743,  0.0136, -0.0014, -0.0330, -0.0316,  0.0075, -0.0322,  0.0021,
        -0.0160,  0.0334, -0.0232, -0.1121, -0.0079, -0.0281, -0.0605, -0.1220,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.9435e-03, -2.8986e-03, -6.4550e-02, -5.1126e-02,  1.5993e-03,
         4.9004e-02, -3.9440e-05, -9.4546e-03, -1.4844e-02, -2.4473e-02,
         6.8421e-03,  5.4594e-02, -5.4785e-02, -1.5649e-02, -2.0585e-02,
         1.8049e-02, -8.4980e-03, -9.3830e-02, -3.8216e-03,  6.5358e-02,
        -9.8204e-03,  5.0484e-02, -6.9184e-02, -3.3443e-02,  8.2504e-02,
         6.2656e-03, -2.8005e-02, -1.8174e-02,  1.2806e-02, -1.4224e-02,
         2.8642e-03,  4.5543e-03,  3.1842e-02, -7.3886e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0624, -0.1562, -0.0228, -0.0097, -0.0518,  0.0170,  0.0018,  0.0024,
        -0.0336, -0.0780, -0.0328,  0.0014, -0.0451, -0.0720,  0.0406,  0.1100,
        -0.0421,  0.0126, -0.0101, -0.0068,  0.0086, -0.0297, -0.0567, -0.0159,
         0.0800,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1272,  0.0512, -0.2487,  0.0051, -0.0174,  0.0644, -0.0039,  0.0396,
        -0.0242, -0.2040, -0.1458, -0.0170, -0.0137, -0.0169, -0.0082,  0.0127,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0645,  0.0721,  0.0513, -0.0583,  0.0084,  0.0027,  0.0083, -0.0138,
        -0.0665,  0.0030,  0.0363,  0.0232,  0.0333,  0.0226,  0.2003,  0.0364,
         0.0044, -0.0250,  0.0084,  0.0478,  0.0072, -0.0587,  0.1477,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1499,  0.0135,  0.0273, -0.1097, -0.0121, -0.0511,  0.0801, -0.0366,
         0.0179, -0.0063, -0.2266, -0.0358, -0.0661, -0.1670,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.2240, -0.0417,  0.0091, -0.0451,  0.0165,  0.0033, -0.0088, -0.0055,
        -0.0685, -0.0106, -0.0351,  0.0011, -0.0241,  0.0068, -0.0391, -0.0185,
         0.0302, -0.0584, -0.0007, -0.0489, -0.0284,  0.0582, -0.1676,  0.0499,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1932,  0.2034,  0.0353, -0.1053, -0.1404,  0.0341, -0.0938, -0.0419,
        -0.1235, -0.0191,  0.0101,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1831, -0.0551, -0.0457,  0.0679, -0.1167, -0.0238, -0.1007,  0.0600,
        -0.0269, -0.0120, -0.0227,  0.0271, -0.0304, -0.0434, -0.0024,  0.0237,
         0.0014, -0.0634, -0.0053,  0.0111,  0.0774,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2127, -0.1248, -0.0625, -0.0090,  0.0256,  0.1093, -0.0646, -0.0615,
         0.0174, -0.0203,  0.0482,  0.0658,  0.1299,  0.0483,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0997, -0.0280, -0.0302,  0.0082, -0.0435, -0.0139,  0.0099, -0.1286,
        -0.0208,  0.0690, -0.0661,  0.1285, -0.0099,  0.0288,  0.0131,  0.0578,
        -0.0508,  0.0905, -0.0091, -0.0052, -0.0887,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0126,  0.0173, -0.1738, -0.2123,  0.1270, -0.0289, -0.0303,  0.0213,
        -0.1799, -0.1097, -0.0708,  0.0151, -0.0009,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.0245e-02, -2.7017e-02, -1.5101e-02, -2.1019e-03, -1.0492e-02,
         2.0543e-02, -1.5340e-02,  1.6117e-02,  5.0965e-02,  3.9410e-02,
        -1.8145e-02,  3.5302e-02, -1.3608e-02,  4.9829e-03, -1.9979e-02,
        -6.3089e-03,  3.1586e-03,  4.0613e-03, -3.8984e-03,  1.7039e-02,
        -3.0845e-02, -1.8826e-02, -2.2117e-02, -5.4510e-02, -8.7953e-02,
         6.6542e-03, -4.9543e-02, -1.9165e-02,  3.4924e-03, -3.3557e-02,
        -1.5232e-02,  7.0571e-02, -3.3206e-05,  3.9920e-03,  3.8466e-02,
        -2.9450e-02, -2.1255e-02,  2.5365e-03,  3.1953e-03, -7.5109e-03,
         1.3606e-02,  2.1584e-02,  1.0436e-02,  8.9983e-02,  1.6711e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0519, -0.0061, -0.0310, -0.0083, -0.0066,  0.0438, -0.0304, -0.0094,
        -0.0593, -0.0136, -0.0322,  0.0008, -0.0428, -0.0410, -0.0381,  0.0006,
         0.0082, -0.0016, -0.0038, -0.0065,  0.0280, -0.0333, -0.0123, -0.0108,
         0.0652,  0.0205, -0.0049, -0.0282, -0.0536, -0.0254,  0.0810,  0.0643,
        -0.0518, -0.0848,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0344, -0.1244, -0.0801, -0.0165, -0.0418,  0.0079, -0.0372, -0.0365,
         0.1451, -0.0092, -0.0545, -0.0015,  0.0707, -0.0700,  0.0013, -0.0589,
         0.0045, -0.0136, -0.0112,  0.0195, -0.0662, -0.0092, -0.0053, -0.0469,
         0.0099, -0.0236,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1965, -0.0864,  0.0815, -0.0025, -0.0531,  0.0763,  0.0771,  0.0096,
         0.0586, -0.0135, -0.0160, -0.0186,  0.0034, -0.0243, -0.0529, -0.0142,
        -0.0528,  0.0302, -0.0228, -0.0457,  0.0059, -0.0557,  0.0026,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0289,  0.0636, -0.0019, -0.0392, -0.0004,  0.0108,  0.0044,  0.0109,
         0.0341,  0.0213,  0.0099,  0.0490, -0.0527, -0.0010,  0.0819, -0.0309,
         0.0025,  0.0441,  0.0246, -0.0231, -0.0152,  0.0426,  0.0344, -0.0060,
         0.0115, -0.0227,  0.0244,  0.0183,  0.0010, -0.0370, -0.0144,  0.0342,
        -0.0206, -0.0748, -0.0480, -0.0132, -0.0253, -0.0008, -0.0176,  0.0028,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.8664e-03, -2.3564e-02, -2.1013e-02,  7.5667e-02, -7.6587e-02,
        -4.0532e-02, -1.6445e-01, -4.4394e-02,  8.3898e-03, -2.9787e-02,
         3.0318e-02, -2.1521e-02,  1.5473e-04,  1.7561e-02, -5.7329e-03,
        -7.8683e-04, -4.5757e-02,  2.4541e-02,  3.1853e-02, -1.6859e-02,
         5.5770e-03, -6.8082e-03,  9.6384e-03,  8.7083e-03, -7.0086e-02,
        -8.1942e-03,  9.1075e-03, -6.0720e-03, -1.5007e-02, -9.1139e-03,
        -6.7829e-02, -2.6588e-02, -1.7565e-02, -5.6365e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0342,  0.0086, -0.0381, -0.0275,  0.0315,  0.0061, -0.0112,  0.0054,
        -0.0359,  0.0769,  0.0019, -0.0202,  0.0711, -0.0018, -0.0708, -0.0171,
        -0.0090, -0.0197, -0.0050,  0.0049, -0.0464,  0.0330,  0.0219,  0.0178,
         0.0071, -0.0709,  0.0205, -0.0293,  0.0317,  0.0086, -0.0060, -0.0095,
         0.0199, -0.0195, -0.0017,  0.0011,  0.0012, -0.0119, -0.0043, -0.0117,
        -0.0123,  0.0142, -0.0143,  0.0007,  0.0063, -0.0338, -0.0061,  0.0050,
         0.0102,  0.0065,  0.0077,  0.0120], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0790,  0.0103, -0.0051,  0.0773, -0.0476,  0.0172, -0.0305, -0.0418,
        -0.0330, -0.0046, -0.0045,  0.0337, -0.0328, -0.0130, -0.0202, -0.0158,
        -0.0620, -0.0083, -0.0099,  0.0312, -0.0251, -0.0049, -0.0274, -0.0893,
        -0.0157,  0.0190, -0.0321, -0.0127, -0.0834,  0.0074,  0.1055,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0757,  0.0609, -0.0350, -0.0245, -0.0047, -0.0220, -0.0388, -0.0462,
        -0.0422, -0.0853, -0.0312, -0.0103, -0.1427,  0.0174, -0.0498, -0.0808,
        -0.0123,  0.0542,  0.0124, -0.0160, -0.0127,  0.0261,  0.0417,  0.0134,
         0.0435,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1026,  0.1036, -0.0226, -0.0114,  0.0785, -0.0478,  0.0493,  0.0514,
        -0.1072,  0.0809, -0.0908,  0.2540,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1074, -0.0356, -0.0013,  0.0015, -0.0049, -0.0165, -0.0377,  0.0045,
        -0.0348,  0.0760, -0.0485, -0.0185, -0.0248, -0.0389, -0.0240, -0.0086,
         0.0195, -0.0466, -0.0131, -0.0063, -0.0151,  0.0017,  0.0225,  0.1756,
        -0.0085, -0.0363, -0.0701, -0.0606, -0.0407,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1210, -0.0023, -0.1788, -0.0686, -0.0514, -0.1133, -0.0993,  0.0860,
        -0.0618, -0.0545,  0.0768, -0.0863,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0463, -0.0245, -0.0955, -0.0463,  0.1134, -0.0606, -0.2227,  0.1406,
        -0.0760,  0.0018,  0.0430,  0.1293,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0024,  0.1535, -0.0536, -0.0298, -0.0491, -0.0848,  0.0266, -0.0127,
        -0.0547,  0.0416, -0.0209, -0.0512, -0.0060,  0.1272, -0.0711, -0.0038,
        -0.0242, -0.0671, -0.0588, -0.0608,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.1857e-01, -5.0525e-02, -4.2411e-02, -4.5572e-02, -6.0889e-02,
        -3.2190e-02, -1.1850e-02,  2.3404e-02,  1.1308e-02, -9.7312e-03,
         1.3492e-02, -2.6260e-02,  1.6802e-02,  3.3979e-02, -1.0079e-02,
         2.1810e-02, -1.0387e-02, -1.0980e-02,  7.5569e-03, -3.3130e-02,
         3.3515e-04, -2.5026e-02,  4.1006e-02,  2.2744e-02,  4.9296e-03,
        -4.2796e-03, -1.5691e-02, -1.9992e-02, -2.1169e-02, -3.5448e-02,
        -2.8367e-02, -4.2606e-02, -3.7182e-05, -1.0303e-02, -4.9553e-03,
         9.8860e-02, -3.3320e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1428, -0.3142, -0.0226,  0.0452,  0.1782,  0.1414, -0.0159,  0.0467,
        -0.0929,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1659, -0.0227, -0.0007, -0.0880, -0.0346,  0.0482, -0.0381, -0.0203,
        -0.0310,  0.0475,  0.0332,  0.0515,  0.0312, -0.0415, -0.0188, -0.0233,
        -0.0165,  0.0837, -0.0310, -0.0031,  0.0597,  0.0282,  0.0379,  0.0055,
        -0.0116, -0.0152,  0.0110,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.0043e-01, -4.1699e-02, -1.4647e-02, -3.2627e-02, -4.6982e-02,
        -3.1206e-03, -9.7929e-03, -8.0943e-04,  7.4008e-03, -5.7042e-02,
        -2.7636e-02,  1.7567e-02,  6.2224e-05, -6.3954e-03, -2.9956e-02,
        -8.1594e-03,  5.9783e-03,  1.3651e-02,  5.3406e-03,  4.6770e-02,
         9.7598e-02, -9.9399e-03, -1.3908e-02, -9.3907e-03, -9.7161e-03,
        -2.5947e-02, -9.8319e-03, -8.4388e-03, -3.7327e-02,  4.6389e-03,
        -2.7004e-03, -2.3125e-02,  2.6627e-02, -4.4739e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0289, -0.0097, -0.0121,  0.0217,  0.0629,  0.0468,  0.0182,  0.0084,
        -0.0028,  0.0461,  0.1171,  0.0284,  0.0164,  0.0086, -0.0058, -0.0386,
         0.0501, -0.0085, -0.0034, -0.0045,  0.0114, -0.0208,  0.0159, -0.0183,
         0.0566,  0.0104,  0.0162, -0.0163,  0.0010, -0.0124, -0.0165,  0.0301,
         0.0078,  0.0110,  0.0080,  0.0163,  0.0301, -0.0038, -0.0585, -0.0109,
        -0.0022,  0.0022, -0.0601, -0.0006, -0.0236,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1120, -0.0869,  0.0302,  0.0157, -0.0583, -0.0377, -0.0174, -0.0599,
        -0.0518, -0.0537, -0.1865, -0.0630, -0.0066,  0.0531, -0.1046, -0.0625,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2018, -0.0679, -0.0474,  0.0420,  0.0253, -0.0821, -0.0442,  0.0227,
        -0.0320, -0.0702, -0.0114, -0.0472, -0.0464, -0.0162, -0.0334, -0.0054,
        -0.0176, -0.0314,  0.0045,  0.0099, -0.0157, -0.0052,  0.0105, -0.0138,
         0.0302,  0.0241, -0.0415,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1448, -0.0510, -0.0996,  0.0303, -0.0136,  0.0209,  0.0071,  0.0054,
         0.0036,  0.0340, -0.0043,  0.0123,  0.0055, -0.0022, -0.0071,  0.0084,
        -0.0020, -0.0047, -0.0056,  0.0364, -0.0238,  0.0405, -0.0238, -0.0041,
        -0.0389, -0.0060, -0.0010, -0.0047, -0.0301, -0.0015,  0.0008,  0.0054,
        -0.0153, -0.0195, -0.0521, -0.0212, -0.0296, -0.0057, -0.0311, -0.0229,
         0.0098, -0.0454, -0.0069,  0.0095,  0.0172,  0.0345], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0300, -0.0385, -0.0370, -0.0010, -0.0321, -0.0172, -0.0200, -0.0454,
        -0.0390, -0.0567, -0.0622, -0.0487, -0.0117, -0.0296, -0.0579,  0.0059,
         0.0308, -0.0479, -0.0123, -0.0479,  0.0076, -0.0386,  0.0098,  0.0039,
         0.0192,  0.0027, -0.0615,  0.0010,  0.1576, -0.0264,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0168,  0.1048,  0.0461, -0.1420,  0.0333,  0.1335, -0.0025, -0.0906,
        -0.1437,  0.2023, -0.0843,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1022, -0.0637, -0.0307, -0.1260, -0.0012, -0.0162, -0.0262, -0.0861,
        -0.0124, -0.1617,  0.0510,  0.0053, -0.0174, -0.0769, -0.0606, -0.0057,
        -0.0033,  0.0143,  0.0258, -0.1133,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0669,  0.0680, -0.0308, -0.0626, -0.0547, -0.1500, -0.0671, -0.4056,
        -0.0100, -0.0470,  0.0374,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1248, -0.0741, -0.0777, -0.0298,  0.0048, -0.0554, -0.0188, -0.0562,
        -0.0151, -0.0830, -0.0394, -0.0130, -0.0459, -0.0601, -0.0220, -0.0154,
        -0.0649, -0.0382, -0.0213, -0.0545, -0.0856,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0536, -0.0628, -0.0744, -0.0940, -0.0228, -0.1210,  0.1101,  0.1182,
         0.0411, -0.0569,  0.0358, -0.0161,  0.0115,  0.0070,  0.0292,  0.0382,
         0.0702, -0.0372,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0197, -0.0353, -0.0678,  0.0446,  0.0371, -0.0316, -0.0120,  0.0137,
        -0.0087, -0.0074,  0.0188,  0.0357, -0.0159, -0.0189, -0.0359,  0.0137,
        -0.0310, -0.0440,  0.0157, -0.0141, -0.0419, -0.0090, -0.0496, -0.0247,
        -0.0072,  0.0131, -0.0452,  0.0064, -0.0190, -0.0243,  0.0286, -0.0125,
        -0.0221,  0.0212,  0.0004,  0.0361, -0.0615,  0.0317, -0.0242,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0426, -0.0933,  0.0242,  0.0117, -0.0146, -0.0144,  0.0113,  0.0560,
         0.0319,  0.0010, -0.0388, -0.0092, -0.0249,  0.0581, -0.1050, -0.0155,
        -0.0037,  0.0416, -0.0067,  0.0095, -0.0537,  0.0065, -0.0120,  0.0010,
        -0.0239, -0.0212, -0.0139, -0.0286, -0.0533, -0.0454, -0.0048,  0.0063,
        -0.0418,  0.0735,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.3075, -0.1721,  0.0071, -0.0794, -0.0059, -0.0940,  0.0752, -0.1084,
        -0.1503,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0645, -0.0690, -0.0032,  0.0378,  0.0028,  0.0274,  0.0051, -0.0031,
         0.0696, -0.0195, -0.0085, -0.0346,  0.0156,  0.0017,  0.0049,  0.0260,
        -0.0026, -0.1170,  0.0054, -0.0071, -0.0236,  0.0245, -0.0239, -0.0341,
        -0.0104, -0.0381, -0.0334, -0.0268,  0.0169, -0.0132, -0.0272, -0.0403,
        -0.0136, -0.0403,  0.0109, -0.0370, -0.0161, -0.0019,  0.0254,  0.0169,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0801, -0.0174, -0.0460, -0.0103, -0.0327, -0.0049,  0.0080, -0.0028,
        -0.0215, -0.0215, -0.0038, -0.0191, -0.0015, -0.0060, -0.0111,  0.0656,
         0.0016, -0.0103, -0.0052, -0.0037, -0.0002, -0.0195,  0.0073,  0.0207,
         0.0452,  0.0149, -0.0280,  0.0245,  0.0263, -0.0651, -0.0307, -0.0133,
         0.0026, -0.0183, -0.0158, -0.1023, -0.0123, -0.0046,  0.0091,  0.0057,
        -0.0149,  0.0026, -0.0407,  0.0112, -0.0104,  0.0057, -0.0102,  0.0147,
        -0.0071, -0.0036,  0.0339, -0.0053,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0581, -0.0601, -0.0920,  0.0410,  0.0136,  0.0003, -0.0412,  0.0226,
         0.0006,  0.0021, -0.0074, -0.0316, -0.0560, -0.0215, -0.0881, -0.0452,
         0.0505,  0.0860, -0.0256, -0.0537, -0.0846, -0.1182,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1131, -0.2516, -0.0804,  0.0310, -0.0343,  0.0099,  0.0225, -0.0340,
         0.0155,  0.0141,  0.0631,  0.0276,  0.0353,  0.0189,  0.0165,  0.0092,
        -0.0140, -0.0141,  0.0413,  0.0494, -0.1039,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0242, -0.0800, -0.0316, -0.0207, -0.0359,  0.0024, -0.0454,  0.0175,
        -0.0184, -0.0299, -0.0287, -0.0155, -0.0267, -0.0419, -0.0021,  0.0135,
        -0.0053, -0.0164, -0.0298,  0.0002, -0.0059, -0.0101, -0.0112,  0.0202,
         0.0258,  0.0147, -0.0224,  0.0008, -0.0122, -0.0055, -0.0084,  0.0059,
        -0.0134, -0.0061, -0.0049,  0.0077,  0.0191, -0.0088, -0.0008, -0.0040,
        -0.0053, -0.0307, -0.0779, -0.0431, -0.0156, -0.0043, -0.0209, -0.0058,
         0.0085, -0.0292, -0.0194,  0.0105, -0.0080, -0.0205, -0.0066],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.2585e-02,  3.4071e-03,  6.9733e-02, -9.9880e-03, -3.1734e-03,
        -1.2873e-02,  4.5288e-02,  8.0529e-02,  2.4113e-05,  2.5605e-02,
         3.7162e-02,  2.6272e-02, -4.4622e-02, -5.8051e-03, -2.0163e-02,
         1.7599e-02,  4.5229e-02, -1.0535e-02,  9.2633e-03,  1.5391e-02,
        -2.7777e-02, -4.2130e-02, -4.1517e-02, -1.2714e-02,  5.5408e-02,
        -2.7933e-02,  1.7782e-02, -1.9630e-02,  2.4856e-02, -2.6960e-02,
         2.3987e-02, -1.4828e-02, -2.6808e-02, -3.6029e-02, -2.2159e-02,
        -6.6928e-03,  1.3517e-03,  7.1566e-03,  8.1398e-03,  3.1604e-03,
        -1.4838e-02,  6.2419e-03, -1.8197e-02,  1.5735e-02, -2.7221e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0951, -0.0955, -0.0633, -0.0009,  0.0612, -0.0172, -0.0250, -0.0494,
         0.0060,  0.0080, -0.0138,  0.0413,  0.0815,  0.0360,  0.0042,  0.0423,
         0.0182,  0.0197, -0.0075, -0.0113, -0.1243,  0.1242,  0.0041, -0.0500,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0926, -0.0447,  0.0281,  0.0247, -0.0325, -0.0127, -0.0560, -0.0200,
        -0.0383, -0.0168, -0.0843, -0.0149, -0.0136,  0.0203, -0.0109,  0.0123,
        -0.0281, -0.0254, -0.0098,  0.0003, -0.0187,  0.0137, -0.0052, -0.0572,
        -0.0858, -0.0006, -0.0440,  0.0199, -0.0109, -0.0440,  0.0155,  0.0698,
        -0.0286,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0294, -0.0573,  0.1278, -0.0711, -0.0045, -0.1021, -0.0647, -0.0051,
        -0.1161, -0.1012,  0.0317, -0.0074,  0.0151, -0.0798,  0.0312,  0.0461,
        -0.0071, -0.0086,  0.0934,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0246,  0.0128,  0.0972, -0.0212,  0.0612, -0.0143,  0.0159,  0.0102,
         0.0083,  0.0481,  0.0735,  0.0610,  0.0298,  0.0040,  0.0338,  0.0124,
         0.0542, -0.1095, -0.0098, -0.0005, -0.0247,  0.0089,  0.0285, -0.1294,
         0.1061,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1175,  0.0066, -0.0353,  0.0291, -0.1151, -0.0398, -0.1505,  0.1585,
         0.0203, -0.0181, -0.0367, -0.0672, -0.0298, -0.0226, -0.0922,  0.0136,
        -0.0471,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0925, -0.0572, -0.0015, -0.0131,  0.0268, -0.0190, -0.0008, -0.0042,
        -0.0413, -0.0631,  0.0395,  0.0360,  0.0137, -0.0147, -0.0372, -0.0399,
         0.0057, -0.0641,  0.0602,  0.0173,  0.0254, -0.0236, -0.0495, -0.0175,
        -0.0339,  0.0150, -0.0222,  0.0161,  0.0484, -0.0076, -0.0494, -0.0437,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0230, -0.0500, -0.0040,  0.0059, -0.0399, -0.0035, -0.0379, -0.0204,
        -0.0023, -0.0347, -0.0103, -0.0383,  0.0368, -0.0433, -0.0455,  0.0317,
        -0.0799, -0.0404, -0.0045,  0.0446, -0.0019,  0.0242, -0.0126, -0.0767,
         0.0226,  0.0239,  0.2117, -0.0295,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0368, -0.1222,  0.0731,  0.0064, -0.0992, -0.0768, -0.0639, -0.0147,
         0.0032,  0.0242,  0.0371, -0.0275, -0.0333, -0.0309, -0.0043,  0.0162,
        -0.0319, -0.0009, -0.0211,  0.0348,  0.0001, -0.0355, -0.0294, -0.0122,
        -0.0318,  0.0056,  0.0558,  0.0496,  0.0215,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.7421e-02, -7.8146e-05, -2.9496e-02, -5.3716e-02, -6.1550e-02,
        -2.4251e-02, -5.7295e-02,  1.9713e-02, -6.3905e-02, -3.6887e-02,
        -4.8439e-02, -2.2354e-02, -2.6266e-02, -5.7077e-03, -4.0969e-02,
        -1.3276e-02, -1.2643e-03, -4.6502e-03,  2.0553e-03,  4.0934e-03,
        -3.6580e-02, -1.4788e-02, -7.4646e-03, -4.0614e-02, -1.8791e-02,
         1.4094e-02, -2.5257e-02, -1.9653e-02,  9.8560e-02, -1.3517e-02,
        -3.1405e-02,  5.6329e-02, -7.9559e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0336, -0.0647,  0.0214, -0.0575, -0.1081, -0.0408, -0.0704, -0.0798,
        -0.0025, -0.0132,  0.0337, -0.0305, -0.0423, -0.0329, -0.0379, -0.0137,
        -0.0019,  0.0103, -0.0339,  0.0301, -0.0568, -0.0031,  0.1335, -0.0473,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1537,  0.0006, -0.0072,  0.0026, -0.0445, -0.1273, -0.1448, -0.0594,
        -0.0453, -0.0425, -0.0195,  0.0036, -0.0186, -0.0320, -0.0014, -0.0339,
        -0.0535, -0.0454,  0.0019, -0.0137,  0.0156, -0.0555, -0.0773,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.0197e-02, -6.3648e-02, -3.6860e-02, -4.8074e-05, -2.0008e-02,
        -4.3129e-02,  2.4530e-02, -2.5999e-03, -2.9268e-02, -5.3595e-02,
        -3.2998e-02, -5.5158e-02, -4.3098e-02, -2.2295e-02, -2.3577e-02,
         9.3821e-03, -2.3846e-03, -2.2024e-02, -2.0837e-02,  3.5066e-02,
        -2.9451e-02,  3.3169e-02,  9.5325e-03,  3.9749e-02, -1.7246e-02,
        -3.7411e-02, -1.1617e-02, -5.2433e-02, -1.8901e-02, -1.8875e-02,
        -2.2950e-02, -9.4823e-03, -9.6000e-03,  1.5568e-02, -6.5542e-03,
        -2.8491e-02, -9.1122e-03,  4.0539e-02,  8.6154e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.2428e-02, -4.7562e-02,  6.8478e-03,  1.1563e-02, -5.9736e-02,
         4.2346e-03,  1.8699e-02,  2.3388e-02,  2.5103e-02, -2.9178e-03,
        -5.5144e-03, -2.1956e-04,  2.8306e-02,  4.9547e-03, -2.8815e-03,
         1.2318e-03, -7.8512e-02, -1.5401e-02,  8.5991e-03,  2.4182e-02,
        -3.0642e-02, -7.6617e-03, -3.9035e-02,  6.3723e-03,  2.8442e-02,
        -2.3245e-02,  1.7686e-03, -1.8116e-03, -2.5803e-02,  6.2884e-03,
         9.6203e-05,  3.9786e-02,  8.6913e-02,  7.2929e-03, -9.4605e-03,
        -3.6375e-03,  2.1726e-02, -3.1310e-02,  3.4457e-03, -1.2098e-02,
        -2.3056e-02, -1.3590e-03,  1.6647e-01,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0245, -0.0312,  0.0092, -0.0009,  0.0535, -0.0102, -0.0091, -0.0094,
        -0.0541,  0.0245,  0.0381, -0.0056, -0.0185, -0.0158,  0.0046, -0.0408,
        -0.0060,  0.0104, -0.0032, -0.0368,  0.0034, -0.0019,  0.0040, -0.0037,
        -0.0151, -0.0014,  0.0342,  0.0298, -0.0051, -0.0046,  0.0210, -0.0022,
        -0.0088,  0.0488,  0.0271,  0.0209, -0.0035,  0.0029, -0.0243, -0.0048,
        -0.0130, -0.0293, -0.0035, -0.0665, -0.0023,  0.0007, -0.0052, -0.0128,
        -0.0027, -0.0041, -0.0405, -0.0172, -0.0733,  0.0143,  0.0283, -0.0056,
        -0.0070], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1713, -0.1693, -0.0679, -0.0081, -0.0987, -0.1302, -0.0321, -0.0367,
         0.0034,  0.0778,  0.0166,  0.1384,  0.0435, -0.0059,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1261, -0.0582, -0.0374, -0.1060,  0.0025,  0.0365, -0.0483, -0.0348,
        -0.0234,  0.0272, -0.0170, -0.0999, -0.0556, -0.0084, -0.0105,  0.0626,
         0.0080, -0.0136,  0.0099, -0.0047, -0.0126, -0.0344, -0.0325,  0.0070,
        -0.0093, -0.0185,  0.0088,  0.0285,  0.0500,  0.0077,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0643,  0.1808,  0.1606, -0.0742,  0.0491,  0.1695,  0.1620, -0.1396,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0545, -0.0229, -0.0433,  0.0317, -0.0445, -0.1113, -0.0357, -0.0354,
         0.0041, -0.0270, -0.0546, -0.0249, -0.0592, -0.0224, -0.0260, -0.0075,
         0.0042, -0.0017,  0.0151, -0.0198,  0.0311, -0.0103, -0.0183,  0.0447,
        -0.0503, -0.0386, -0.0204,  0.0111,  0.0057, -0.0282,  0.0305, -0.0062,
         0.0588,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0316, -0.0095,  0.0187,  0.0493,  0.0012,  0.0074, -0.0027, -0.0053,
        -0.0242, -0.0051,  0.0001,  0.0044, -0.0037,  0.0024,  0.0014, -0.0005,
         0.0032,  0.0107, -0.0012, -0.0088,  0.0093, -0.0097,  0.0222,  0.0264,
         0.0085, -0.0077,  0.0099, -0.0167,  0.0156,  0.0248,  0.0059, -0.0040,
         0.0009, -0.0236,  0.0204, -0.0084,  0.0024, -0.0064,  0.0046,  0.0219,
        -0.0062,  0.0225, -0.0088, -0.0205,  0.0462,  0.0231,  0.0176, -0.0225,
         0.0137, -0.0004, -0.0159, -0.0073, -0.0130, -0.0111, -0.0127,  0.0029,
        -0.0049, -0.0703, -0.0165,  0.0078,  0.0036, -0.0135, -0.0083, -0.0703,
         0.0089,  0.0175,  0.0096,  0.0267, -0.0050,  0.0403, -0.0117],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.8653e-02,  2.5279e-02, -3.1840e-02,  3.4268e-02, -2.6338e-02,
        -1.2904e-02, -1.6593e-01, -1.6362e-02, -3.3686e-02,  1.2902e-02,
         7.8023e-03,  1.4270e-02, -6.0575e-03,  6.6890e-04,  3.1206e-03,
        -1.5282e-02,  4.1619e-03,  3.7022e-02, -1.3757e-02, -2.5128e-02,
        -8.7256e-03, -2.0778e-03,  5.9998e-02, -1.4808e-02,  9.0111e-03,
         4.9541e-03, -2.3128e-02,  2.5035e-02,  2.8942e-02,  3.0664e-02,
        -1.3570e-02,  1.6737e-02, -5.6467e-03, -8.6184e-03, -1.2308e-02,
        -2.5983e-02, -4.7849e-03, -1.5533e-02, -2.7993e-03, -1.2946e-02,
         9.3310e-05, -1.1820e-01,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0117, -0.0548, -0.0080, -0.0667,  0.0285, -0.0143,  0.0072,  0.0071,
         0.0087,  0.0206, -0.0111,  0.0150,  0.0030, -0.0078,  0.0161, -0.0476,
        -0.0018, -0.0267, -0.0497,  0.0005, -0.0134,  0.0283, -0.0427,  0.0560,
        -0.0121, -0.0026, -0.0585,  0.0325, -0.0262,  0.0167,  0.0064, -0.0114,
         0.0205,  0.0314,  0.0093,  0.0754,  0.1497,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1209, -0.0663,  0.0172, -0.0137, -0.1570,  0.1568, -0.0341, -0.0790,
        -0.0108, -0.1520, -0.0278,  0.0332, -0.0526, -0.0406, -0.0381,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2121,  0.0126, -0.0890, -0.0180, -0.0526,  0.0099,  0.0172, -0.0355,
         0.0443,  0.0130, -0.0185, -0.1252,  0.0383,  0.0106, -0.0840,  0.0291,
        -0.0243,  0.0599, -0.0217, -0.0841,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.4635e-02,  2.5748e-02, -7.2694e-02,  4.8940e-02, -2.9311e-02,
        -3.1061e-02, -2.1436e-02, -3.9597e-02, -6.6884e-03,  2.1578e-02,
        -1.2055e-02, -5.3098e-03, -2.5200e-02,  6.0041e-03, -9.2673e-03,
        -1.5861e-02, -2.9524e-02, -8.4761e-05,  2.2170e-03, -6.0300e-03,
         6.4776e-04,  5.1674e-03,  3.2924e-02,  1.0450e-02,  5.2430e-03,
         7.8707e-03, -1.8210e-02,  1.2468e-02, -1.3793e-02,  1.2261e-02,
        -2.0360e-02, -3.7458e-02,  1.4781e-03,  2.2553e-03,  8.1000e-03,
        -4.9032e-03, -2.1924e-02,  8.8142e-03, -3.0897e-02,  8.0939e-03,
         4.4682e-02,  9.1106e-02, -1.1765e-01,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0539,  0.0089, -0.0697, -0.0882, -0.0048, -0.0067,  0.0051, -0.0880,
         0.0132,  0.0156,  0.0098, -0.0364, -0.0072,  0.0310, -0.0171,  0.0092,
        -0.0032, -0.0339,  0.0002, -0.0293,  0.0263,  0.0098, -0.0082,  0.0105,
         0.0147,  0.0223, -0.0296, -0.0056, -0.0169,  0.0621,  0.0129, -0.0194,
        -0.0121, -0.0080,  0.0043,  0.0366,  0.0796, -0.0896,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1299, -0.0590, -0.0661, -0.0965,  0.0427, -0.0417,  0.0189, -0.0799,
         0.0459, -0.0719, -0.0178, -0.0111, -0.0310,  0.0463, -0.0030, -0.0368,
        -0.0341,  0.0190, -0.0203, -0.0183, -0.0713,  0.0387,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0744,  0.0480,  0.0403, -0.0425,  0.0444, -0.0398, -0.0397,  0.0089,
        -0.0307, -0.0188, -0.0127,  0.0173, -0.0057, -0.0085,  0.0021, -0.0007,
        -0.0331, -0.0362, -0.0163, -0.0186, -0.0168,  0.0094, -0.0118, -0.0207,
        -0.0483, -0.0165, -0.0033,  0.0492, -0.0144, -0.0258, -0.0150, -0.0360,
        -0.0069, -0.0328, -0.0084, -0.0489,  0.0166, -0.0137,  0.0485,  0.0180,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0525e-01,  1.2931e-02,  6.1449e-03, -1.6546e-02,  5.2882e-03,
        -2.7514e-02,  4.4180e-02,  2.0590e-02, -2.4151e-02, -4.3832e-02,
         3.4043e-02,  2.6575e-03, -8.2180e-05, -1.6002e-02, -6.4829e-02,
        -3.8208e-02,  3.4763e-02, -6.3601e-02, -4.0327e-02, -1.3299e-02,
         2.1323e-02, -1.9715e-03, -1.4344e-02,  1.8438e-02,  2.9451e-02,
        -1.7642e-02, -2.9641e-02, -2.5295e-02, -7.2557e-02, -1.9842e-02,
         6.3242e-02, -7.2016e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1159, -0.0571, -0.0652,  0.0347,  0.0013, -0.0066, -0.0151,  0.0068,
        -0.0262, -0.0149, -0.0223,  0.0151,  0.0187, -0.0078,  0.0059, -0.0123,
        -0.0268, -0.0391, -0.0152, -0.0010, -0.0020, -0.0749, -0.1356, -0.0551,
        -0.0337, -0.0316,  0.0018, -0.0031, -0.0115,  0.0058, -0.0046, -0.0193,
        -0.0062,  0.0595,  0.0472,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-4.7754e-02, -2.9148e-01, -1.0872e-01, -3.8891e-02,  3.3127e-02,
         3.5276e-02,  2.2013e-02, -4.7573e-02, -7.7917e-02, -3.9141e-02,
         1.3176e-01,  1.6867e-02, -1.1193e-02, -2.4915e-03,  2.2816e-04,
         1.0978e-02,  1.2978e-02,  2.7314e-03,  8.0808e-03, -4.0008e-02,
        -2.0792e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1298e-01, -8.1369e-02, -1.2802e-02, -1.1152e-01, -1.0884e-01,
        -3.0325e-02, -1.3872e-02, -2.0827e-02, -5.0625e-02, -5.5062e-02,
        -2.1931e-02,  1.1476e-02,  8.2425e-02, -6.0674e-02, -2.9485e-02,
        -7.5865e-05, -6.0254e-02,  1.2064e-01, -1.4823e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.4068e-02, -3.5251e-01, -8.0498e-02, -3.7576e-02, -1.1763e-02,
         2.3960e-02, -9.7717e-04, -5.0422e-02,  7.5572e-02, -1.1514e-01,
        -7.3230e-02,  8.1064e-03,  1.6771e-02, -4.3722e-02,  1.8808e-02,
        -2.4540e-02,  2.2363e-02, -2.5189e-02, -4.4616e-03,  3.2871e-04,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1114, -0.1520, -0.0733,  0.0147, -0.0028,  0.0124,  0.1042, -0.0581,
         0.0102,  0.0191, -0.0265, -0.0267, -0.0013, -0.0504, -0.0683,  0.0072,
        -0.0417,  0.0180, -0.0216, -0.0157,  0.0378, -0.0395,  0.0872,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2055, -0.0457, -0.0424,  0.0720, -0.0199, -0.0320, -0.0028, -0.0117,
        -0.0618, -0.1103, -0.0030,  0.0080,  0.0041, -0.0025,  0.0409, -0.0282,
        -0.0426,  0.0227,  0.0309,  0.0182, -0.0768, -0.0005,  0.0211, -0.0348,
         0.0618,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.1459e-02, -1.5928e-01,  5.0611e-02,  8.1982e-05,  5.2531e-03,
        -1.7506e-02, -2.6573e-03, -2.7094e-02, -1.0409e-02, -5.6381e-02,
        -1.3868e-02, -2.4081e-02,  3.5614e-04, -5.5844e-02,  2.4227e-02,
        -1.0226e-02,  7.9574e-03, -1.2231e-02,  3.4936e-02,  8.6243e-03,
        -3.4412e-02, -2.3851e-02, -1.8217e-02, -2.1938e-02, -1.0369e-02,
         3.9978e-03, -2.0818e-02, -4.3303e-02, -3.0083e-02,  1.3537e-02,
        -1.4010e-02,  9.6498e-03, -1.8569e-02,  2.0801e-02, -1.4336e-01,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1188,  0.0826,  0.0761, -0.0294, -0.0093,  0.0780, -0.0613,  0.0071,
        -0.1565, -0.0230, -0.0038,  0.1044,  0.0160, -0.0435, -0.0389, -0.0454,
        -0.0149, -0.0145, -0.0147,  0.0177,  0.0027,  0.0413,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0192,  0.0347,  0.0165, -0.0767,  0.0034, -0.0577, -0.0060,  0.0328,
        -0.0380, -0.0533, -0.1427, -0.0041,  0.0490, -0.0924, -0.0916, -0.0077,
         0.0036,  0.0164, -0.0700, -0.1409, -0.0433,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2056, -0.1223, -0.0885, -0.0282, -0.0179, -0.0570, -0.0220, -0.0121,
         0.0191, -0.0495, -0.0007,  0.0985, -0.0166, -0.0141,  0.0251, -0.0141,
        -0.0851, -0.0211, -0.1026,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0549,  0.0848, -0.0148,  0.0161,  0.0165, -0.0172,  0.0176,  0.0222,
         0.0028,  0.0166,  0.0198,  0.0110, -0.0122, -0.0024, -0.0174,  0.0274,
         0.0049, -0.0059,  0.0018, -0.0114,  0.0136,  0.0105, -0.0342, -0.0067,
        -0.0680, -0.0569,  0.0270, -0.0087,  0.0109, -0.0358,  0.0073, -0.0101,
        -0.0179, -0.0275,  0.0067,  0.0165,  0.0238, -0.0040, -0.0409, -0.0120,
        -0.0061, -0.0145, -0.0070,  0.0031, -0.0146,  0.0869, -0.0507],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0550, -0.0425, -0.0414, -0.0090,  0.0091, -0.0531,  0.0094,  0.0061,
        -0.0048, -0.0167, -0.0027,  0.0162,  0.0141,  0.0351, -0.0333, -0.0217,
        -0.0870, -0.0237,  0.0275,  0.0128, -0.0023,  0.0511,  0.0271, -0.0948,
        -0.0311,  0.0445, -0.0273,  0.0166, -0.0255, -0.0167,  0.0131,  0.0481,
         0.0389, -0.0417,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.0618e-01, -3.2011e-02,  6.6249e-03,  2.6804e-02,  5.9182e-02,
        -4.6018e-02, -3.2142e-02, -1.7074e-02, -3.1759e-02, -1.6007e-02,
         2.8307e-02, -1.8183e-02, -2.6677e-02, -2.2619e-02, -2.9365e-02,
        -3.9127e-02, -3.3097e-02,  1.4494e-01, -4.1088e-02, -1.4786e-02,
        -8.6631e-03, -8.4639e-05, -1.0299e-02, -1.7376e-03, -1.4022e-02,
        -4.2444e-03, -6.6605e-03, -1.4870e-02, -1.9110e-03, -1.6938e-02,
         3.4385e-02, -2.2789e-02,  1.6280e-02, -7.5126e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0703, -0.1509,  0.1541, -0.0396, -0.0060, -0.0388, -0.0605, -0.0986,
        -0.0178, -0.0697, -0.0525, -0.0296, -0.0175, -0.0314, -0.0543, -0.0486,
         0.0193, -0.0407,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0529,  0.0366, -0.0812, -0.0123, -0.0112,  0.0042, -0.0141, -0.0468,
        -0.0082, -0.0155,  0.0546,  0.0398,  0.0223, -0.0053, -0.0251,  0.0791,
         0.0545, -0.0224,  0.0275, -0.1759, -0.0344, -0.0163, -0.0272, -0.0414,
        -0.0040,  0.0436,  0.0433,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.5726e-02, -1.3340e-01,  1.3015e-02, -3.5412e-02, -1.9245e-01,
        -1.9123e-02, -6.9414e-02, -8.8081e-02, -8.4590e-02, -3.0938e-02,
        -9.4446e-03, -1.0015e-02, -2.6609e-02, -2.6142e-02, -7.4768e-03,
        -7.1415e-03, -1.5960e-02, -3.1291e-05, -2.0421e-02, -1.9629e-02,
        -2.3866e-03, -1.3410e-02, -2.9656e-02,  1.6311e-02,  2.4130e-02,
         1.9087e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1425, -0.0758, -0.0235, -0.0044, -0.0854, -0.0027,  0.0462, -0.0384,
        -0.0383, -0.0057, -0.0483,  0.0208, -0.0539, -0.0174,  0.0497, -0.0487,
         0.0255, -0.0467, -0.1393,  0.0709, -0.0158,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0356,  0.0114, -0.1536, -0.1727, -0.0423,  0.0169,  0.1416, -0.0051,
        -0.0410,  0.0721, -0.0259, -0.0299, -0.0232, -0.1026, -0.1259,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.0456e-01, -1.8648e-02, -7.6866e-03, -8.6523e-02,  1.3815e-01,
        -1.2186e-02,  1.9653e-04, -1.9370e-02,  3.3688e-02, -1.6912e-01,
        -9.0792e-03,  2.9372e-02, -1.0357e-02, -4.7434e-03,  5.8285e-02,
         3.7432e-02,  2.4434e-01,  1.6271e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1410, -0.0560, -0.0428, -0.2230, -0.0711, -0.0342,  0.0585, -0.0609,
         0.0143, -0.0961,  0.0090,  0.0113,  0.0403,  0.0356, -0.0062, -0.0082,
        -0.0117, -0.0374, -0.0426,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1215, -0.0364,  0.0064,  0.0169,  0.0263, -0.0446,  0.0388,  0.0056,
        -0.0097,  0.0185,  0.0368,  0.0151, -0.0159,  0.0065,  0.0401,  0.0537,
        -0.0713, -0.0304,  0.0259,  0.0328, -0.0198,  0.0290, -0.0178,  0.0066,
         0.0098,  0.0114,  0.0107, -0.0186,  0.0015, -0.0098, -0.0042, -0.0026,
         0.0045, -0.0016, -0.0195, -0.0068, -0.0040,  0.0545, -0.0076, -0.0044,
         0.0051,  0.0240,  0.0657,  0.0071], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2195, -0.0133, -0.0187, -0.0217,  0.0069, -0.0134,  0.0073,  0.0226,
        -0.0385,  0.0337, -0.0193,  0.0128, -0.0086,  0.0069,  0.0263, -0.0241,
        -0.0089, -0.0625,  0.0166,  0.0171, -0.0205,  0.0011,  0.0095, -0.0021,
        -0.0174, -0.0045,  0.0023, -0.0126,  0.0197,  0.0847, -0.0047, -0.0528,
        -0.0250, -0.1096,  0.0350,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0193,  0.0896, -0.0234,  0.0320,  0.0151,  0.1578,  0.0696, -0.0473,
        -0.0341, -0.0046,  0.0582, -0.0050, -0.0125,  0.0331,  0.0248,  0.0493,
         0.0323,  0.0101,  0.0433,  0.0568,  0.0232,  0.0060, -0.1174, -0.0352,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0334, -0.0155, -0.0083, -0.0590,  0.0116, -0.0375, -0.0064, -0.0442,
         0.0074,  0.0438, -0.0390, -0.0219,  0.0233, -0.0089, -0.0144, -0.0197,
        -0.0303,  0.0034, -0.0559, -0.0644, -0.0169, -0.0331, -0.0309,  0.0040,
         0.0532, -0.0418, -0.0558, -0.0220,  0.0085, -0.0259, -0.0323,  0.0051,
        -0.0656, -0.0566,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2183,  0.0107,  0.0094, -0.0927, -0.0022,  0.0345, -0.0709, -0.0276,
        -0.0198,  0.0222,  0.0147,  0.0239, -0.0335,  0.0057, -0.0193,  0.0136,
         0.0579, -0.0104,  0.0011, -0.0477,  0.0111,  0.0019,  0.0734, -0.0081,
         0.0090,  0.0266,  0.0104, -0.0067, -0.1153,  0.0016,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0394,  0.2061, -0.1529, -0.0494,  0.0176, -0.0073,  0.0385, -0.0254,
        -0.0088,  0.0042, -0.0636, -0.2630,  0.1236,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0658, -0.1649,  0.1369,  0.0003, -0.0701, -0.0494,  0.0388, -0.0196,
        -0.0593, -0.0031, -0.0406, -0.0505,  0.0648, -0.0073, -0.0073, -0.0718,
        -0.1468,  0.0027,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2436,  0.0145, -0.1751, -0.1977, -0.1691,  0.1667,  0.0008, -0.0325,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0405,  0.0470,  0.0118,  0.0113,  0.0179,  0.0043,  0.0019,  0.0202,
        -0.0025,  0.0103, -0.0025,  0.0040, -0.0279, -0.0366, -0.0027, -0.0132,
        -0.0050,  0.0084, -0.0155, -0.0145,  0.0851,  0.0151,  0.0300, -0.0367,
         0.0172, -0.0228, -0.0273, -0.0457, -0.0288,  0.0110,  0.0166,  0.0012,
         0.0193, -0.0553,  0.0312, -0.0309, -0.0170, -0.0113,  0.0316, -0.0051,
        -0.0072,  0.0044, -0.0263,  0.0383,  0.0409, -0.0105,  0.0353],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1196,  0.0625, -0.1431, -0.1943, -0.0251,  0.1150, -0.0980,  0.0388,
        -0.1044, -0.0659, -0.0333,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.0238e-01,  1.2058e-04, -1.1836e-01, -2.6106e-02, -1.3293e-01,
        -7.4729e-02, -4.5302e-02, -1.3279e-01, -3.5450e-02, -4.4399e-02,
        -2.6555e-02, -6.9268e-02, -2.2116e-02, -3.0638e-02,  2.9039e-02,
        -5.2812e-03,  8.3499e-02,  2.1051e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1316,  0.0033, -0.1537, -0.0363, -0.0027,  0.0518, -0.0009, -0.0097,
         0.0098, -0.0523, -0.0521,  0.0109, -0.0905,  0.1506, -0.2437,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2518, -0.0769,  0.0792,  0.0123, -0.0645, -0.0025, -0.1564,  0.0865,
        -0.0164,  0.0811, -0.0203,  0.0115,  0.1018,  0.0063, -0.0325,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1694, -0.0384,  0.0768, -0.0247, -0.0065,  0.0156,  0.0060, -0.0263,
        -0.0683, -0.0479, -0.0322, -0.0273, -0.0340,  0.0145, -0.0083,  0.0295,
        -0.0356,  0.0305,  0.0081,  0.0043, -0.0637, -0.0299,  0.0659, -0.0634,
        -0.0304, -0.0424,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.1564e-02, -3.4970e-01, -9.5636e-02, -4.8592e-02, -6.2651e-02,
        -2.7900e-04,  1.8872e-03, -4.9067e-02, -9.9032e-02, -1.3328e-02,
        -2.9597e-02,  4.6571e-03, -1.0579e-02, -4.3092e-02, -1.0034e-01,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0685, -0.0274,  0.0099,  0.0184, -0.0679, -0.0382,  0.0064, -0.0087,
         0.0066, -0.1579,  0.0353, -0.0036, -0.0065, -0.0074, -0.1152, -0.0142,
        -0.0152, -0.0217, -0.0249, -0.0539, -0.0349, -0.0669, -0.0270,  0.0486,
         0.0886, -0.0263,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1603,  0.0609, -0.0890, -0.0054, -0.0353, -0.0163,  0.0223,  0.0329,
         0.0438, -0.0020, -0.0450, -0.0280, -0.0138, -0.0280, -0.0201, -0.0338,
        -0.0299, -0.0302,  0.0222, -0.0394,  0.0191,  0.0027, -0.0154, -0.0376,
        -0.0898, -0.0769,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.0914, -0.0619, -0.0601, -0.0317, -0.0238, -0.0055, -0.0299, -0.0331,
         0.0172, -0.0364,  0.0646, -0.0262, -0.0401,  0.0195, -0.0185, -0.0047,
        -0.0080, -0.0041, -0.0066, -0.0340, -0.0226, -0.0088,  0.0049, -0.0241,
        -0.0115, -0.0649, -0.0322,  0.0336, -0.0424,  0.0286, -0.0021, -0.0119,
        -0.0017, -0.0093, -0.0459,  0.0382,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0454, -0.0509,  0.0368, -0.0195,  0.0101,  0.0031, -0.0327, -0.0298,
        -0.0017,  0.0165, -0.0122,  0.0255, -0.0274,  0.0114, -0.0366, -0.0410,
        -0.0082,  0.0107, -0.0523, -0.0167, -0.0207,  0.0987, -0.0425,  0.0435,
        -0.0050, -0.0322, -0.0098, -0.0159, -0.0100,  0.0210, -0.0004, -0.0284,
        -0.0280, -0.0279, -0.0160, -0.1113,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0512,  0.1703,  0.0365,  0.0087, -0.0407,  0.0832, -0.0016,  0.0249,
        -0.0231, -0.0390,  0.0105,  0.0574,  0.0380, -0.0009,  0.0988, -0.0072,
         0.0828,  0.0979, -0.0766, -0.0508,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1821,  0.0009, -0.0185,  0.0497,  0.1167, -0.0234,  0.0372,  0.0012,
        -0.0370,  0.0096,  0.0097,  0.0027,  0.0024,  0.0232, -0.0057,  0.0074,
         0.0139, -0.0136, -0.0171, -0.0069, -0.0359,  0.0050, -0.0004, -0.0185,
         0.0520,  0.0059,  0.0141, -0.0064, -0.0008, -0.0517,  0.0139, -0.0163,
        -0.0502, -0.0135,  0.0285,  0.0668,  0.0314, -0.0097,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0209, -0.0001,  0.0147,  0.0413, -0.0073, -0.0339, -0.0033,  0.0756,
        -0.0117, -0.0359, -0.0138, -0.0645, -0.0028, -0.0080, -0.0649, -0.0387,
         0.0046, -0.0101,  0.0096,  0.0237,  0.0047,  0.0386,  0.0181, -0.0431,
         0.0345,  0.0031, -0.0080, -0.0065, -0.0852, -0.0451, -0.0169,  0.0402,
         0.0199,  0.0050, -0.0134, -0.0017, -0.0330, -0.0390, -0.0124, -0.0230,
         0.0230], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0034, -0.1225,  0.1041,  0.0320, -0.0218,  0.0970, -0.1228, -0.0203,
         0.0331,  0.0293,  0.0176,  0.0123, -0.0167,  0.0024, -0.1132, -0.0618,
        -0.0287, -0.0377, -0.0305, -0.0549, -0.0107, -0.0273,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0063, -0.0114, -0.0102, -0.0234,  0.0144, -0.0402, -0.0133, -0.0309,
        -0.0330,  0.0045, -0.0250, -0.0427,  0.0147, -0.0187,  0.0435, -0.0215,
        -0.0356, -0.0193,  0.0093, -0.0636, -0.0337, -0.0540,  0.0283, -0.0023,
        -0.0173,  0.0074, -0.0054, -0.0291, -0.0729, -0.0835,  0.0040, -0.0326,
        -0.0321,  0.0169, -0.0992,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0080, -0.0704, -0.0375, -0.0346, -0.0949,  0.0491, -0.0231,  0.0530,
         0.0170,  0.0099, -0.0391,  0.0339, -0.0015,  0.0198,  0.0653, -0.0735,
         0.0319, -0.0009, -0.0247,  0.0511, -0.0281,  0.0309, -0.0206, -0.0050,
        -0.0384, -0.0884,  0.0043,  0.0045,  0.0406,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2442,  0.0186, -0.0282,  0.0217,  0.0043, -0.0542,  0.0216, -0.0257,
        -0.0040, -0.0377,  0.0176, -0.0568,  0.0234, -0.0174, -0.0300, -0.0054,
        -0.0822, -0.0164, -0.0473, -0.0109, -0.0049, -0.1377,  0.0581, -0.0312,
        -0.0005,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0324, -0.0689,  0.0346, -0.0379, -0.1022, -0.0070, -0.0275, -0.0345,
        -0.0097,  0.0877, -0.0173, -0.0770, -0.0343, -0.0794, -0.0236, -0.0543,
         0.0034, -0.0134, -0.0100, -0.0154, -0.0484, -0.1174,  0.0637,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.9506e-03, -4.8745e-02, -3.8434e-02, -4.2558e-02, -3.8362e-02,
         7.0037e-02,  7.4281e-02, -2.4308e-02, -3.4738e-02,  5.3654e-04,
         1.0342e-02,  2.4878e-04, -4.3538e-03, -2.9328e-02,  4.4750e-02,
         2.3253e-02,  2.3354e-02,  6.5561e-05, -2.0927e-02, -1.3505e-02,
        -1.5136e-02, -3.8330e-02, -3.5274e-02, -1.2106e-02, -5.5572e-02,
        -4.7661e-02, -3.4655e-02,  1.5961e-02, -1.1935e-02, -1.9058e-02,
        -1.3982e-02, -7.3746e-03,  2.0495e-03, -2.4091e-02, -2.1207e-02,
        -5.3550e-02, -1.8547e-02, -1.5055e-02, -8.3775e-03,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0054, -0.2019,  0.0040, -0.0614, -0.1449, -0.0250, -0.0439, -0.0910,
        -0.2068, -0.1222, -0.0191,  0.0745,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0710, -0.0617, -0.1764, -0.0129, -0.0053, -0.0275, -0.0044, -0.0292,
        -0.0100, -0.0182, -0.0416, -0.0281, -0.0050,  0.0227, -0.0033, -0.0168,
        -0.0078, -0.0189,  0.0095, -0.0311,  0.0034, -0.0209, -0.0044, -0.0231,
         0.0374,  0.0291, -0.0132,  0.0029, -0.0679,  0.0174,  0.0033,  0.0058,
        -0.0065, -0.0063, -0.0099, -0.0045, -0.0142,  0.0210, -0.0207, -0.0089,
        -0.0076,  0.0153, -0.0157,  0.0077,  0.0316,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.3181, -0.1327,  0.0431, -0.0424,  0.1309,  0.0179, -0.0586, -0.0536,
        -0.0822, -0.0255,  0.0058, -0.0892,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0676, -0.4133, -0.0955,  0.0043,  0.1756, -0.0135,  0.0403, -0.1899,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.3428e-01, -1.3986e-02, -9.2426e-03,  4.3237e-03,  8.8884e-03,
        -1.9829e-03, -1.9272e-02,  1.5705e-03,  3.2315e-03, -4.8147e-02,
        -5.0446e-02,  2.2743e-02,  4.9637e-03, -5.0301e-04, -5.6918e-03,
         8.3564e-03, -4.0125e-02, -6.3866e-03,  9.7881e-03,  1.6909e-03,
        -2.2991e-03, -3.6416e-03,  2.3849e-03,  1.0305e-02, -1.0985e-02,
        -8.7093e-03, -4.1947e-03, -2.3907e-03, -1.4064e-03,  1.5492e-02,
        -1.9117e-02, -1.1181e-02, -7.6067e-03, -9.6116e-03, -3.5684e-03,
         2.5367e-03,  5.4900e-02, -1.3773e-02, -3.1926e-05, -1.4152e-02,
        -6.4102e-03,  4.5933e-03,  2.3653e-02, -5.3267e-03,  6.0367e-02,
         7.7778e-02, -1.8760e-02,  3.9092e-02, -1.5597e-02, -2.4000e-02,
        -2.8295e-02, -1.1206e-02, -1.2352e-02,  3.3760e-02,  5.6603e-03,
         5.2982e-03,  4.4468e-03, -3.9935e-03, -6.3715e-03,  2.6443e-03,
         1.6486e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0077, -0.0656, -0.0059, -0.0073, -0.0808, -0.0523, -0.0023,  0.0144,
         0.0734,  0.0464,  0.0367,  0.0154,  0.0135,  0.0195, -0.0876, -0.0179,
         0.0101,  0.0016,  0.0013,  0.0316, -0.0104,  0.0384, -0.0169,  0.0329,
         0.0157, -0.0165, -0.0045, -0.0103, -0.0150, -0.0416,  0.0161,  0.0953,
         0.0951,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1078,  0.0272, -0.0841,  0.0145, -0.0283,  0.0525, -0.0035,  0.0094,
         0.0173, -0.0623, -0.0070, -0.0136, -0.0102,  0.0153, -0.0561,  0.0152,
        -0.0739, -0.0511, -0.1245,  0.0049,  0.0482, -0.1118,  0.0465, -0.0150,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0950, -0.0847, -0.0512, -0.0242, -0.0033, -0.0979, -0.0198, -0.0205,
         0.0481,  0.0316, -0.0212,  0.0173, -0.0816, -0.0050, -0.0257, -0.0151,
         0.0243,  0.0294, -0.0014,  0.0093, -0.0090,  0.0538,  0.0785,  0.1522,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0939,  0.0038, -0.0352, -0.0346, -0.0082,  0.0135, -0.0271, -0.0104,
         0.0283, -0.0016, -0.0279,  0.0259, -0.0928, -0.0054,  0.0044, -0.0170,
         0.0017,  0.0157, -0.0017,  0.0283, -0.0282, -0.0117, -0.0010, -0.0885,
        -0.0475,  0.0912,  0.0084, -0.0060,  0.0992,  0.0115, -0.0271, -0.0101,
         0.0472,  0.0450,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0546, -0.0181, -0.0319,  0.0034,  0.0065,  0.0959,  0.0196,  0.0089,
         0.0147, -0.1005, -0.1055, -0.0768, -0.0499, -0.1215,  0.0799,  0.0073,
         0.0697, -0.0013, -0.0201, -0.0031, -0.0417, -0.0263, -0.0194, -0.0210,
         0.0024,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1448, -0.1236,  0.0187, -0.0771, -0.0021,  0.0849, -0.0473,  0.0211,
        -0.0394, -0.1307, -0.0583, -0.0488, -0.0022, -0.0321, -0.1266,  0.0423,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.9499e-01, -1.1223e-01, -9.6126e-03, -3.5192e-02, -2.3531e-02,
        -4.6800e-02, -7.4344e-03, -3.1722e-03,  4.9059e-02, -1.8342e-02,
         2.1189e-02, -4.8976e-02, -6.7486e-03, -2.4512e-02, -4.1053e-02,
         1.1353e-02, -9.3285e-02, -2.1041e-02,  9.4840e-03,  1.3144e-04,
        -2.7694e-02, -3.3374e-02, -6.0791e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1541, -0.2950, -0.0362, -0.0614, -0.0661,  0.0109,  0.0188, -0.0250,
        -0.0162, -0.0107,  0.1247, -0.0093,  0.0911,  0.0806,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.2544, -0.0752,  0.0284, -0.0572,  0.0099,  0.0068, -0.0823, -0.0885,
        -0.0020,  0.0352, -0.0593,  0.0038,  0.1065, -0.0196,  0.0016, -0.0107,
        -0.0067, -0.0402, -0.0034, -0.0050,  0.0036, -0.0356, -0.0523,  0.0118,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0973,  0.3289,  0.0231, -0.0348,  0.0613,  0.0850, -0.1192, -0.0282,
         0.0215,  0.0115,  0.1891,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.0207e-02, -1.3851e-01, -2.3753e-03,  3.8022e-03, -1.3512e-01,
        -6.5876e-02, -1.4781e-01,  6.8510e-02,  3.2856e-03,  1.4217e-04,
        -1.9082e-02, -2.1417e-02, -3.2403e-02, -8.7949e-02, -4.2059e-02,
        -3.2964e-02,  3.2959e-04,  1.7760e-02,  5.5631e-03,  5.2702e-02,
        -9.2138e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0211,  0.0710, -0.1504, -0.0938,  0.0833,  0.0519, -0.1204, -0.0289,
        -0.0174, -0.0224,  0.1151, -0.0050,  0.1046, -0.1147,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0321,  0.0692,  0.0018,  0.0903,  0.0130,  0.1096,  0.0192,  0.0374,
        -0.0153,  0.0058, -0.0154, -0.1875, -0.0772, -0.0046,  0.0245,  0.0334,
         0.0318, -0.0072, -0.0765, -0.0065, -0.1417,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0134, -0.0195, -0.0050,  0.0063,  0.0914, -0.0390,  0.0083, -0.2464,
        -0.0200, -0.1711, -0.2168, -0.0688, -0.0941,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0188, -0.0231, -0.0123,  0.0077, -0.0125, -0.0104,  0.0141, -0.0251,
        -0.0388,  0.0221, -0.0218, -0.0033,  0.0203, -0.0265, -0.0211, -0.0150,
        -0.0107, -0.0155,  0.0096, -0.0084, -0.0134,  0.0305, -0.0508, -0.0347,
        -0.1256,  0.0264,  0.0100, -0.0106, -0.0193,  0.0052, -0.0009,  0.0226,
         0.0179,  0.0272, -0.0473,  0.0887, -0.0157,  0.0086, -0.0041, -0.0009,
         0.0004,  0.0013,  0.0042,  0.0481, -0.0482], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0221, -0.0368, -0.0385, -0.0389, -0.0312, -0.0222, -0.0157, -0.0362,
        -0.0132, -0.0382, -0.0201,  0.0125, -0.0345,  0.0026,  0.0003,  0.0031,
        -0.1249,  0.0122, -0.0334,  0.0075,  0.0240, -0.0481,  0.0192,  0.0246,
         0.0130,  0.0427, -0.0622, -0.0263, -0.0566, -0.0499,  0.0098,  0.0577,
        -0.0113,  0.0104,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0096, -0.2178, -0.0097, -0.0759, -0.0513, -0.0089, -0.0375, -0.0794,
         0.0047, -0.0467, -0.0052, -0.0268, -0.0277, -0.0267, -0.0389, -0.0051,
         0.0090, -0.0253, -0.0249, -0.0210, -0.1079,  0.0082, -0.0219,  0.0448,
        -0.0189,  0.0464,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0017, -0.0925,  0.1659, -0.0312,  0.1472,  0.0107,  0.0735, -0.0267,
         0.0506, -0.0135, -0.0286,  0.0034, -0.0089, -0.0414, -0.0894,  0.0197,
        -0.0277,  0.0669, -0.0029, -0.0230, -0.0205, -0.0458,  0.0082,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0209, -0.0200,  0.0804, -0.0081, -0.0124,  0.0621,  0.0018, -0.0142,
        -0.0058,  0.0180,  0.0208, -0.0196,  0.0307, -0.0433, -0.1225,  0.0776,
         0.0569, -0.0046, -0.0020,  0.0069, -0.0003, -0.0008,  0.0027,  0.0128,
         0.0093,  0.0196, -0.0040, -0.0546, -0.0225,  0.0121,  0.0654,  0.0004,
         0.0362, -0.0343,  0.0143,  0.0228, -0.0010, -0.0078, -0.0188, -0.0316,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0522, -0.0324, -0.0529,  0.0035, -0.0022,  0.0068,  0.0337, -0.0258,
        -0.0802, -0.0307,  0.0110, -0.0276, -0.0015, -0.0393, -0.0115, -0.0306,
         0.0009,  0.0238,  0.0339,  0.0011, -0.0299,  0.0133,  0.0208,  0.0715,
        -0.0375, -0.0171,  0.0068, -0.0475,  0.0026,  0.1265, -0.0610, -0.0456,
        -0.0177,  0.0002,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0891,  0.0190, -0.0881,  0.0004, -0.0262, -0.0034,  0.0233, -0.0073,
         0.0304,  0.0176,  0.0068, -0.0078,  0.1325,  0.0048, -0.0363,  0.0404,
         0.0196,  0.0078, -0.0061, -0.0119, -0.0098, -0.0027,  0.0309,  0.0338,
        -0.0186, -0.0157, -0.0531, -0.0264, -0.0358,  0.0128,  0.0087, -0.0034,
         0.0243,  0.0375,  0.0009, -0.0091,  0.0067, -0.0019,  0.0018, -0.0012,
         0.0019, -0.0110, -0.0051, -0.0027, -0.0015,  0.0044,  0.0017,  0.0093,
         0.0017,  0.0018,  0.0238, -0.0218], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0581, -0.0704,  0.0015,  0.0098, -0.0808, -0.0492, -0.0886,  0.0022,
         0.0016,  0.0325, -0.0215,  0.0388, -0.0803, -0.0347, -0.0340,  0.0003,
        -0.0625, -0.0086, -0.0080,  0.0622, -0.0080,  0.0092,  0.0021, -0.0677,
         0.0190,  0.0112,  0.0169, -0.0218, -0.0441,  0.0101,  0.0445,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2098, -0.0754,  0.0002,  0.0355, -0.0060, -0.0368, -0.0224,  0.0007,
         0.0170,  0.0461, -0.0234, -0.0195, -0.0930, -0.0025, -0.0394, -0.0223,
        -0.0987, -0.0184, -0.0239, -0.0608, -0.0207, -0.0259, -0.0244, -0.0159,
         0.0614,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2190, -0.0157,  0.0456, -0.0392, -0.0988, -0.0280, -0.1105, -0.0570,
         0.0499, -0.0487,  0.1169, -0.1706,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 6.6984e-02, -1.2958e-01, -5.7201e-02, -1.7842e-02, -2.8039e-02,
         1.6349e-02, -1.3411e-02,  9.2838e-03, -1.1293e-02, -1.1677e-04,
        -1.5755e-02,  9.1543e-03, -2.8624e-03, -7.4575e-02, -1.2780e-02,
        -5.3815e-02, -2.6495e-02, -1.0202e-02,  1.5313e-02, -2.1277e-02,
        -4.1368e-03, -3.1784e-02,  2.8820e-02,  1.5999e-01,  6.3248e-02,
         6.4726e-02, -4.0607e-02,  1.3175e-02,  1.1895e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0891, -0.0079, -0.0264,  0.0568, -0.0199, -0.0217, -0.0571,  0.0632,
         0.1697, -0.3742, -0.0987,  0.0155,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.2763e-01, -2.7200e-01,  1.2832e-04, -4.5137e-02, -1.5597e-01,
        -6.0950e-02, -1.4161e-01, -7.0941e-02, -2.8520e-02, -1.1987e-02,
        -5.8153e-02,  2.6971e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2350,  0.0323, -0.0490,  0.0260, -0.0548, -0.0170,  0.0028,  0.0352,
        -0.0573,  0.0467, -0.0104, -0.0466,  0.0340,  0.0525, -0.0215, -0.0231,
         0.0092,  0.0203, -0.1749, -0.0512,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.3142e-02, -7.3419e-03, -3.9071e-02, -7.0782e-02, -6.7773e-02,
        -4.9076e-02, -2.5242e-02,  5.1847e-04, -1.6622e-02, -5.2805e-03,
         2.2415e-02, -1.0730e-02, -1.6340e-02,  5.2020e-02,  4.5465e-02,
        -2.6785e-03, -9.4695e-03,  2.8800e-02,  2.1851e-02, -1.8325e-02,
        -1.7164e-02,  1.8485e-02,  9.7199e-03, -1.8289e-02, -7.3237e-03,
         3.2412e-02,  1.7319e-02, -1.2508e-02, -2.8154e-02, -3.0936e-05,
         2.6827e-02,  1.6329e-03, -3.2976e-02, -9.6830e-02, -1.3540e-02,
         4.7336e-02, -9.6510e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1031, -0.3323, -0.1947, -0.0461, -0.0810,  0.0263,  0.1771,  0.0173,
        -0.0221,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1942,  0.0156,  0.0383, -0.0976, -0.0016, -0.0599, -0.0493, -0.0522,
        -0.0034,  0.0129,  0.0229,  0.1076, -0.0020, -0.0186,  0.0173,  0.0066,
        -0.0207,  0.0517, -0.0292, -0.0016, -0.0171,  0.0428,  0.0067, -0.0241,
        -0.0079, -0.0574, -0.0409,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1350, -0.0435, -0.0221, -0.0178, -0.0168, -0.0027, -0.0115, -0.0155,
        -0.0078, -0.0395, -0.0356,  0.0024, -0.0492,  0.0462,  0.0106, -0.0091,
        -0.0197, -0.0166, -0.0228,  0.0295,  0.0366, -0.0488, -0.0125,  0.0423,
         0.0087, -0.0013, -0.0178, -0.0088, -0.1631, -0.0170,  0.0012, -0.0304,
         0.0390, -0.0188,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0305, -0.0242, -0.0035, -0.0418, -0.0069,  0.0146,  0.0003, -0.0154,
         0.0142,  0.0162, -0.0115, -0.0614, -0.0304, -0.0235,  0.0222, -0.0127,
        -0.0685, -0.0246,  0.0014,  0.0280,  0.0222,  0.0125, -0.0262,  0.0040,
        -0.0138,  0.0091, -0.0141, -0.0165, -0.0431, -0.0232,  0.0197,  0.0074,
        -0.0358,  0.0603,  0.0474, -0.0058,  0.0015, -0.0139,  0.0053,  0.0125,
        -0.0183, -0.0385,  0.0384,  0.0445, -0.0144,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1983, -0.0169,  0.0050,  0.0339, -0.1206, -0.0700, -0.1729, -0.0795,
        -0.0272,  0.0211,  0.0348, -0.0027, -0.0359, -0.0007,  0.0333, -0.1470,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1440, -0.0191, -0.0385, -0.0551,  0.0308,  0.0467,  0.0029,  0.0634,
        -0.0268, -0.0377, -0.0238, -0.0522, -0.0576, -0.0162, -0.0181,  0.0309,
        -0.0320,  0.0411,  0.0403,  0.0273, -0.0390, -0.0099, -0.0222, -0.0002,
         0.0393, -0.0380,  0.0467,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0204,  0.2202, -0.0166, -0.0229, -0.0083,  0.0055, -0.0023,  0.0041,
        -0.0009, -0.0148, -0.0050, -0.0100, -0.0104, -0.0185, -0.0037, -0.0106,
         0.0016,  0.0223, -0.0313,  0.0064,  0.0034, -0.0103, -0.0105, -0.0247,
         0.0307, -0.0105, -0.0081, -0.0146, -0.0268,  0.0142,  0.0065,  0.0119,
         0.0098,  0.0007,  0.0379,  0.0511, -0.0069, -0.0313, -0.0211, -0.0233,
        -0.0324, -0.0164, -0.0202,  0.0098, -0.1166,  0.0144], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0427, -0.0309,  0.0075, -0.0283, -0.0119,  0.0184, -0.0294, -0.0238,
         0.0086, -0.0516, -0.0105, -0.0127,  0.0027, -0.0286, -0.0162, -0.0286,
        -0.0320, -0.0639, -0.0727, -0.1579, -0.0159,  0.0103,  0.0173, -0.0087,
        -0.0182,  0.0182, -0.0229, -0.1316,  0.0066,  0.0714,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2010,  0.1954, -0.0317,  0.0027, -0.0294, -0.1428, -0.0670, -0.0911,
         0.0690, -0.1273, -0.0426,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1386, -0.1371, -0.0669, -0.0172, -0.0273, -0.0056, -0.0035, -0.0516,
        -0.0268, -0.1955, -0.0140, -0.0380, -0.0241, -0.0147, -0.0052,  0.0137,
        -0.0005, -0.0100,  0.0480, -0.1616,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0667, -0.1968,  0.0236, -0.1499, -0.0377, -0.0864, -0.1359, -0.0242,
        -0.1935, -0.0027, -0.0827,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1184, -0.0344, -0.0525, -0.1129, -0.1174,  0.0713, -0.0071, -0.0563,
        -0.0491,  0.0670, -0.0490, -0.0698, -0.0169, -0.0242, -0.0035, -0.0107,
        -0.0344, -0.0409,  0.0098, -0.0356, -0.0190,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0440, -0.0466, -0.0266, -0.0868, -0.0217, -0.2838, -0.0386,  0.0558,
         0.0145,  0.2297, -0.0301, -0.0085,  0.0374,  0.0164, -0.0046,  0.0348,
         0.0069,  0.0134,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.7090e-02, -8.0350e-02, -1.1224e-01, -4.4360e-02,  1.8166e-02,
        -7.4476e-05, -1.6895e-02,  2.4665e-03, -7.6277e-03, -1.3172e-02,
         2.6862e-02,  5.5756e-02, -1.8149e-03, -1.1155e-02, -3.2626e-02,
        -3.2804e-02, -5.8822e-03, -4.0382e-02, -3.4176e-03, -7.8617e-03,
        -2.0764e-02, -2.7258e-02, -5.8485e-02,  1.6503e-02, -2.4578e-02,
        -2.6122e-02, -2.1645e-02, -1.6360e-02, -1.4901e-02, -1.9969e-02,
         4.3139e-03, -7.8867e-03, -3.0305e-02,  5.3239e-03, -3.0459e-02,
         1.2231e-02, -2.8462e-02, -7.1969e-03, -8.6235e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0836, -0.1576,  0.0127,  0.0585, -0.0127, -0.0400,  0.0272, -0.0440,
        -0.0071,  0.0111, -0.0440, -0.0243, -0.0293,  0.0187, -0.0196,  0.0097,
         0.0101,  0.0235,  0.0177, -0.0185, -0.0092, -0.0400,  0.0153, -0.0194,
        -0.0006,  0.0005, -0.0292,  0.0101, -0.0296, -0.0777,  0.0107,  0.0139,
        -0.0058, -0.0680,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.2581, -0.1816, -0.0589, -0.0335, -0.1114, -0.1493, -0.0840, -0.0413,
        -0.0818,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.9110e-02, -1.8858e-02,  7.7159e-03,  2.6661e-03, -1.2354e-02,
         4.2039e-02, -4.2554e-03, -2.5728e-02, -7.2812e-03, -1.1672e-02,
        -2.2667e-02, -7.3149e-03,  7.0461e-03, -1.7109e-03,  7.4550e-02,
        -9.4088e-03, -3.0298e-02, -2.1857e-01, -4.6293e-03, -2.5916e-02,
         1.8657e-02,  1.5363e-03, -2.7296e-02,  1.3678e-02, -2.7735e-02,
         9.2816e-03,  6.0592e-04, -8.2751e-03,  4.6972e-03,  1.8113e-03,
         7.4727e-03,  1.2261e-04, -1.1325e-02, -2.1972e-02,  1.8086e-02,
        -2.5204e-02, -3.9601e-04, -7.1144e-02,  3.3095e-02, -6.3815e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0356, -0.0375, -0.0670,  0.0003, -0.0338, -0.0002, -0.0005, -0.0324,
         0.0050, -0.0141, -0.0017,  0.0081, -0.0030, -0.0010,  0.0015,  0.0461,
        -0.0226,  0.0371, -0.0072, -0.0032, -0.0121,  0.0081,  0.0203, -0.0087,
         0.0206, -0.0044, -0.0553, -0.0116,  0.0416, -0.0734, -0.0455, -0.0503,
        -0.0032,  0.0083, -0.0169,  0.0042, -0.0053, -0.0205,  0.0032,  0.0231,
         0.0317,  0.0071,  0.0014, -0.0014, -0.0052,  0.0040, -0.0085,  0.0137,
        -0.0156,  0.0262,  0.0289,  0.0617,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0518, -0.0692, -0.0785, -0.0538,  0.0404, -0.0301, -0.0454, -0.0537,
        -0.0118,  0.0353, -0.0554, -0.0221, -0.0242, -0.0643, -0.0030, -0.0504,
        -0.0165, -0.0676, -0.0656, -0.0616, -0.0510,  0.0483,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0872, -0.1972, -0.0135, -0.0152,  0.0150, -0.0300, -0.0275, -0.0169,
         0.0632, -0.0413, -0.0645, -0.0436, -0.0051, -0.0364, -0.0314, -0.0380,
        -0.0556,  0.0416,  0.0172,  0.0266,  0.1331,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1329,  0.0020,  0.0054, -0.0536,  0.0249, -0.0022, -0.0213, -0.0107,
        -0.0087,  0.0115,  0.0002,  0.0127,  0.0032,  0.0154,  0.0060,  0.0051,
         0.0024,  0.0141, -0.0287, -0.0109, -0.0021, -0.0190,  0.0076, -0.0016,
        -0.0145,  0.0154,  0.0064, -0.0211,  0.0003,  0.0083, -0.0166, -0.0024,
         0.0038, -0.0124, -0.0130, -0.0298,  0.0144, -0.0043, -0.0153, -0.0376,
        -0.1365, -0.0120, -0.0254, -0.0162, -0.0068, -0.0004, -0.0307, -0.0297,
         0.0109, -0.0064, -0.0198, -0.0199,  0.0161,  0.0409,  0.0105],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.8356e-02,  2.1563e-03,  4.8921e-02, -4.2030e-02, -5.4798e-03,
        -1.0293e-02,  4.8013e-02,  4.4350e-02,  7.0092e-03, -1.7786e-05,
         1.2996e-02,  1.3724e-02, -1.1696e-02, -1.0967e-02,  6.9660e-03,
         1.2799e-02,  4.3746e-02, -2.5923e-02,  5.8927e-03, -2.6357e-02,
        -3.5245e-02, -3.0361e-02, -2.9419e-02, -6.1165e-03,  2.2784e-02,
        -5.2740e-03, -2.5149e-02, -7.7382e-02,  2.2699e-03, -1.5440e-02,
        -3.5054e-02, -1.2466e-02, -1.4068e-02, -8.7325e-02, -4.1342e-02,
        -1.1890e-02,  1.2167e-02, -1.8663e-02,  1.0490e-02,  2.2326e-03,
        -1.4380e-02,  2.7051e-03,  1.7751e-02, -2.4688e-02, -1.7644e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0366, -0.1075, -0.0913, -0.0553, -0.0231,  0.0044, -0.0186,  0.0036,
        -0.0313,  0.0108,  0.0434,  0.0047,  0.0246, -0.0066, -0.0291, -0.0141,
        -0.0581, -0.1011,  0.0240, -0.0332, -0.0669,  0.0678,  0.1273, -0.0164,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0177,  0.0359,  0.0097,  0.0802,  0.0828,  0.0350,  0.0340, -0.0038,
         0.0192,  0.0023, -0.0323,  0.0281,  0.0228,  0.0102,  0.0227,  0.0579,
         0.0235,  0.0168,  0.0179,  0.0428, -0.0212, -0.0136,  0.0297,  0.0645,
         0.0969,  0.0007,  0.0422, -0.0162,  0.0232,  0.0478,  0.0213,  0.0106,
        -0.0162,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0423, -0.1221, -0.0454, -0.1152, -0.0104, -0.0807, -0.0284, -0.0020,
        -0.1465, -0.0251,  0.0086, -0.0119, -0.0526, -0.0459, -0.0314, -0.0915,
        -0.0403,  0.0129,  0.0868,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2242, -0.0010, -0.0222, -0.0242, -0.0387,  0.0357, -0.0028,  0.0006,
        -0.0559,  0.0560,  0.0430, -0.0581,  0.0290, -0.0394,  0.0220,  0.0121,
         0.0025, -0.0152,  0.0089,  0.0101, -0.0392, -0.0876, -0.0060, -0.0499,
         0.1158,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1127, -0.0234, -0.0982,  0.0380,  0.0463, -0.0259, -0.3005,  0.0560,
         0.0263,  0.0072, -0.0255, -0.0120, -0.0066, -0.0247, -0.0842,  0.0934,
         0.0190,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0785,  0.0603, -0.0380, -0.0071, -0.0192, -0.0133, -0.0245,  0.0114,
        -0.1734,  0.0023,  0.0009,  0.0167, -0.0027, -0.0375, -0.0123, -0.0497,
        -0.0063, -0.0478,  0.0278,  0.0105,  0.0200, -0.0060, -0.0523,  0.0071,
         0.0129, -0.0219, -0.0087,  0.0501,  0.0085,  0.0134,  0.0028, -0.1560,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1580,  0.0549, -0.0126, -0.0132,  0.0040,  0.0163, -0.0045,  0.0083,
         0.0042, -0.0052, -0.0205,  0.0912, -0.0625, -0.0840, -0.1041,  0.0751,
         0.0195, -0.0228, -0.0069, -0.0142, -0.0138, -0.0280,  0.0097, -0.0142,
         0.0488,  0.0010,  0.0468, -0.0557,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.0359e-01,  1.8584e-01, -2.4321e-02,  1.3267e-02, -1.6774e-03,
         6.0343e-03,  4.2392e-02,  1.9061e-02,  2.6814e-02,  3.3066e-02,
         3.4138e-03,  3.1825e-02,  1.6838e-02, -2.5113e-02,  4.8783e-02,
         1.4441e-04,  1.4079e-02,  1.1586e-02,  2.6427e-02,  2.5980e-02,
         2.2428e-02,  3.3796e-02,  6.0847e-03,  6.2950e-02,  1.6974e-02,
        -1.5759e-02, -2.5028e-02,  1.0923e-01, -4.7499e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1236, -0.0269,  0.0172, -0.0093, -0.0177, -0.0521,  0.0277, -0.0460,
         0.0028, -0.0484, -0.0244, -0.0167,  0.0165, -0.0708,  0.0082, -0.0093,
        -0.0321, -0.0671,  0.0039,  0.0370, -0.0160, -0.0268,  0.0047,  0.0126,
        -0.0506, -0.0196, -0.0021, -0.0193,  0.0549,  0.0079, -0.0024, -0.0498,
         0.0755,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1054,  0.0414,  0.0004, -0.0196, -0.0573, -0.0230,  0.0154, -0.0229,
         0.0449, -0.0617, -0.0678, -0.0704, -0.0234, -0.0385, -0.0515, -0.0218,
        -0.0209, -0.0501, -0.0659, -0.0668, -0.0355,  0.0102,  0.0523,  0.0328,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1982, -0.1037, -0.0497, -0.0038, -0.0031, -0.0818, -0.0808, -0.0127,
         0.0106, -0.0038, -0.0192, -0.0180, -0.0107, -0.0431, -0.0075,  0.0184,
        -0.0189, -0.0148, -0.0140, -0.1810, -0.0145,  0.0009, -0.0908,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0469, -0.0347,  0.0231, -0.0095, -0.0051, -0.0212, -0.0045,  0.0156,
         0.0071,  0.0093, -0.0416, -0.0129, -0.0303, -0.0396, -0.0494,  0.0351,
        -0.0093, -0.0287, -0.0282,  0.0166,  0.0069,  0.0158,  0.0030, -0.0086,
        -0.0258, -0.0163,  0.0091, -0.0215, -0.0830, -0.0247, -0.0329, -0.0286,
        -0.0280, -0.0269,  0.0029, -0.0183, -0.0497, -0.0820, -0.0474,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0287, -0.0517,  0.0043, -0.0119, -0.0254,  0.0057, -0.0099,  0.0032,
        -0.0299,  0.0570,  0.0067, -0.0376,  0.0317,  0.0170,  0.0024, -0.0153,
        -0.0151, -0.0176,  0.0313, -0.0194, -0.0279,  0.0100, -0.0009,  0.0351,
         0.0911,  0.0193,  0.0021,  0.0030, -0.0008, -0.0427,  0.0018,  0.0111,
        -0.0732, -0.0070, -0.0556,  0.0229,  0.0098, -0.0105, -0.0146, -0.0010,
        -0.0062, -0.0728,  0.0591,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0218, -0.0169,  0.0144,  0.0114,  0.0258, -0.0364, -0.0403, -0.0293,
        -0.0404,  0.0171, -0.0172, -0.0042, -0.0119, -0.0158, -0.0189,  0.0085,
        -0.0081, -0.0167,  0.0047, -0.0118,  0.0125,  0.0094, -0.0229, -0.0089,
         0.0069,  0.0030, -0.0044, -0.0028,  0.0294, -0.0202, -0.0081,  0.0014,
        -0.0305, -0.0023, -0.0418,  0.0254,  0.0259,  0.0254, -0.0288, -0.0011,
        -0.0170, -0.0246,  0.0139,  0.0152,  0.0038,  0.0123, -0.0092, -0.0153,
        -0.0025, -0.0374, -0.0349,  0.0147, -0.0719, -0.0018,  0.0009,  0.0309,
         0.0112], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1115, -0.2302, -0.1813, -0.0244, -0.0451, -0.0335,  0.0171, -0.0448,
         0.1502,  0.0017, -0.0442, -0.0428, -0.0121, -0.0612,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1582, -0.0377, -0.0772, -0.0246, -0.0099,  0.0120, -0.0344, -0.0010,
        -0.0083, -0.0134, -0.0205, -0.1117, -0.0484,  0.0016, -0.0148, -0.0574,
        -0.0099, -0.0406, -0.0247, -0.0156, -0.0846, -0.0144, -0.0206, -0.0172,
         0.0012, -0.0030,  0.0126, -0.0415,  0.0375, -0.0454,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0371,  0.0224, -0.0119, -0.1310, -0.0176,  0.3298,  0.2549, -0.1954,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0513, -0.0225, -0.0007, -0.0138, -0.0513, -0.1375, -0.0341, -0.0060,
         0.0080, -0.0250, -0.0254,  0.0480, -0.0398, -0.0304,  0.0150, -0.0114,
         0.0536, -0.0088,  0.0157, -0.0262,  0.0321, -0.0053, -0.0105,  0.0103,
        -0.0081, -0.0125, -0.0110,  0.0501, -0.0505, -0.0589, -0.0074,  0.1144,
        -0.0046,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.3583e-02,  5.2034e-03, -1.2493e-02,  3.9314e-02, -9.8180e-04,
        -1.6492e-03,  5.2204e-03, -5.2758e-03,  4.6579e-02, -1.3448e-04,
        -9.3536e-03,  8.4672e-03,  5.3938e-03,  6.1966e-03,  1.1928e-02,
        -6.3993e-04,  6.2668e-03,  1.2697e-02, -7.0520e-05, -7.1729e-04,
         2.3260e-02,  3.9871e-02, -3.7738e-02, -9.8556e-04,  1.8916e-03,
        -9.3432e-03, -7.9492e-04, -1.2111e-02, -1.7633e-02,  1.0297e-02,
         4.1440e-04, -1.8769e-04, -9.8493e-03, -2.5049e-02, -1.6469e-03,
        -8.3915e-03, -9.8141e-04, -1.2967e-03,  1.1941e-02,  2.6036e-02,
         1.4747e-02, -2.6216e-02, -1.7283e-02,  2.2650e-02,  4.1205e-02,
         6.7288e-03,  4.7051e-02, -2.0493e-02,  4.1387e-02,  4.2156e-02,
        -3.9990e-02, -1.4275e-03,  3.8205e-03,  4.9956e-03, -9.0906e-03,
        -6.0446e-03,  1.7585e-02, -1.2862e-03, -8.3300e-03,  6.3759e-03,
        -9.2469e-03, -1.3861e-02, -5.0083e-02, -9.1804e-03,  2.8500e-03,
         2.5377e-04,  6.8793e-03,  1.1877e-02,  1.2624e-02,  3.9055e-02,
        -2.3341e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0045, -0.0205, -0.0148,  0.0166, -0.0263, -0.0816, -0.2085, -0.0099,
        -0.0054,  0.0007, -0.0112,  0.0079, -0.0284,  0.0041, -0.0013, -0.0203,
        -0.0144,  0.0215, -0.0012, -0.0340,  0.0181, -0.0087,  0.0085, -0.0076,
         0.0041,  0.0154, -0.0187, -0.0135,  0.0235,  0.0045,  0.0089,  0.0172,
        -0.0314, -0.0164,  0.1052, -0.0211, -0.0262, -0.0171, -0.0038, -0.0017,
        -0.0075, -0.0879,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0767, -0.1136, -0.0083, -0.0272,  0.0469,  0.0036,  0.0020,  0.0167,
        -0.0431, -0.0244, -0.0616, -0.0189, -0.0333,  0.0042,  0.0198,  0.0120,
         0.0011, -0.0123, -0.0534, -0.0142, -0.0169, -0.0104,  0.0106, -0.0087,
        -0.0174, -0.0385, -0.0017,  0.0006, -0.0604,  0.0277,  0.0470, -0.0140,
         0.0868,  0.0047,  0.0261,  0.0063,  0.0289,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0366,  0.2512,  0.1000,  0.0148, -0.0422, -0.1267,  0.0649, -0.0882,
        -0.0539,  0.0323,  0.0391,  0.0753,  0.0227, -0.0243,  0.0277,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0090,  0.0959,  0.0769,  0.0519,  0.0183, -0.0209,  0.0364, -0.0008,
         0.0435,  0.0326, -0.0592, -0.0232,  0.1547,  0.0419,  0.0677,  0.0691,
         0.0675,  0.0585,  0.0638,  0.0083,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.4138e-02, -8.0013e-02, -4.2767e-02, -9.8960e-03, -7.2243e-02,
        -5.0083e-02, -3.1676e-02,  2.3716e-02,  1.0405e-02, -1.6669e-02,
        -1.6583e-02,  2.6331e-03, -3.1581e-02, -1.1946e-02, -5.3068e-03,
        -2.1827e-03,  1.8772e-02, -1.7591e-02,  1.0027e-02, -1.5161e-02,
         1.3842e-03,  6.5152e-05, -1.6093e-02, -2.5396e-02, -1.0518e-02,
        -1.7325e-02, -3.2725e-02, -2.3517e-03,  3.2665e-03, -1.3814e-02,
        -1.9651e-02, -6.2306e-02, -4.4342e-03, -2.0096e-02,  3.5334e-03,
        -1.5783e-02, -7.9677e-03, -3.7182e-03, -4.2481e-02, -5.3836e-02,
        -1.2364e-03,  1.4724e-01,  1.3951e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0795, -0.0466,  0.0229, -0.0959, -0.0378,  0.0009,  0.0079, -0.0281,
        -0.0029,  0.0187, -0.0402, -0.0290, -0.0183, -0.0083, -0.0529,  0.0116,
         0.0124, -0.0067,  0.0206,  0.0812, -0.0148, -0.0177, -0.0269, -0.0087,
         0.0107,  0.0365,  0.0275,  0.0077, -0.0497,  0.0403, -0.0292, -0.0264,
        -0.0024, -0.0146, -0.0228,  0.0073,  0.0216, -0.0130,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0287, -0.0592, -0.1177, -0.1298, -0.0565, -0.0368, -0.0701,  0.0752,
         0.0382,  0.0376, -0.0494, -0.0276, -0.0176,  0.0109,  0.0298, -0.0194,
        -0.1006,  0.0113, -0.0081, -0.0129, -0.0181, -0.0443,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0145, -0.0469,  0.0337,  0.0113,  0.0207,  0.0783, -0.0197,  0.0136,
        -0.0465, -0.0213,  0.0069,  0.0085,  0.0061, -0.0157, -0.0040, -0.0160,
        -0.0770, -0.0191, -0.0530, -0.0314,  0.0558,  0.0030, -0.0047, -0.0282,
        -0.0187,  0.0202, -0.0340,  0.0127,  0.0097, -0.0353,  0.0358, -0.0080,
        -0.0263,  0.0075, -0.0206, -0.0270, -0.0121, -0.0434, -0.0306,  0.0220,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0607,  0.0121, -0.0117, -0.0290, -0.0131, -0.0156, -0.0353, -0.0245,
        -0.0114, -0.1090, -0.0291, -0.0342,  0.0111, -0.0102, -0.0276, -0.0695,
        -0.0252,  0.0750, -0.0419, -0.0130, -0.0367,  0.0141, -0.0089, -0.0719,
        -0.0166, -0.0130, -0.0149, -0.0134, -0.0168, -0.0298, -0.0113, -0.0933,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0005, -0.1491,  0.0584,  0.0026, -0.0694,  0.0738, -0.0491, -0.0568,
        -0.1054,  0.0027,  0.0051,  0.0128,  0.0184, -0.0104, -0.0393,  0.0760,
        -0.0226, -0.0010, -0.0043,  0.0479, -0.0050,  0.0054,  0.0082, -0.0297,
        -0.0060,  0.0303, -0.0343, -0.0121, -0.0012, -0.0082, -0.0079, -0.0003,
         0.0233,  0.0100,  0.0125,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0898, -0.0948, -0.0081,  0.0223,  0.0016, -0.0078, -0.0195, -0.2096,
        -0.0789, -0.1602,  0.0332, -0.0344, -0.0365, -0.0022, -0.0192,  0.0174,
        -0.0008, -0.0419, -0.0444, -0.0099, -0.0673,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0927, -0.1787,  0.0437, -0.0901, -0.0305, -0.0779,  0.1313, -0.0463,
        -0.1365, -0.0126, -0.0063,  0.0071,  0.0018,  0.0041, -0.0341,  0.0029,
        -0.0231,  0.0460,  0.0343,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0250, -0.3364, -0.0416, -0.0445, -0.0145, -0.0155,  0.0122, -0.0075,
        -0.0122, -0.0785, -0.0437, -0.0152,  0.0217, -0.0495,  0.0081, -0.0191,
         0.0159,  0.0006, -0.0463,  0.1920,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0118, -0.1452,  0.0026, -0.0259,  0.0075,  0.0486,  0.0714, -0.0110,
        -0.0350, -0.0118, -0.0596,  0.0056, -0.0156, -0.0459, -0.0463, -0.0267,
        -0.0869, -0.0157, -0.0083, -0.0477,  0.0183, -0.1215,  0.1310,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7325e-01, -3.2256e-02, -1.4715e-02,  1.9749e-05, -2.8902e-02,
        -1.4001e-03, -1.8495e-02, -6.5393e-03, -7.8788e-02,  7.7333e-03,
         2.2876e-02, -3.8764e-04,  1.2557e-01, -1.9487e-02,  3.8309e-02,
        -1.1595e-02, -4.4123e-02,  2.4218e-02,  2.3618e-02,  2.9086e-02,
         2.5907e-02,  2.8362e-03,  5.2978e-02, -2.1363e-01,  3.2825e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1167,  0.2187,  0.0411, -0.0505, -0.0021,  0.0068, -0.0329, -0.0128,
        -0.0711,  0.0288,  0.0272,  0.0045,  0.0083,  0.0946, -0.0516,  0.0161,
         0.0279,  0.0030,  0.0047, -0.0046,  0.0031,  0.0155,  0.0031,  0.0067,
         0.0036,  0.0094,  0.0103,  0.0074,  0.0130,  0.0095,  0.0283,  0.0025,
         0.0215, -0.0382,  0.0041,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1715,  0.0051, -0.0490, -0.0456, -0.0110, -0.0135, -0.0533,  0.0507,
        -0.1336, -0.0041, -0.0444, -0.0268,  0.0263, -0.0113, -0.0567, -0.0409,
         0.0146, -0.0234,  0.1010,  0.0094, -0.0426, -0.0650,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1229,  0.0907,  0.0421, -0.0011,  0.0338, -0.0293,  0.0094, -0.0138,
        -0.0154, -0.0054,  0.0044,  0.0025, -0.0394, -0.0043, -0.1733, -0.0426,
        -0.1015, -0.0300, -0.0664, -0.1568, -0.0151,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0615, -0.2063, -0.0600,  0.0343,  0.0096,  0.0366, -0.0510, -0.1090,
        -0.0027, -0.0017, -0.0661,  0.0443, -0.0962, -0.0804, -0.0510,  0.0093,
        -0.0079, -0.0272, -0.0448,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0392, -0.1138, -0.0113, -0.0484, -0.0291, -0.0523,  0.0449,  0.0056,
         0.0069,  0.0104, -0.0103, -0.0178,  0.0027, -0.0163,  0.0204,  0.0120,
        -0.0009, -0.0156,  0.0135, -0.0161, -0.0005, -0.0333, -0.0254, -0.0132,
         0.0273, -0.0151,  0.0078, -0.0369, -0.0063, -0.0764, -0.0654, -0.0143,
        -0.0031, -0.0392, -0.0024,  0.0028,  0.0146,  0.0062, -0.0046, -0.0013,
         0.0107, -0.0270,  0.0033,  0.0094, -0.0075, -0.0359, -0.0226],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2215, -0.0602, -0.0413,  0.0048,  0.0187, -0.0018,  0.0113, -0.0042,
        -0.0258,  0.0095,  0.0042,  0.0149, -0.0107,  0.0113,  0.1063, -0.0299,
         0.0218, -0.0101, -0.0082,  0.0147,  0.0109, -0.0517, -0.0043, -0.0386,
        -0.0210,  0.0322,  0.0291,  0.0314, -0.0261,  0.0011,  0.0190,  0.0019,
        -0.0632, -0.0382,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1091, -0.0437,  0.0019,  0.0022,  0.0115, -0.0311,  0.0309, -0.0115,
        -0.0004, -0.0118,  0.0616,  0.0376, -0.0076, -0.1588,  0.0147,  0.0094,
         0.0093,  0.0603,  0.0497, -0.0488, -0.0396,  0.0043,  0.0034,  0.0511,
        -0.0241, -0.0125, -0.0348, -0.0025,  0.0237, -0.0260,  0.0175,  0.0151,
        -0.0158, -0.0179,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.1056, -0.1829, -0.1038, -0.0448, -0.0334, -0.0686, -0.0148, -0.0377,
        -0.0318, -0.0639, -0.0932, -0.0639, -0.0239, -0.0644, -0.0015, -0.0361,
        -0.0130,  0.0166,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2324,  0.0083, -0.0093, -0.0238, -0.0322,  0.0123,  0.0179,  0.0004,
        -0.0077, -0.0700,  0.0455, -0.0390, -0.0312, -0.0312, -0.0129, -0.0205,
        -0.0680,  0.0107, -0.0251, -0.1016,  0.0143,  0.0611, -0.0200, -0.0552,
        -0.0028,  0.0445,  0.0022,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1760,  0.0832,  0.0480,  0.0016,  0.0379, -0.0335,  0.0955,  0.1048,
         0.0455,  0.0407, -0.0026,  0.0100,  0.0010,  0.0604,  0.0026,  0.0041,
         0.0380, -0.0031,  0.0396, -0.0463,  0.0052,  0.0067,  0.0067, -0.0070,
        -0.0707,  0.0294,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2691, -0.0773, -0.0248, -0.0272, -0.1808, -0.0817,  0.0041,  0.0187,
        -0.0140, -0.0185, -0.0525,  0.0078, -0.0455,  0.0061, -0.0258,  0.0050,
        -0.0164, -0.0132, -0.0145,  0.0443,  0.0528,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0126, -0.0835, -0.0677, -0.2699,  0.0597, -0.0098,  0.0833, -0.0188,
        -0.0018,  0.0473, -0.0413, -0.0280, -0.1255,  0.0235, -0.1273,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1004,  0.0572, -0.0726, -0.0080,  0.1820, -0.0203,  0.0230,  0.0180,
        -0.0392,  0.0107, -0.0269,  0.0662, -0.0631, -0.0454,  0.0350, -0.0126,
         0.1185,  0.1009,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0755, -0.0830, -0.0589, -0.1642, -0.0718, -0.1682,  0.0437, -0.0701,
        -0.0038, -0.0626, -0.0353,  0.0008,  0.0252, -0.0316,  0.0013, -0.0455,
         0.0390,  0.0087, -0.0109,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0450, -0.0498, -0.0280, -0.0132,  0.0299,  0.0118,  0.0205, -0.0228,
        -0.0340,  0.0062,  0.0205,  0.0161, -0.0129,  0.0089,  0.0014, -0.0177,
        -0.0630, -0.0426,  0.0458, -0.0325, -0.0685, -0.0032, -0.0079, -0.0002,
        -0.0080, -0.0235, -0.0011,  0.0021, -0.0109, -0.0114,  0.0013,  0.0250,
        -0.0423, -0.0169, -0.0259, -0.0131, -0.0756,  0.0081, -0.0046, -0.0099,
        -0.0053,  0.0044,  0.0326, -0.0755], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.0218e-01, -3.7861e-02, -3.1513e-02, -1.4435e-02,  2.3353e-03,
        -1.2875e-02, -2.7661e-02,  9.5626e-03, -3.7862e-02,  1.8551e-02,
        -1.8058e-02,  1.1296e-03,  4.5100e-03,  6.1437e-05,  1.2795e-02,
        -1.8156e-02,  2.5025e-02, -3.8784e-02,  2.0656e-03, -8.4775e-03,
        -3.1447e-02,  2.2232e-04,  2.2102e-03,  1.4511e-03,  1.1404e-02,
        -1.5650e-02,  3.1512e-03, -1.9452e-02,  7.5952e-03,  6.1500e-02,
        -4.2971e-03, -5.3721e-02, -9.3716e-03, -2.2633e-01,  2.8300e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0569, -0.1788, -0.0169, -0.0296,  0.0104,  0.0192, -0.0075, -0.0663,
        -0.1264, -0.0135, -0.0214, -0.0123, -0.1193, -0.0219, -0.0083, -0.0623,
         0.0164, -0.0303,  0.1058,  0.0071, -0.0171, -0.0076, -0.0218,  0.0227,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1198,  0.0031, -0.0073, -0.0096,  0.0077, -0.0364,  0.0048, -0.0686,
        -0.0609,  0.0172, -0.0644, -0.0217,  0.0680, -0.0358, -0.0389,  0.0192,
         0.0321,  0.0019, -0.0349, -0.0114, -0.0163, -0.0197, -0.0108, -0.0139,
         0.0083, -0.0348, -0.0140, -0.0154,  0.0136, -0.0455, -0.0214,  0.0165,
        -0.1034, -0.0027,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0066, -0.1663, -0.0663,  0.0545, -0.0671,  0.0241, -0.0404, -0.0079,
         0.0085, -0.0105, -0.0307,  0.0056,  0.0036, -0.0455, -0.0203, -0.0483,
        -0.0187, -0.0130,  0.0016,  0.0776, -0.0075,  0.0431, -0.0551, -0.0276,
         0.0050,  0.0021,  0.0070,  0.0458,  0.0732, -0.0163,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.3340, -0.1993, -0.1171, -0.0797, -0.0084, -0.0186, -0.0857,  0.0059,
         0.0104, -0.0436, -0.0121, -0.0396,  0.0457,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1001,  0.1069,  0.0303, -0.0954,  0.0885,  0.0440, -0.1228,  0.0079,
        -0.0250,  0.0351, -0.0132, -0.0730,  0.0252, -0.0405, -0.0151, -0.0607,
        -0.0636,  0.0528,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.3672, -0.1294,  0.0201, -0.0085, -0.0481,  0.1849, -0.0476, -0.1941,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.0607e-02, -1.1822e-02, -1.7531e-02, -1.5862e-02,  3.8810e-03,
        -2.1019e-02, -6.5402e-03, -1.4082e-02, -2.4571e-02, -9.1377e-03,
         3.1165e-03, -1.1669e-02, -2.4481e-02,  2.7870e-02, -1.3140e-02,
        -1.0789e-02,  4.8216e-04, -2.7310e-03,  4.0839e-04,  2.1982e-03,
         3.6526e-02, -4.2207e-02,  9.9793e-05, -1.0063e-03,  8.9442e-04,
        -2.8386e-02, -1.8566e-02, -7.0308e-02, -7.5732e-02, -3.9495e-02,
         6.0502e-03, -1.1466e-02,  1.7625e-02, -5.1525e-02, -2.6686e-02,
        -1.1568e-02, -5.8038e-02, -1.1875e-02, -1.2682e-02, -8.5281e-03,
        -1.2370e-02, -1.5430e-02, -1.7748e-02,  5.6346e-02, -2.3631e-02,
         6.7299e-02,  3.5976e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0031, -0.0434, -0.1024, -0.1897,  0.0119, -0.0887, -0.1056, -0.0864,
        -0.2702,  0.0297, -0.0690,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1791, -0.0674,  0.0024, -0.0993, -0.0442, -0.0266, -0.0357, -0.1281,
         0.0544, -0.0290, -0.0306, -0.0284, -0.0523, -0.0455,  0.0641, -0.0602,
        -0.0133,  0.0394,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1202, -0.1483, -0.0920, -0.1443,  0.0668,  0.0172,  0.0739, -0.0038,
        -0.0397, -0.0401,  0.0094,  0.0332,  0.0567,  0.0933, -0.0610,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0145, -0.1595,  0.0208, -0.0476, -0.1585, -0.0457, -0.0271, -0.0743,
        -0.0974, -0.0715, -0.1110,  0.0696,  0.0081,  0.0601, -0.0343,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1675, -0.0180, -0.0721, -0.0749,  0.0057, -0.0885, -0.0193, -0.0585,
        -0.0258,  0.0070,  0.0110, -0.0284, -0.0540, -0.0019,  0.0049, -0.0042,
        -0.0380,  0.0410,  0.0005, -0.0320, -0.0417, -0.0428, -0.0260,  0.0091,
        -0.0728,  0.0544,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0176, -0.1572, -0.1896,  0.0284, -0.0658,  0.0046,  0.0504, -0.0786,
        -0.0957,  0.0066,  0.0135,  0.0788, -0.1112,  0.0790, -0.0228,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1732e-01, -4.5899e-02,  1.2350e-02, -1.1870e-02, -1.5074e-02,
         3.4772e-02, -6.2301e-02,  5.2446e-03, -1.9005e-02, -2.0160e-01,
        -1.3170e-02,  2.2395e-02, -1.5327e-02, -2.4842e-02, -3.9839e-02,
         2.1623e-02, -1.5484e-02, -1.4811e-02,  1.7655e-02,  5.4816e-05,
        -1.7027e-02,  1.9418e-02,  3.0622e-03,  2.8275e-02,  1.9096e-01,
        -3.0630e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0408, -0.0042,  0.0262,  0.0400, -0.0096, -0.0422,  0.0785, -0.0452,
        -0.0149, -0.0579, -0.1600, -0.0151,  0.0115, -0.0114, -0.0259, -0.0241,
         0.0502, -0.0320, -0.0184, -0.0320,  0.0198, -0.0349,  0.0237,  0.0138,
         0.1525,  0.0150,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.0185, -0.0252, -0.0808, -0.0147, -0.0244,  0.0469, -0.0181, -0.0356,
        -0.0113, -0.0234, -0.0159, -0.0136, -0.0257, -0.0214, -0.0171, -0.0878,
        -0.0323,  0.0274, -0.0251, -0.0340, -0.0194,  0.0302,  0.0122, -0.0056,
        -0.0164, -0.0274, -0.0240, -0.0113, -0.0809, -0.0242, -0.0264, -0.0226,
         0.0059, -0.0054,  0.0468, -0.0422,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0617, -0.0916, -0.0902, -0.1111, -0.0383, -0.0143,  0.0169, -0.0351,
        -0.0010, -0.0300, -0.0156, -0.0016,  0.0057, -0.0227, -0.0066,  0.0034,
         0.0078, -0.0154,  0.0537, -0.0456, -0.0007,  0.0094,  0.0120,  0.0141,
         0.0117, -0.0166,  0.0011, -0.0175, -0.0172, -0.0177,  0.0342, -0.0282,
        -0.0126,  0.0280, -0.0471, -0.0638,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0245, -0.1542, -0.0830, -0.0923,  0.0380, -0.1004, -0.0327, -0.0247,
         0.0144, -0.0877, -0.0717, -0.0067,  0.0312, -0.0778, -0.0660,  0.0179,
         0.0016,  0.0123, -0.0358,  0.0272,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2803, -0.0051,  0.0053, -0.0525, -0.0339, -0.0169,  0.0099, -0.0052,
         0.0117,  0.0189,  0.0008,  0.0027,  0.0168, -0.0067,  0.0013, -0.0261,
        -0.0302, -0.0046,  0.0026,  0.0045,  0.0196, -0.0014,  0.0004, -0.0099,
        -0.0208,  0.0101, -0.0139,  0.0188,  0.0273, -0.0467, -0.0238,  0.0438,
         0.0292,  0.0105,  0.0099,  0.0585,  0.0828,  0.0366,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.4129e-02, -5.9220e-02,  5.3434e-03,  2.1229e-02, -3.2778e-03,
        -3.2805e-03, -2.1137e-02,  3.8782e-02,  6.7387e-03,  1.8156e-02,
        -7.2331e-02, -8.2958e-02,  1.8289e-03,  1.9321e-02,  1.2732e-02,
        -2.2513e-02, -6.4750e-04, -8.0819e-02,  2.1158e-02, -1.1467e-03,
         9.2938e-04,  6.0940e-03,  1.5438e-02, -5.9238e-03,  3.1837e-02,
        -2.7927e-02, -5.5023e-03,  2.3079e-03,  1.7385e-02, -2.9266e-05,
        -1.1298e-02,  1.3638e-02,  6.9075e-03,  3.6235e-03, -4.5124e-02,
         4.0592e-02,  3.5352e-02, -8.4682e-02,  9.6739e-03, -4.1079e-02,
         6.7909e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2528, -0.3304, -0.0077,  0.0147, -0.0317, -0.0148,  0.0357, -0.0021,
        -0.0069,  0.0049,  0.0022, -0.0423, -0.0027,  0.0383, -0.0381, -0.0265,
         0.0185,  0.0025, -0.0012, -0.0110, -0.0257, -0.0892,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.4131e-02,  1.0784e-02,  3.4554e-02, -1.0446e-02,  1.7359e-03,
        -1.3036e-02,  3.0655e-02, -4.7131e-02, -1.5409e-01, -1.8096e-02,
         4.2403e-02, -2.0133e-02,  4.0140e-02, -3.8967e-02,  3.3678e-02,
        -1.2463e-04, -8.0438e-03, -2.6857e-02, -2.7473e-02, -1.5198e-02,
        -1.0247e-01, -3.6238e-02,  1.2585e-02,  1.0249e-02, -6.5360e-03,
        -2.3260e-02, -2.9044e-03, -1.9473e-02, -7.5673e-02, -3.4068e-02,
        -1.3062e-02, -9.3317e-03,  1.5335e-03, -3.0277e-02, -1.4671e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1007, -0.0159, -0.0382, -0.0660, -0.0344,  0.0414,  0.0091, -0.0022,
         0.1068,  0.0197, -0.0296, -0.0223, -0.0264,  0.0137,  0.0402, -0.0584,
        -0.0285,  0.0160, -0.0335,  0.0234, -0.0647, -0.0372, -0.0231, -0.0336,
         0.0099, -0.0349,  0.0046,  0.0559,  0.0096,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0932, -0.0228, -0.0525, -0.0458, -0.0191, -0.0436, -0.0011, -0.0329,
        -0.0364, -0.0350, -0.0270, -0.0470, -0.0023, -0.0613, -0.0504, -0.0309,
        -0.0095, -0.0172, -0.0782, -0.0115, -0.0526,  0.0091,  0.0429,  0.0893,
        -0.0887,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0204, -0.0542, -0.0781, -0.2268, -0.0746, -0.0216, -0.0451, -0.0219,
        -0.0432,  0.0082,  0.0259,  0.0171,  0.0058,  0.0801, -0.0356, -0.0479,
        -0.0147, -0.0122,  0.0055, -0.0294, -0.0306, -0.0783, -0.0228,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0385,  0.0122, -0.0026, -0.0246, -0.0417, -0.0203,  0.1553, -0.0192,
        -0.0070, -0.0060, -0.0083, -0.0461, -0.0077, -0.0126, -0.0018,  0.0204,
         0.0127,  0.0202, -0.0005, -0.0043, -0.0266,  0.0867, -0.0121,  0.0471,
        -0.0131,  0.0267, -0.0225,  0.0062, -0.0148, -0.0113, -0.0203,  0.0071,
        -0.0102, -0.0162, -0.0263, -0.0164, -0.0032,  0.1425,  0.0288,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1389, -0.1133, -0.0606, -0.1243, -0.1235, -0.0467,  0.0307, -0.0561,
        -0.0999,  0.0090, -0.1729, -0.0241,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0636, -0.0594, -0.1167,  0.0055, -0.0098, -0.0572, -0.0219, -0.0343,
        -0.0019, -0.0268, -0.0037, -0.0312, -0.0020, -0.0191, -0.0063,  0.0082,
        -0.0118, -0.0319, -0.0057, -0.0189, -0.0054, -0.0191, -0.0255, -0.0153,
        -0.0113, -0.0140, -0.0071, -0.0120,  0.0424,  0.0064,  0.0418,  0.0232,
        -0.0095, -0.0138, -0.0232, -0.0412, -0.0204, -0.0066, -0.0246, -0.0154,
        -0.0161, -0.0116, -0.0153,  0.0149,  0.0279,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1042, -0.0772, -0.0122, -0.0376,  0.1847, -0.0734, -0.2870, -0.0281,
        -0.0447, -0.0103,  0.1128, -0.0278,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.4122, -0.0825, -0.1614, -0.0031,  0.1588, -0.0852, -0.0268,  0.0701,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0474, -0.0435, -0.0152,  0.0418,  0.0055,  0.0010, -0.0126, -0.0004,
        -0.0172, -0.0570, -0.0188,  0.0291, -0.0050, -0.0095, -0.0089,  0.0098,
         0.0104, -0.0173,  0.0094, -0.0054, -0.0094,  0.0117, -0.0033, -0.0024,
        -0.0074, -0.0002, -0.0041, -0.0109, -0.0010, -0.0034,  0.0056, -0.0132,
        -0.0012, -0.0124, -0.0023, -0.0088, -0.0141, -0.0089, -0.0137, -0.0163,
        -0.0064, -0.0011,  0.0177, -0.0020, -0.0446,  0.0497, -0.0536, -0.0380,
        -0.0653, -0.0527, -0.0148,  0.0129, -0.0075,  0.0421, -0.0078, -0.0136,
        -0.0075, -0.0036,  0.0275, -0.0117,  0.0045], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1337, -0.0500,  0.0372, -0.0595, -0.0355, -0.0077, -0.0475,  0.0026,
         0.0392, -0.0256, -0.0540,  0.0033, -0.0082, -0.0190, -0.1025, -0.0198,
        -0.0097, -0.0191, -0.0302, -0.0126, -0.0365, -0.0283,  0.0301, -0.0181,
        -0.0251, -0.0052, -0.0148, -0.0154, -0.0198, -0.0584, -0.0062, -0.0116,
        -0.0136,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0104, -0.0431, -0.0790, -0.0174,  0.0194,  0.0143, -0.0165,  0.0274,
         0.0038, -0.0105, -0.0435, -0.0347, -0.0396, -0.0333, -0.0322, -0.0329,
        -0.0447, -0.0902, -0.2056,  0.0085,  0.0752, -0.0842, -0.0102,  0.0235,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0769, -0.2414, -0.0251, -0.0341, -0.0064, -0.0086, -0.0270, -0.0419,
        -0.0066, -0.0450, -0.0129, -0.0535, -0.0787, -0.0391, -0.0180, -0.0248,
         0.0086,  0.0252, -0.0244, -0.0139, -0.0361, -0.0199, -0.0502, -0.0819,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2348, -0.0162, -0.0298, -0.0080,  0.0050,  0.0347, -0.0200, -0.0187,
         0.0128, -0.0135,  0.0097, -0.0253,  0.0361,  0.0005, -0.0334, -0.0387,
        -0.0247, -0.0107,  0.0188, -0.0426, -0.0222, -0.0162, -0.0319, -0.0804,
        -0.0196,  0.0175, -0.0036, -0.0216,  0.0450,  0.0146, -0.0081,  0.0070,
         0.0364,  0.0420,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0759, -0.1002, -0.0092,  0.0102, -0.0496,  0.0183, -0.0153, -0.0318,
         0.0959, -0.0548,  0.0593,  0.0434,  0.0424, -0.1089,  0.0337,  0.0349,
        -0.0354, -0.0236, -0.0020,  0.0052, -0.0010, -0.0628, -0.0048,  0.0286,
         0.0529,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0102, -0.0061, -0.0121,  0.0466, -0.0231,  0.0256,  0.0590,  0.0186,
         0.0473,  0.3527,  0.1586,  0.0245,  0.0059, -0.0315, -0.1386, -0.0396,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2557, -0.0086, -0.0409, -0.0667, -0.0204, -0.0394, -0.0578, -0.0107,
         0.0337, -0.0226, -0.0193, -0.0432, -0.0458, -0.0225, -0.0175,  0.0092,
        -0.0418, -0.0231, -0.0320, -0.0302, -0.0053, -0.1143, -0.0393,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0805, -0.2555,  0.0463,  0.0532, -0.0963, -0.0825,  0.0226,  0.0119,
        -0.0451,  0.0871, -0.0242, -0.0167,  0.0136, -0.1645,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0110, -0.0628, -0.0120, -0.0999, -0.0282,  0.0311, -0.0087,  0.0129,
         0.0199,  0.0064, -0.1256, -0.0370, -0.1199,  0.0122, -0.0229, -0.0214,
        -0.0927, -0.1533, -0.0209,  0.0139, -0.0274, -0.0347, -0.0063,  0.0187,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1063,  0.2145,  0.1520, -0.0293, -0.1116, -0.0436, -0.0884, -0.0166,
         0.0667, -0.1371, -0.0338,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0339, -0.1928,  0.0850,  0.0906, -0.0316,  0.0368, -0.1406,  0.0629,
         0.0209, -0.0052,  0.0252,  0.0113, -0.0130, -0.0431, -0.0300,  0.0036,
        -0.0152,  0.0199, -0.0281,  0.0936, -0.0167,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0283, -0.1204, -0.1536, -0.0312,  0.0631, -0.0671, -0.1166, -0.1226,
        -0.0667, -0.0402, -0.0061,  0.0101,  0.1328, -0.0413,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0093, -0.0692,  0.0545,  0.0050, -0.0039,  0.0089,  0.0172,  0.0620,
         0.0630, -0.0645, -0.0227,  0.0821,  0.1341, -0.0488,  0.0022, -0.0363,
         0.0293,  0.0672,  0.1254,  0.0141,  0.0802,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0020, -0.1196, -0.1057, -0.0403, -0.0535, -0.0817, -0.0463, -0.1810,
        -0.0793, -0.1871, -0.0022, -0.0624, -0.0390,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1162, -0.0469, -0.0054, -0.0050,  0.0067,  0.0094, -0.0175, -0.0253,
        -0.0129,  0.0178, -0.0173,  0.0165, -0.0153,  0.0053,  0.0018, -0.0018,
        -0.0184,  0.0285,  0.0012,  0.0134, -0.0458,  0.0202,  0.0065,  0.0450,
        -0.0611,  0.0017, -0.0437, -0.0220, -0.0310, -0.0063, -0.0303,  0.0897,
        -0.0087,  0.0331, -0.0184,  0.0116, -0.0197, -0.0030,  0.0002, -0.0110,
        -0.0270,  0.0098, -0.0026,  0.0039,  0.0652], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1119, -0.0230, -0.0237, -0.0253, -0.0308,  0.0088, -0.0111,  0.0049,
        -0.0139, -0.0004, -0.0165,  0.0097,  0.0134, -0.0094, -0.0135,  0.0480,
        -0.0149,  0.0059, -0.0013, -0.0197,  0.0079, -0.0557, -0.0162, -0.0396,
         0.0088, -0.0017,  0.0176, -0.0245, -0.0761, -0.0282,  0.0634,  0.0868,
        -0.0856, -0.0819,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0156, -0.0997, -0.0280, -0.0632, -0.0568, -0.0045, -0.0754, -0.0711,
         0.0881, -0.0466, -0.0192, -0.0306,  0.0232, -0.0541, -0.0436,  0.0219,
        -0.0053, -0.0550,  0.0068,  0.0064, -0.0234, -0.0394, -0.0077,  0.0791,
        -0.0101, -0.0249,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0677, -0.1313,  0.0555, -0.0423,  0.0403, -0.0237,  0.0328, -0.0173,
         0.0178, -0.1127,  0.0094, -0.0079,  0.0238,  0.0182, -0.0069, -0.0382,
        -0.1276,  0.0495, -0.0321, -0.0110,  0.0086, -0.0880,  0.0374,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0097, -0.0419, -0.0212, -0.0576, -0.0017, -0.0374,  0.0250, -0.0189,
        -0.0398, -0.0010, -0.0158,  0.0646, -0.0044, -0.0128, -0.1010,  0.0561,
         0.0130, -0.0160, -0.0185, -0.0228, -0.0093,  0.0035, -0.0022, -0.0162,
         0.0096,  0.0338, -0.0243, -0.0825,  0.0281,  0.0027, -0.0349, -0.0282,
         0.0484, -0.0060,  0.0005,  0.0158, -0.0163,  0.0019, -0.0397, -0.0170,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0421, -0.0133,  0.0016,  0.0271, -0.0276, -0.0364,  0.0035, -0.0394,
         0.0036, -0.0635, -0.0076, -0.0118, -0.0146, -0.0336, -0.0130, -0.0266,
        -0.0248, -0.0026,  0.0677,  0.0118, -0.0172, -0.0114,  0.0084,  0.0184,
        -0.0321, -0.0072, -0.0151, -0.0694, -0.0133, -0.1435, -0.1516,  0.0211,
         0.0068,  0.0124,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0789, -0.0532, -0.0284,  0.0029,  0.0486, -0.0003,  0.0208,  0.0003,
        -0.0161,  0.0210, -0.0226, -0.0048,  0.0022, -0.0230, -0.0583, -0.0606,
        -0.0448,  0.0124,  0.0074, -0.0090, -0.0372,  0.0138,  0.0197,  0.0487,
        -0.0048, -0.0357,  0.0721,  0.0227,  0.0030,  0.0279,  0.0035, -0.0196,
        -0.0263, -0.0044, -0.0124,  0.0115,  0.0128, -0.0125, -0.0041,  0.0003,
        -0.0147, -0.0072, -0.0153, -0.0097, -0.0107,  0.0027,  0.0026,  0.0047,
        -0.0043,  0.0075, -0.0034,  0.0086], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0139, -0.0314, -0.0271, -0.0282, -0.0944, -0.0862,  0.0222, -0.0195,
        -0.0274, -0.0458,  0.0076, -0.0086, -0.0997, -0.0527, -0.0559, -0.0110,
        -0.0321,  0.0434, -0.0275, -0.0203, -0.0148, -0.0088, -0.0231, -0.0771,
         0.0130, -0.0094,  0.0229,  0.0085, -0.0168, -0.0129, -0.0375,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0134, -0.0663, -0.0399, -0.0068, -0.0294, -0.0159, -0.0700, -0.0104,
        -0.0403,  0.0150, -0.0273, -0.0389, -0.0379, -0.0534, -0.0732, -0.0119,
         0.0302,  0.0416, -0.0140, -0.0592,  0.0277, -0.0085, -0.0711,  0.0866,
         0.1111,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.3932,  0.0194,  0.0010, -0.0524, -0.0702, -0.1278, -0.0727,  0.0334,
         0.0119, -0.0889, -0.0363,  0.0929,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0142, -0.1209, -0.0678, -0.0686, -0.0270, -0.0228, -0.0523,  0.0013,
        -0.0060,  0.0169, -0.0582, -0.0203, -0.0282, -0.0531, -0.0207, -0.0489,
        -0.0086, -0.0341,  0.0114, -0.0062, -0.0171, -0.0026,  0.0756, -0.0507,
         0.0644, -0.0511,  0.0167, -0.0054, -0.0288,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2048,  0.0091, -0.0435, -0.0089, -0.0149, -0.0766, -0.1356,  0.1295,
         0.2032,  0.0266, -0.0311, -0.1163,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.5477e-01, -1.8438e-01, -8.9503e-02, -5.9139e-02,  7.7757e-02,
        -1.2722e-01, -1.3600e-01, -4.3463e-02, -6.3151e-02, -7.6208e-06,
         5.0071e-02,  1.4540e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0949, -0.0066, -0.2427,  0.0027, -0.0147,  0.0472, -0.0032, -0.0144,
        -0.1644, -0.0047, -0.0216, -0.0823, -0.0204,  0.0848,  0.0009, -0.0270,
         0.0018, -0.0613, -0.0576,  0.0467,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2020,  0.0602,  0.0150,  0.0395, -0.0186, -0.0146, -0.0354, -0.0115,
        -0.0277, -0.0142,  0.0027, -0.0112, -0.0219, -0.0305,  0.0130,  0.0063,
         0.0122, -0.0047,  0.0081, -0.0139, -0.0108, -0.0059, -0.0210,  0.0172,
         0.0659, -0.0177,  0.0774, -0.0183, -0.0031, -0.0101,  0.0141,  0.0040,
        -0.0128, -0.0498, -0.0026, -0.0627,  0.0431,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2218, -0.4032,  0.0086, -0.1252, -0.0468, -0.0096,  0.0711,  0.0168,
        -0.0969,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0200, -0.0853,  0.1547, -0.0386, -0.0071, -0.0724, -0.0293, -0.0129,
        -0.0196,  0.0164, -0.0159,  0.0103, -0.0309, -0.0283,  0.0199,  0.0266,
        -0.0158, -0.0079, -0.0167,  0.0126,  0.0095, -0.0097,  0.0076, -0.0246,
         0.0219, -0.1913,  0.0938,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.0587e-02, -7.6627e-02, -3.3210e-02,  1.2361e-02, -2.5001e-02,
        -3.6537e-02, -1.9051e-02, -7.8033e-03,  5.4735e-04,  5.5072e-02,
         2.2266e-02, -1.9753e-02, -1.2640e-02, -3.7949e-02, -6.6941e-03,
        -2.4384e-02, -3.1007e-02, -8.9170e-03, -2.0830e-03,  1.2241e-03,
        -9.6756e-03, -6.5598e-03, -2.4229e-02,  2.4993e-02, -4.3815e-03,
         5.2396e-02,  2.6139e-03, -2.6667e-02, -2.8008e-01,  1.3499e-02,
        -2.5031e-04, -2.0354e-03,  4.5086e-02,  4.3816e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([-2.2250e-02,  2.2618e-02,  3.6057e-02, -1.9620e-02, -4.3052e-02,
        -6.0942e-02,  3.0703e-02, -3.9518e-03,  4.5872e-03, -1.7853e-02,
        -7.0802e-02, -3.9396e-02, -1.8650e-02,  8.9970e-03,  1.9271e-02,
         1.0085e-03, -4.7596e-02,  2.0510e-02,  4.2114e-02,  4.9727e-03,
         2.0162e-02, -1.9746e-02, -7.9975e-03,  1.4018e-02,  8.0109e-03,
        -6.1990e-03, -2.1143e-02,  2.3927e-02,  1.4274e-02,  4.2421e-02,
         1.8341e-03, -3.6751e-02, -9.1252e-03,  4.2324e-02,  1.1322e-03,
        -2.5047e-02, -2.1672e-02,  1.9299e-03, -3.2503e-05,  9.7638e-03,
        -1.5037e-02, -3.6100e-02,  3.5512e-02,  4.8724e-03, -4.6018e-02,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1057, -0.1316, -0.0735, -0.0234, -0.1107, -0.0188, -0.1499,  0.0052,
        -0.0478, -0.0502,  0.0334, -0.0026, -0.1114,  0.0380,  0.0488,  0.0489,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1580,  0.0072,  0.0970,  0.0584,  0.0444, -0.0040,  0.0280,  0.0251,
        -0.0252, -0.0378, -0.0270, -0.0276, -0.0961, -0.0071, -0.0315,  0.0107,
        -0.0052,  0.0010,  0.0365,  0.0009, -0.0429, -0.0212, -0.0028, -0.0081,
         0.0736,  0.0548,  0.0680,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0276,  0.0590, -0.1153, -0.0428, -0.0501, -0.0284, -0.0065, -0.0025,
        -0.0173, -0.0093,  0.0062, -0.0063, -0.0148,  0.0203, -0.0111, -0.0043,
        -0.0103,  0.0071,  0.0741,  0.0020,  0.0196, -0.0055, -0.0209, -0.0358,
         0.0339,  0.0008, -0.0112, -0.0253,  0.0015, -0.0177,  0.0198, -0.0104,
        -0.0332,  0.0127, -0.0178,  0.0171, -0.0089, -0.0342, -0.0190,  0.0031,
         0.0043, -0.0085, -0.0104, -0.0060, -0.0108,  0.0963], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1830, -0.0359, -0.0068,  0.0068, -0.0372, -0.0053, -0.0284,  0.0063,
        -0.0017,  0.0097, -0.0196,  0.0104, -0.0019, -0.0057, -0.0571,  0.0228,
        -0.0330, -0.0328, -0.0453, -0.1801, -0.0201, -0.0475, -0.0162, -0.0290,
         0.0188, -0.0102, -0.0261, -0.0438, -0.0232,  0.0355,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.3858, -0.2498, -0.0028,  0.1016,  0.0501, -0.0314, -0.0068,  0.0174,
        -0.0522, -0.0078,  0.0944,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1842, -0.2062,  0.0221, -0.0235, -0.0827,  0.0182, -0.0193, -0.1315,
        -0.0091, -0.0152, -0.0223, -0.0030, -0.1074,  0.0175,  0.0108, -0.0096,
         0.0407,  0.0061, -0.0568,  0.0137,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0099, -0.2166, -0.0826, -0.2087, -0.0988, -0.0674, -0.1756, -0.0279,
        -0.0616,  0.0394,  0.0115,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1417, -0.1869, -0.0150, -0.0530,  0.0116,  0.0227, -0.0307, -0.0238,
        -0.0903, -0.0096, -0.0789,  0.0047, -0.0472, -0.0137, -0.0306, -0.0611,
        -0.0513, -0.0085,  0.0195, -0.0614, -0.0376,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1019,  0.0778,  0.0670, -0.0227, -0.0147,  0.1974, -0.0257, -0.0506,
        -0.0332, -0.0281, -0.0413,  0.0154, -0.0353,  0.0064, -0.0697, -0.0603,
        -0.0352,  0.1173,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0660, -0.0349, -0.1093,  0.0186, -0.0035,  0.0291,  0.0192, -0.0053,
        -0.0016, -0.0068,  0.0196, -0.0013, -0.0042, -0.0213, -0.0513,  0.0266,
        -0.0137, -0.0646,  0.0092, -0.0304, -0.0189, -0.0285, -0.0728, -0.0081,
        -0.0043, -0.0108, -0.0065, -0.0219, -0.0130, -0.0423, -0.0009, -0.0172,
        -0.0212, -0.0202, -0.0295, -0.0173, -0.0321,  0.0359, -0.0619,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0431, -0.2554,  0.0174, -0.0255, -0.0081,  0.0168, -0.0165,  0.0365,
         0.0399,  0.0298,  0.0016, -0.0058,  0.0119, -0.0023, -0.0452, -0.0048,
        -0.0095,  0.0351,  0.0065, -0.0263, -0.0144, -0.0121, -0.0054, -0.0145,
         0.0061, -0.0078, -0.0272,  0.0148, -0.0343, -0.0872,  0.0018, -0.0124,
        -0.0744, -0.0495,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.2096, -0.1883, -0.0325,  0.0372,  0.0490, -0.0725,  0.2581, -0.0466,
        -0.1062,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0287, -0.0834, -0.0492,  0.0356, -0.0372,  0.0478, -0.0234,  0.0056,
        -0.0059, -0.0293, -0.0086, -0.0007,  0.0209, -0.0063,  0.0259,  0.0205,
        -0.0157, -0.2209,  0.0094,  0.0198, -0.0037, -0.0104,  0.0139, -0.0045,
        -0.0100, -0.0017,  0.0116, -0.0128, -0.0634, -0.0038, -0.0219,  0.0070,
         0.0037, -0.0034,  0.0041,  0.0096, -0.0178, -0.0075,  0.0305, -0.0640,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0065,  0.0249,  0.0829, -0.0352,  0.0010, -0.0353,  0.0319, -0.0357,
        -0.0097,  0.0371, -0.0038, -0.0144, -0.0009, -0.0052, -0.0184, -0.0296,
         0.0231, -0.0220,  0.0264,  0.0019,  0.0443,  0.0363, -0.0161,  0.0208,
        -0.0066,  0.0178,  0.0043,  0.0078,  0.0676, -0.0147,  0.0242,  0.0382,
         0.0103, -0.0110,  0.0039,  0.0283,  0.0241,  0.0253, -0.0183, -0.0002,
        -0.0094,  0.0222,  0.0184,  0.0108,  0.0053,  0.0088,  0.0235, -0.0006,
         0.0024, -0.0198,  0.0057, -0.0071,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0792, -0.0911, -0.0829,  0.0254,  0.0476, -0.0326, -0.0044, -0.0095,
        -0.0227,  0.0063,  0.0289,  0.0192, -0.0829, -0.0500, -0.0808, -0.0698,
        -0.0054,  0.0750, -0.0373, -0.1366,  0.0121, -0.0004,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.9879e-01, -3.4504e-01,  1.7291e-02, -1.5257e-02, -2.6140e-02,
         1.5355e-03,  2.5831e-02, -2.3834e-02, -5.2064e-02,  1.3719e-02,
        -8.3133e-02, -1.3767e-04, -7.3686e-03, -3.8915e-03, -1.0049e-02,
        -1.5140e-02,  3.4453e-02, -3.4923e-04,  1.5074e-02,  4.1828e-03,
        -1.0672e-01,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4946e-02, -1.1475e-01, -9.7289e-03, -4.4180e-02,  4.0632e-02,
        -1.5501e-03,  1.3344e-02, -1.4873e-02, -2.9876e-02, -7.6443e-03,
        -1.6770e-02, -1.2185e-03, -1.4385e-03,  3.4383e-02, -4.3709e-03,
        -2.2290e-02, -2.4915e-02,  3.6127e-03, -2.5986e-02, -1.5927e-02,
        -9.2276e-04, -4.0309e-03, -7.4087e-03, -3.0454e-03, -5.4246e-05,
        -2.0632e-04, -2.3625e-02, -6.9020e-03, -1.2137e-02,  7.6809e-03,
        -4.4825e-02, -2.3568e-02, -5.2625e-04, -1.4522e-02, -1.0078e-02,
         9.1996e-03, -7.5469e-03, -1.3996e-02, -1.3467e-02, -1.2869e-02,
        -7.6910e-02, -3.0987e-02, -6.9892e-02, -4.1147e-03, -1.3912e-02,
        -1.3460e-02, -3.9404e-02, -3.4551e-02,  1.4552e-05, -1.4537e-02,
        -1.0340e-02,  5.1358e-03,  8.5301e-03, -2.1880e-03,  1.6983e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0555, -0.0241, -0.0876, -0.0787,  0.0489, -0.0113,  0.0169, -0.0172,
         0.0083,  0.0163, -0.0034,  0.0105, -0.0012, -0.0123, -0.0048, -0.0158,
         0.0094, -0.0230,  0.0085, -0.0055,  0.0109, -0.0245, -0.0548, -0.0029,
         0.0063,  0.0090, -0.0289, -0.0403, -0.0127, -0.0080,  0.0778, -0.0049,
         0.0027, -0.0427, -0.0295, -0.0027, -0.0159, -0.0105, -0.0036, -0.0185,
        -0.0040, -0.0160, -0.0596,  0.0219,  0.0320,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0766, -0.0595, -0.1053, -0.1104, -0.0321, -0.0160, -0.0477, -0.0694,
        -0.0290,  0.0251,  0.0024, -0.0202, -0.0133, -0.0042,  0.0026, -0.0284,
         0.0548, -0.1320,  0.0503, -0.0002, -0.0055,  0.0124,  0.0334, -0.0691,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0536,  0.0179,  0.0127, -0.0259, -0.0763, -0.0022, -0.0607, -0.0523,
        -0.0458,  0.0029,  0.0478, -0.0169, -0.0408, -0.0262, -0.0337,  0.0120,
        -0.0058, -0.0144,  0.0197, -0.0278, -0.0287,  0.0326, -0.0035, -0.0466,
        -0.0501,  0.0150, -0.0427,  0.0023,  0.0022, -0.0530,  0.0504, -0.0356,
         0.0418,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2829, -0.0992, -0.0505, -0.0840, -0.0470,  0.0241,  0.0057, -0.0071,
        -0.0755,  0.0166,  0.0015, -0.0221, -0.0338,  0.0505,  0.0706, -0.0570,
        -0.0518, -0.0026, -0.0175,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0488, -0.0678,  0.0690,  0.0376, -0.0447,  0.0173, -0.0545,  0.0271,
        -0.0736, -0.0311,  0.0121, -0.0293,  0.0078, -0.0200,  0.0279, -0.0091,
        -0.0179,  0.0448, -0.0121,  0.0245, -0.0210, -0.1096, -0.1193, -0.0251,
        -0.0481,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2092, -0.0824,  0.0037,  0.0416, -0.0747, -0.0928, -0.2045, -0.0268,
        -0.0349, -0.0519, -0.0126, -0.0345, -0.0182, -0.0308, -0.0278, -0.0141,
        -0.0395,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 8.2482e-02, -9.6865e-03,  6.2208e-02,  2.1191e-02,  1.4213e-02,
         3.4131e-02, -3.9268e-05, -5.6123e-02,  3.9874e-04,  5.4837e-02,
        -2.6428e-02, -2.5578e-02, -2.6673e-02,  7.4866e-02, -6.9752e-03,
         7.2932e-03,  2.0981e-02,  3.8325e-02, -9.5653e-02,  2.4075e-03,
         1.0548e-02,  3.0194e-02,  5.3973e-02, -1.1206e-02,  4.3694e-02,
         3.1714e-03,  4.8081e-02,  4.0163e-02,  3.5634e-03, -5.5629e-04,
        -2.9490e-02,  6.4869e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2121, -0.0634, -0.0328, -0.0127,  0.0203,  0.0034, -0.0171, -0.0505,
        -0.0018, -0.0034,  0.0113,  0.0092, -0.0920,  0.0145,  0.0633, -0.0093,
        -0.0520, -0.0525, -0.0544,  0.0094, -0.0017,  0.0104, -0.0568, -0.0299,
         0.0057, -0.0328, -0.0519,  0.0254,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0877, -0.1247,  0.0808,  0.0267, -0.0003, -0.0172, -0.0456, -0.0828,
        -0.0107, -0.0084, -0.0114,  0.0009, -0.0007,  0.0130,  0.0082, -0.0311,
        -0.0056, -0.0399,  0.0231, -0.0297, -0.0179,  0.0003,  0.0483, -0.0179,
        -0.0131,  0.0690,  0.0803, -0.0683, -0.0363,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0748, -0.0245, -0.0166, -0.0109, -0.0734, -0.0522,  0.0104, -0.0806,
        -0.0061, -0.0453, -0.0212, -0.0092,  0.0051, -0.0238,  0.0391, -0.0018,
        -0.0254, -0.0421, -0.0206,  0.0361,  0.0913, -0.0390, -0.0187, -0.0242,
        -0.0134, -0.0063, -0.0215, -0.0119,  0.0616, -0.0040,  0.0017, -0.0543,
        -0.0332,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0845,  0.0221, -0.0421, -0.0486,  0.0496, -0.0462,  0.0941, -0.0274,
         0.0141, -0.0346,  0.0303, -0.0472,  0.0426, -0.0388, -0.0162, -0.0057,
        -0.0262, -0.0379, -0.0915, -0.0202, -0.0448, -0.0346, -0.0330, -0.0678,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7970e-02, -9.5467e-02, -8.7481e-02, -2.5009e-02, -4.5831e-03,
        -8.6931e-02, -5.6153e-02, -3.8047e-02, -1.3677e-02, -4.4614e-02,
         1.4278e-02,  7.7175e-05, -3.0093e-02, -1.1139e-02, -1.4224e-02,
        -1.1137e-02, -1.7140e-03, -2.4191e-02, -9.0564e-03, -1.9230e-01,
        -7.3961e-02, -6.6263e-03, -1.4127e-01,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0128,  0.1034,  0.0017,  0.0238,  0.0393,  0.0565, -0.0120,  0.0097,
        -0.0139,  0.0130,  0.0348,  0.0161,  0.0472,  0.0328,  0.0418, -0.0257,
         0.0080,  0.0278,  0.0222,  0.0587,  0.0123, -0.0207, -0.0090, -0.0153,
         0.0203,  0.0089, -0.0186,  0.0200,  0.0202,  0.0369,  0.0535,  0.0281,
         0.0133,  0.0006,  0.0288,  0.0387,  0.0073, -0.0272,  0.0190,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0198, -0.2240,  0.0104,  0.0151, -0.0285,  0.0054,  0.0031,  0.0139,
        -0.0332,  0.0255, -0.0395, -0.0105, -0.0043, -0.0221, -0.0076, -0.0310,
         0.0511, -0.0080,  0.0051, -0.0058, -0.0033, -0.0132, -0.0009, -0.0218,
        -0.0017, -0.0171, -0.0075,  0.0052, -0.0169, -0.0054, -0.0221,  0.0062,
         0.0018, -0.0088, -0.0061,  0.0014, -0.0093, -0.0251, -0.0264, -0.0034,
         0.0136,  0.0246, -0.1942,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0418, -0.0932, -0.0041,  0.0117, -0.0417, -0.0033, -0.0163,  0.0017,
        -0.0537,  0.0023, -0.0168,  0.0004, -0.0160,  0.0023,  0.0011, -0.0252,
        -0.0186, -0.0469, -0.0023, -0.0020,  0.0007, -0.0157, -0.0077, -0.0088,
        -0.0089,  0.0001,  0.0307,  0.0113,  0.0218, -0.0045,  0.0012, -0.0299,
        -0.0451, -0.0187, -0.0106, -0.0276, -0.0851,  0.0286, -0.0203,  0.0048,
        -0.0163, -0.0367, -0.0059, -0.0017,  0.0032, -0.0062,  0.0004, -0.0004,
         0.0054, -0.0359, -0.0087, -0.0034,  0.0022, -0.0377, -0.0240,  0.0263,
        -0.0020], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0989, -0.1385, -0.1317, -0.0584, -0.0149, -0.0681, -0.1212, -0.0831,
        -0.0039,  0.0772, -0.0198, -0.0516, -0.0835, -0.0491,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1096,  0.0962,  0.0151, -0.0616,  0.0062, -0.0089,  0.0279,  0.0152,
         0.0060,  0.0104,  0.0171,  0.1113,  0.0680, -0.0078,  0.0135, -0.0492,
         0.0012,  0.0554, -0.0711,  0.0131,  0.1056,  0.0282,  0.0179, -0.0140,
         0.0180, -0.0024, -0.0070, -0.0012, -0.0175,  0.0232,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0575, -0.1678, -0.2065,  0.0083, -0.1892, -0.2237, -0.1080, -0.0390,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0907, -0.0258,  0.0188, -0.0106, -0.0082, -0.1361,  0.0331,  0.0348,
         0.0032, -0.0712,  0.0020, -0.0340, -0.0533, -0.0399,  0.0028, -0.0173,
        -0.0019,  0.0030, -0.0037, -0.0159,  0.0015, -0.0443, -0.0187, -0.0230,
        -0.0431, -0.0498, -0.0332,  0.0061, -0.0228, -0.0931,  0.0232, -0.0268,
        -0.0080,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1854e-01, -5.1679e-03, -4.0493e-02, -1.8058e-02, -1.5708e-03,
        -6.3365e-03, -5.4778e-04, -6.9303e-03,  1.5584e-02, -5.3880e-03,
        -2.6732e-03,  1.0137e-03, -1.1006e-03, -4.8934e-03,  6.5366e-03,
         6.0018e-03, -2.1324e-03, -1.4905e-02, -3.0460e-03,  1.6298e-03,
        -1.0101e-02,  4.0676e-05,  5.5586e-03,  3.1129e-03,  1.0916e-02,
         7.4548e-04, -6.7426e-03, -3.2278e-03, -3.6262e-03, -2.4378e-03,
        -4.3876e-03, -1.1506e-03, -4.5909e-03, -6.0363e-02, -3.0045e-03,
         3.5414e-03, -1.0871e-02,  1.2266e-02,  6.4668e-03, -7.9109e-03,
        -1.1651e-03, -6.4156e-03, -1.6109e-02, -2.4222e-02,  4.8118e-02,
        -1.4975e-02, -2.9603e-02, -1.1423e-02, -4.5313e-02, -5.8650e-02,
        -5.1788e-02, -1.8811e-02, -1.2565e-02, -9.5413e-03, -2.9086e-02,
         3.9894e-04, -8.5030e-03, -7.3302e-04, -2.3294e-02, -1.5621e-02,
        -2.3977e-02,  1.2262e-02, -5.0317e-02, -1.6598e-02,  3.3344e-03,
         3.5344e-03, -6.9643e-03,  8.4915e-04,  9.0465e-03,  1.1175e-02,
         1.2001e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2493, -0.0116,  0.0044, -0.0147, -0.0191, -0.1036, -0.1048,  0.0496,
         0.0081,  0.0022, -0.0049,  0.0043, -0.0111, -0.0008,  0.0082, -0.0173,
         0.0005,  0.0101,  0.0075, -0.0265, -0.0213, -0.0019,  0.0331, -0.0328,
        -0.0121, -0.0160,  0.0156, -0.0073,  0.0040, -0.0096,  0.0107,  0.0116,
        -0.0175, -0.0037,  0.0162, -0.0334,  0.0080,  0.0139, -0.0040,  0.0222,
         0.0455, -0.0007,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0700, -0.1166, -0.0360,  0.0035, -0.0073, -0.0138, -0.0002, -0.0091,
        -0.0402, -0.0283, -0.0180, -0.0250, -0.0412, -0.0163,  0.0058,  0.0160,
        -0.0123, -0.0139, -0.1022, -0.0246, -0.0206, -0.0069, -0.0174, -0.0592,
        -0.0096, -0.0254,  0.0062, -0.0188, -0.0570, -0.0216, -0.0212, -0.0146,
        -0.0195, -0.0204,  0.0333,  0.0011,  0.0470,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1355,  0.1055, -0.0797,  0.0461, -0.0002,  0.0376,  0.0586,  0.0114,
         0.0678,  0.0475,  0.0417,  0.0903,  0.0602,  0.1305,  0.0875,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0581, -0.0281,  0.0714,  0.0752,  0.0333,  0.0490, -0.0046,  0.0160,
         0.0487,  0.0551, -0.0149, -0.0075,  0.0795,  0.0574,  0.0504, -0.0543,
         0.0103, -0.0928,  0.1094, -0.0839,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1802, -0.1130, -0.0383,  0.0754, -0.0448, -0.0182, -0.0120,  0.0258,
         0.0019, -0.0099,  0.0080, -0.0114, -0.0234, -0.0032, -0.0213, -0.0041,
         0.0140, -0.0115,  0.0175, -0.0101, -0.0084,  0.0156, -0.0223,  0.0177,
        -0.0425,  0.0126, -0.0380, -0.0065, -0.0081,  0.0092, -0.0133, -0.0220,
        -0.0010, -0.0149,  0.0171, -0.0139, -0.0185, -0.0027, -0.0105, -0.0125,
        -0.0082,  0.0364, -0.0040,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0689,  0.0058, -0.1304, -0.0465, -0.0075, -0.0109, -0.0088, -0.0966,
        -0.0114,  0.0015, -0.0091, -0.0144, -0.0415, -0.0077, -0.0267, -0.0087,
         0.0034, -0.0238,  0.0138,  0.0531, -0.0091, -0.0277, -0.0222, -0.0193,
         0.0316, -0.0321, -0.0884, -0.0069, -0.0016, -0.0012,  0.0008, -0.0163,
         0.0074, -0.0046, -0.0095,  0.0094, -0.0264, -0.0950,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1304, -0.1239, -0.0334, -0.1835,  0.0060, -0.0816,  0.0131,  0.0572,
        -0.0301, -0.1449, -0.0064, -0.0134, -0.0162, -0.0081,  0.0151, -0.0334,
        -0.0217, -0.0246, -0.0290,  0.0029, -0.0163, -0.0089,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7143e-01, -6.4752e-02,  4.9728e-03, -2.3791e-02, -1.7897e-02,
        -2.6679e-02,  8.3484e-04,  2.6396e-02, -6.8476e-02, -1.8134e-02,
        -1.3839e-02, -2.0241e-02, -2.1888e-02, -1.8476e-02, -9.2771e-03,
        -2.2990e-02, -5.5301e-02, -3.9733e-02, -2.2821e-02, -1.1160e-02,
         2.0199e-02,  1.2889e-04,  9.2397e-03,  8.2788e-03, -7.3122e-03,
         9.2225e-03,  6.2903e-04, -2.6188e-02,  1.4141e-02,  1.3306e-02,
         2.4396e-02, -1.9993e-02,  4.4564e-04, -1.3987e-02, -8.8740e-03,
        -8.5884e-02, -1.8646e-02, -2.5840e-02,  2.2255e-02,  1.1943e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0485, -0.0822,  0.0401, -0.0233, -0.0015, -0.1282, -0.0509, -0.0017,
        -0.0302, -0.0373,  0.0009, -0.0137, -0.0025, -0.0318, -0.0303, -0.0214,
        -0.0196,  0.0290, -0.0845, -0.0184, -0.0126, -0.0172,  0.0196, -0.0037,
        -0.0352, -0.0466, -0.0001, -0.0286, -0.0510, -0.0391,  0.0218, -0.0283,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.5111e-02, -2.1153e-01,  4.0535e-03, -5.3548e-03, -1.6065e-02,
         1.6954e-02, -2.4290e-02, -3.9932e-02, -4.1706e-02, -5.6619e-03,
        -2.4742e-02, -1.5839e-02, -3.2768e-02, -1.0541e-02, -5.4678e-03,
        -3.8321e-02, -8.8033e-03, -1.7596e-02,  1.9096e-03, -2.6907e-02,
        -4.6416e-03,  1.0022e-04, -9.8700e-03, -1.9472e-02,  8.8860e-03,
         3.2604e-02, -2.0873e-02, -9.0273e-03,  1.7640e-02,  1.1016e-02,
         9.6772e-03, -1.0033e-02,  1.1136e-02, -5.0522e-02, -1.5095e-01,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0194,  0.0050, -0.0582,  0.0741, -0.0438,  0.0052, -0.0350,  0.0736,
         0.0050,  0.2027, -0.0194, -0.0045, -0.0292,  0.0511,  0.0571, -0.0168,
         0.0152,  0.0186,  0.0216, -0.1351,  0.1095,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0799, -0.2239, -0.0704, -0.0931, -0.0862, -0.0487,  0.0176, -0.0274,
        -0.0841, -0.0205, -0.0050, -0.0063,  0.0399, -0.0019, -0.0117, -0.0322,
         0.0016, -0.0689,  0.0809,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2437, -0.1192,  0.0112,  0.1024, -0.0102,  0.0170, -0.0612, -0.0018,
         0.0636, -0.1139, -0.0532,  0.0049, -0.0255, -0.0503,  0.0045, -0.0127,
         0.0303,  0.0040, -0.0676, -0.0027,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1654, -0.1699,  0.0006, -0.0275, -0.0219, -0.0231,  0.1977, -0.0322,
        -0.0195,  0.0043, -0.0120,  0.0095,  0.0021, -0.0149, -0.0400,  0.0020,
        -0.0790, -0.0252, -0.0139, -0.0416, -0.0231, -0.0044, -0.0702,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0876, -0.0689, -0.0049, -0.1310,  0.0016, -0.0110,  0.0174,  0.0604,
        -0.0134, -0.0359, -0.0127, -0.0323, -0.0017, -0.0783,  0.1941,  0.0182,
        -0.0411,  0.0014, -0.0328, -0.0016, -0.0338,  0.0217, -0.0864,  0.0092,
        -0.0026,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0430,  0.2222, -0.0605,  0.0354, -0.0035,  0.0318,  0.0065, -0.0081,
        -0.0019,  0.0121,  0.0473, -0.0214,  0.0135,  0.0471, -0.0169,  0.0060,
         0.0182,  0.0156,  0.0075, -0.0323,  0.0162,  0.0155,  0.0240,  0.0057,
         0.0150,  0.0067,  0.0142,  0.0307,  0.0436, -0.0229,  0.0265, -0.0037,
         0.0183, -0.0234,  0.0828,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0168,  0.0947,  0.0005, -0.1029, -0.0151,  0.0051, -0.0104,  0.0309,
        -0.0632, -0.0543,  0.0063, -0.1150, -0.0391, -0.0148, -0.1124, -0.0730,
        -0.0810, -0.0742, -0.0344, -0.0003, -0.0153,  0.0403,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1044,  0.0493, -0.0024,  0.0048, -0.0162, -0.0383, -0.0385,  0.0042,
         0.0157, -0.1138, -0.0935, -0.0794, -0.0782, -0.0263, -0.0253, -0.0538,
        -0.0258, -0.0370, -0.0287,  0.0292, -0.1354,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2400, -0.1153,  0.0028, -0.0551, -0.0335,  0.0020, -0.0557, -0.0962,
         0.0141,  0.0230,  0.0286,  0.0538, -0.0973, -0.0483, -0.0080, -0.0379,
        -0.0293, -0.0308, -0.0282,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.1322e-02,  1.1928e-01,  3.5189e-03, -3.4667e-02, -2.8417e-02,
        -4.3350e-02,  6.0264e-02,  1.4085e-02,  1.5026e-03, -6.6797e-03,
        -2.7264e-02,  2.6305e-03, -1.3917e-05, -2.3108e-02, -5.6681e-03,
        -7.7173e-03,  9.2307e-04, -5.5930e-02, -1.1449e-02,  4.0334e-03,
        -1.4022e-02, -1.2909e-02,  2.2774e-03, -1.6004e-02,  1.1428e-02,
        -1.3531e-02,  1.1439e-02, -4.1046e-02,  7.2656e-03, -1.0485e-01,
         4.3534e-03, -1.7915e-02, -1.9438e-02,  4.3430e-03, -1.3860e-02,
         1.3537e-02, -2.3983e-03,  1.7444e-03, -2.0552e-02, -4.2054e-03,
         1.2427e-02,  1.4926e-02, -1.6957e-03,  2.7793e-04, -2.7592e-02,
         5.0742e-02,  7.3394e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0420, -0.1041, -0.0281, -0.0018,  0.0235, -0.0180, -0.0027, -0.0104,
        -0.0236, -0.0062,  0.0087,  0.0024, -0.0034,  0.0152,  0.0582,  0.0042,
        -0.0807, -0.0082, -0.0200,  0.0230, -0.0070, -0.0139, -0.0260, -0.0971,
        -0.0197,  0.0491,  0.0551, -0.0021, -0.0061, -0.0059,  0.0359,  0.0319,
         0.0764, -0.0895,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0503, -0.1842,  0.0029, -0.0159, -0.0442, -0.0598, -0.0227,  0.0243,
        -0.0249, -0.0275,  0.0175, -0.0015, -0.0326, -0.0627, -0.0080,  0.0141,
        -0.0017, -0.0234, -0.0227, -0.0234, -0.0220, -0.0343, -0.0242, -0.0274,
        -0.0187, -0.0243, -0.0189,  0.0076,  0.0326, -0.0460,  0.0241, -0.0042,
        -0.0209,  0.0306,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.1038, -0.0663,  0.0008, -0.1059, -0.0179, -0.0425,  0.0061,  0.0383,
        -0.1768, -0.0391, -0.0431, -0.0113, -0.0194, -0.0751, -0.0379,  0.0032,
         0.1020,  0.1106,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0182,  0.0282,  0.0021, -0.0140, -0.0201, -0.0136, -0.0392, -0.0219,
        -0.0519, -0.0782,  0.0268, -0.0279,  0.0036, -0.1039,  0.0261,  0.0645,
         0.0186,  0.0323,  0.0334,  0.0332,  0.0813, -0.0351, -0.0654,  0.0102,
        -0.0086, -0.0681,  0.0734,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0321, -0.1788,  0.0335, -0.0388, -0.0149, -0.0279, -0.0557, -0.0063,
        -0.0752, -0.0529, -0.0258, -0.0314, -0.0421, -0.0543, -0.0108,  0.0185,
        -0.0453, -0.0273, -0.0188, -0.0907, -0.0179,  0.0024, -0.0020, -0.0070,
         0.0133, -0.0765,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1626, -0.0622, -0.0259, -0.0327, -0.1248, -0.0009, -0.0733,  0.0009,
        -0.0146, -0.0387, -0.0696,  0.0083, -0.0233, -0.0288, -0.0203, -0.0377,
        -0.0243, -0.0358, -0.0382,  0.0683,  0.1089,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0654, -0.1840, -0.0417, -0.2054, -0.0273, -0.0268,  0.0989, -0.0109,
         0.0076, -0.0660, -0.0256, -0.0515,  0.0400, -0.0988, -0.0499,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1952, -0.0113, -0.0151, -0.0986,  0.0368, -0.0530,  0.0182, -0.0122,
        -0.0425, -0.0163, -0.0633, -0.0887, -0.0637, -0.0409, -0.0831, -0.0430,
         0.0239, -0.0944,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1274,  0.0891,  0.0768,  0.2016,  0.0755,  0.0670, -0.0206,  0.0406,
         0.0505,  0.0438,  0.0234,  0.0223, -0.0576,  0.0112,  0.0177,  0.0311,
         0.0232,  0.0092,  0.0115,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 6.6053e-02, -7.0346e-02, -4.5843e-02, -1.1978e-02, -5.1476e-02,
        -2.9501e-02,  1.7723e-02, -9.5457e-03, -1.5447e-02,  1.7115e-03,
        -7.9252e-05,  2.5467e-02, -6.5541e-03,  4.5651e-03,  2.0589e-02,
        -3.3759e-02, -3.0943e-02, -4.2044e-02,  7.4176e-03,  1.3752e-02,
        -4.8095e-02,  4.3697e-02, -2.2380e-02, -4.8413e-03,  1.4034e-02,
        -3.4425e-02, -2.2745e-02, -9.8555e-03, -1.2574e-02, -8.3532e-03,
        -1.1536e-02, -1.8295e-03, -2.0132e-02, -2.2815e-02, -2.2867e-02,
        -2.7786e-04, -4.1830e-02,  6.3165e-03, -1.8178e-02,  1.3436e-02,
        -4.7518e-03,  6.0464e-03,  5.6560e-02,  4.7629e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0293, -0.0390, -0.0193, -0.0632, -0.0681, -0.0242, -0.0145,  0.0023,
        -0.0483,  0.0348, -0.0317, -0.0028, -0.0315, -0.0221,  0.0478, -0.0196,
         0.1089, -0.0892, -0.0279,  0.0115, -0.0286, -0.0173,  0.0037, -0.0028,
        -0.0137, -0.0076, -0.0175, -0.0052,  0.0047, -0.0361, -0.0263,  0.0009,
        -0.0030, -0.0624, -0.0340,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1844, -0.2300,  0.0030, -0.0285, -0.0087,  0.0052, -0.0593, -0.0718,
        -0.0329, -0.0170,  0.0138, -0.0326, -0.0065, -0.0141,  0.0038, -0.0728,
        -0.0149,  0.0046,  0.0626, -0.0152, -0.0382, -0.0061, -0.0728, -0.0012,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0121, -0.0679, -0.0364, -0.0370, -0.0422, -0.0385,  0.0794, -0.0228,
        -0.0438,  0.0324, -0.0145, -0.0034, -0.0283, -0.0445, -0.0471,  0.0094,
        -0.0186, -0.0382, -0.0528,  0.0199, -0.0622, -0.0051, -0.0016, -0.0056,
         0.0081, -0.0361, -0.0277, -0.0035, -0.0256, -0.0011, -0.0248, -0.0068,
         0.0397, -0.0626,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0565, -0.0776, -0.0901, -0.0060,  0.0253, -0.0477, -0.0644, -0.0229,
         0.0044,  0.0130, -0.0062,  0.0399, -0.0398, -0.0407, -0.0227, -0.1004,
         0.0033,  0.0358, -0.0090, -0.0615, -0.0205,  0.0109, -0.0039, -0.0360,
        -0.0282, -0.0144, -0.0110,  0.0184,  0.0377,  0.0521,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0276, -0.2195, -0.1448, -0.1164,  0.0264, -0.0732, -0.0699, -0.0611,
        -0.0076, -0.1522, -0.0330, -0.0214, -0.0471,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1524, -0.0425,  0.0096,  0.1286, -0.0698, -0.0727,  0.0058, -0.0182,
        -0.0239, -0.0520, -0.0230, -0.0590, -0.0421, -0.0059, -0.0109, -0.1746,
        -0.0723,  0.0367,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.3706, -0.0392, -0.0907, -0.0864, -0.0980, -0.0804, -0.0754, -0.1593,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0594,  0.0131, -0.0148, -0.0446, -0.0134,  0.0364,  0.0070,  0.0196,
        -0.0057, -0.0118, -0.0218, -0.0176, -0.0135, -0.0241,  0.0051, -0.0195,
         0.0040,  0.0158,  0.0067, -0.0084,  0.0051, -0.0016, -0.0479,  0.0070,
        -0.0120, -0.0096, -0.0275, -0.0044, -0.1401, -0.0089, -0.0046, -0.0197,
         0.0144, -0.0607,  0.0272, -0.0208, -0.0093, -0.0121, -0.0157, -0.0045,
        -0.0038, -0.0145, -0.0078,  0.0302,  0.0695,  0.0164,  0.0424],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1515, -0.1041, -0.1232, -0.0973,  0.0883, -0.0210, -0.0316, -0.0305,
        -0.2102, -0.0067, -0.1357,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0923, -0.0672, -0.1158, -0.0952, -0.0482, -0.0241, -0.0121, -0.0463,
        -0.0444, -0.0395, -0.0125, -0.0915, -0.0508, -0.0478,  0.0535, -0.0689,
        -0.0404, -0.0496,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1260, -0.1225, -0.1775,  0.0215, -0.0068,  0.0763, -0.0649,  0.0161,
        -0.0871, -0.0865, -0.0294,  0.0439,  0.0596,  0.0388,  0.0430,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1091, -0.1403,  0.0278, -0.0533, -0.0675, -0.0346, -0.1145, -0.1369,
        -0.0598, -0.0859, -0.1418, -0.0077, -0.0060, -0.0041,  0.0106,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0223, -0.0828, -0.0429, -0.0572, -0.0367,  0.0019,  0.0315, -0.0338,
        -0.0537, -0.0269, -0.0242, -0.0623, -0.1323, -0.0226,  0.0086, -0.0563,
        -0.0853, -0.0573, -0.0305, -0.0138, -0.0255,  0.0023, -0.0423,  0.0197,
         0.0044,  0.0230,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1592, -0.3448, -0.0473, -0.0179, -0.0539, -0.0132, -0.0007, -0.0546,
        -0.1259, -0.0154,  0.0030,  0.0246, -0.0088, -0.0863, -0.0442,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0248, -0.1292, -0.0367, -0.0451,  0.0036, -0.0543, -0.0524,  0.0249,
        -0.0420, -0.0802,  0.0239, -0.0111, -0.0182, -0.0079, -0.0202,  0.0010,
        -0.0096,  0.0148, -0.0232,  0.0488, -0.0471,  0.0743,  0.0786,  0.0475,
         0.0595, -0.0211,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0275, -0.0781, -0.0109, -0.0372, -0.0074, -0.0101,  0.0107, -0.0478,
        -0.0007, -0.0223, -0.0861, -0.0389, -0.0347,  0.1249, -0.0326, -0.0209,
         0.2004, -0.0294, -0.0224, -0.0299, -0.0299, -0.0154, -0.0018, -0.0284,
         0.0003, -0.0513,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.0784, -0.0861, -0.0795, -0.0381, -0.0725, -0.0073, -0.0081, -0.0498,
         0.0126, -0.0179,  0.0141, -0.0140, -0.1186, -0.0012, -0.0185, -0.0160,
         0.0238,  0.0315, -0.0403, -0.0083, -0.0058, -0.0180, -0.0067,  0.0015,
        -0.0459, -0.0091, -0.0255,  0.0125, -0.0148, -0.0133, -0.0402, -0.0101,
        -0.0069,  0.0165, -0.0303,  0.0063,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0072, -0.0752, -0.0750, -0.0019, -0.0090, -0.0063, -0.0278, -0.0670,
         0.0213,  0.0132, -0.0008,  0.0236, -0.0522,  0.0398, -0.0155, -0.0217,
        -0.0116,  0.0017, -0.0025, -0.0202,  0.0228,  0.0061,  0.0194, -0.0520,
        -0.0043, -0.0532,  0.0042, -0.0127, -0.0506,  0.0332,  0.0169, -0.0287,
        -0.0274,  0.0399,  0.0298, -0.1052,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.3202,  0.1209,  0.0073, -0.0627,  0.0529,  0.0312,  0.0158,  0.0761,
        -0.0222,  0.0011,  0.0085,  0.0084, -0.0227,  0.0551,  0.0337, -0.0054,
        -0.0286, -0.0039,  0.0697, -0.0537,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.9581e-01, -4.6660e-02, -4.8247e-03,  1.6019e-02,  1.7291e-02,
        -3.5762e-02, -5.1798e-03, -1.6518e-04, -5.0391e-03,  4.9808e-03,
        -4.7180e-03, -1.9839e-02, -1.8425e-02,  3.4631e-02, -1.7608e-02,
        -9.3149e-03,  6.3396e-03, -1.1419e-02, -1.8307e-02, -1.0857e-02,
        -6.8577e-02,  1.0105e-02, -8.3205e-03,  1.8108e-02,  2.1337e-02,
        -1.3003e-02,  5.7522e-03,  7.7826e-02, -2.1560e-02, -6.0473e-02,
         2.4236e-02, -3.1150e-02,  2.0307e-02, -2.3545e-02,  7.6159e-03,
         1.1343e-02, -2.0972e-02, -7.2575e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0813, -0.0464, -0.0285, -0.0754, -0.0169, -0.0021, -0.0065,  0.0122,
        -0.0085, -0.0499, -0.0667, -0.0762,  0.0020,  0.0052,  0.0386, -0.0168,
         0.0464,  0.0214, -0.0108,  0.0030, -0.0094,  0.0106,  0.0024, -0.0046,
        -0.0119,  0.0013,  0.0013, -0.0075, -0.0527, -0.0529, -0.0351,  0.0002,
         0.0345,  0.0054, -0.0090,  0.0198,  0.0175,  0.0486, -0.0191, -0.0117,
         0.0297], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0653,  0.2914, -0.0323,  0.0927, -0.0348,  0.0306,  0.0654,  0.0394,
         0.0139,  0.0101,  0.0204,  0.0139,  0.0027,  0.0182, -0.0386,  0.0436,
         0.0097,  0.0417,  0.0018, -0.0089,  0.1180, -0.0067,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0334, -0.0008, -0.0091,  0.0107, -0.0172, -0.0357,  0.0102,  0.0075,
        -0.0754, -0.0696, -0.0187, -0.0555,  0.0081, -0.0151,  0.0177, -0.0150,
        -0.0235, -0.0299,  0.0036, -0.0405, -0.1474, -0.0377, -0.0473,  0.0032,
        -0.0173, -0.0294,  0.0039, -0.0317,  0.0185, -0.0152, -0.0205, -0.0209,
        -0.0238,  0.0432, -0.0427,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1092, -0.0575, -0.0555, -0.0152, -0.1139,  0.0186, -0.0170,  0.0281,
         0.0212, -0.0133, -0.0640, -0.0556, -0.0526,  0.0043, -0.0074, -0.0325,
         0.0192, -0.0327, -0.0464, -0.0082, -0.0369, -0.0051,  0.0241, -0.0136,
        -0.0237,  0.0104, -0.0041,  0.0829,  0.0269,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0200, -0.0371,  0.0189, -0.0529, -0.0249, -0.0048, -0.0205, -0.0342,
        -0.0363,  0.0067,  0.0285, -0.0303,  0.0084,  0.0612, -0.0086,  0.0329,
         0.0900,  0.0373, -0.0662, -0.0072, -0.0791, -0.1340, -0.0192,  0.0440,
        -0.0969,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0057, -0.1221, -0.0542, -0.0648, -0.1010, -0.0299, -0.0393, -0.0113,
        -0.0483, -0.0945, -0.0313,  0.0144, -0.0750, -0.0211, -0.0361, -0.0547,
         0.0090, -0.0367, -0.0114,  0.0160, -0.0046,  0.0144, -0.1042,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1580,  0.0460, -0.0181, -0.0608,  0.0324,  0.0308, -0.0142, -0.0107,
         0.0023, -0.0272, -0.0005,  0.0090, -0.0127, -0.0689, -0.0024, -0.0216,
         0.0005, -0.0272,  0.0211, -0.0041, -0.0040,  0.0288, -0.0347,  0.0042,
        -0.0232,  0.0047, -0.0521, -0.0179, -0.0025, -0.0070, -0.0494, -0.0311,
        -0.0287, -0.0400, -0.0273, -0.0047, -0.0373,  0.0305, -0.0037,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0342, -0.0282, -0.2492, -0.1113, -0.1655, -0.0721, -0.0403, -0.0777,
        -0.0913, -0.0688, -0.0546,  0.0068,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0656, -0.0599, -0.2195, -0.0251,  0.0010, -0.0158, -0.0063, -0.0117,
        -0.0075, -0.0054,  0.0011, -0.0022,  0.0149, -0.0212, -0.0318,  0.0057,
         0.0143,  0.0015,  0.0128,  0.0098, -0.0120, -0.0268, -0.0011,  0.0148,
        -0.0006,  0.0030, -0.0133,  0.0214,  0.0213,  0.0087, -0.0425, -0.0398,
        -0.0240, -0.0320, -0.0151, -0.0295,  0.0029,  0.0027, -0.0054, -0.0114,
        -0.0026,  0.0047, -0.0015,  0.0421,  0.0882,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1925, -0.2675, -0.0090, -0.0840, -0.0269, -0.0030, -0.0540, -0.0433,
        -0.0979, -0.1815,  0.0162,  0.0242,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0335, -0.3390, -0.2702, -0.0115,  0.1379, -0.0286,  0.1338,  0.0455,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0285, -0.0991,  0.0159,  0.0582, -0.0053,  0.0063,  0.0061,  0.0057,
         0.0031, -0.0609, -0.0070,  0.0347, -0.0023, -0.0039, -0.0035,  0.0178,
        -0.0136, -0.0239,  0.0596,  0.0072,  0.0016,  0.0004, -0.0007,  0.0205,
        -0.0010,  0.0195,  0.0120, -0.0009,  0.0041,  0.0073,  0.0287, -0.0067,
        -0.0050, -0.0006, -0.0002,  0.0008, -0.0012, -0.0166,  0.0044, -0.0027,
        -0.0034, -0.0002,  0.0079,  0.0072, -0.0276, -0.0107, -0.0773, -0.0005,
        -0.0330, -0.0250, -0.0123,  0.0039, -0.0111,  0.0527,  0.0210,  0.0073,
        -0.0190, -0.0034, -0.0176,  0.0009,  0.0605], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1552, -0.0050,  0.0106, -0.0205, -0.0318, -0.0142, -0.0475, -0.0469,
        -0.0352, -0.0542, -0.0539, -0.0134,  0.0069,  0.0008, -0.1206, -0.0117,
         0.0178, -0.0185, -0.0109, -0.0198, -0.0301, -0.0237, -0.0146, -0.0208,
        -0.0163, -0.0263, -0.0222,  0.0013,  0.0058, -0.0577, -0.0359, -0.0115,
         0.0382,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0165, -0.0623, -0.0482, -0.0309,  0.0017, -0.0237, -0.0410, -0.0172,
        -0.0210, -0.0668,  0.0091, -0.0219, -0.0230, -0.0280, -0.0317,  0.0762,
         0.0296, -0.0616, -0.2718,  0.0054, -0.0488,  0.0059,  0.0367,  0.0212,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0774, -0.1371, -0.0527, -0.0578, -0.0207, -0.0737, -0.0179, -0.0525,
         0.0991, -0.0282, -0.0170, -0.0434, -0.1101, -0.0212, -0.0035, -0.0226,
        -0.0091, -0.0026, -0.0456, -0.0244, -0.0103, -0.0271, -0.0219, -0.0240,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1116, -0.0247, -0.0594, -0.0491, -0.0254,  0.0556, -0.0312, -0.0309,
         0.0044, -0.0021, -0.0115,  0.0145, -0.0004,  0.0059, -0.0377, -0.0395,
        -0.0206,  0.0134, -0.0035, -0.0358, -0.0263, -0.0048,  0.0176, -0.0453,
         0.0134, -0.0077, -0.0243, -0.0105,  0.0751, -0.0371, -0.0453,  0.0327,
         0.0228,  0.0600,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1766,  0.0481,  0.0533, -0.0228,  0.0177, -0.1155,  0.0061, -0.0255,
         0.0011,  0.0163, -0.0132,  0.0260,  0.0102,  0.2216,  0.0360,  0.0652,
         0.0199, -0.0028,  0.0051,  0.0074, -0.0013,  0.0282, -0.0395, -0.0237,
        -0.0165,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2272, -0.0763,  0.0431,  0.0429,  0.0088, -0.0342,  0.0829,  0.0305,
         0.0570,  0.1402,  0.0575,  0.0037,  0.0436,  0.0387, -0.0683, -0.0450,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2332, -0.0170, -0.0228, -0.0594, -0.0382, -0.0267, -0.0727, -0.0455,
         0.0715,  0.0069, -0.0295,  0.0162, -0.0118, -0.0202, -0.0475,  0.0059,
        -0.0239, -0.0187, -0.0414, -0.0413, -0.0058, -0.0982, -0.0456,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0066,  0.2674,  0.1033,  0.0585,  0.1469,  0.0891,  0.0025,  0.0326,
         0.0144, -0.0139, -0.1063,  0.0334, -0.0353,  0.0897,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.1357,  0.0409,  0.0276, -0.0100, -0.0014, -0.0592,  0.0446, -0.0313,
        -0.0495, -0.0291,  0.0031,  0.0449, -0.0232,  0.0382,  0.0034, -0.0260,
         0.0510, -0.0418, -0.0278, -0.0185, -0.0139, -0.1174,  0.0577, -0.1038,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0568,  0.0976, -0.0854,  0.0708, -0.0530, -0.0918, -0.0901,  0.0337,
         0.1363,  0.1201, -0.1644,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0506, -0.1290,  0.0084, -0.0162, -0.0250, -0.0762, -0.1250, -0.0149,
        -0.0517, -0.0088,  0.0432,  0.0403,  0.0053, -0.0040, -0.0012, -0.0104,
        -0.0681, -0.0608,  0.1214,  0.0271, -0.1125,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1914, -0.0084,  0.0240,  0.0541,  0.0142, -0.0336, -0.1464, -0.1165,
         0.0106, -0.0806, -0.0016, -0.0737, -0.1993, -0.0455,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.3791e-01,  6.7618e-02, -4.0790e-04,  7.4108e-02, -7.4179e-02,
        -2.7375e-02, -1.6335e-02,  9.6498e-02, -4.9929e-02, -6.2607e-02,
        -4.3222e-02, -9.4148e-05, -1.0621e-01,  5.5994e-03,  1.2947e-02,
        -3.8725e-02, -4.3536e-02,  1.4367e-02, -4.3685e-02, -1.2824e-04,
         8.4519e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1565, -0.0409, -0.0579, -0.0959,  0.1607, -0.0769, -0.0360, -0.1603,
        -0.0667, -0.0683, -0.0506,  0.0070, -0.0223,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0539, -0.0688,  0.0022, -0.0025, -0.0151, -0.0099, -0.0187, -0.0580,
         0.0103,  0.0094,  0.0116, -0.0234,  0.0263, -0.0087, -0.0267, -0.0154,
        -0.0254,  0.0150,  0.0075, -0.0475, -0.0379, -0.0166, -0.0042,  0.0102,
        -0.0820, -0.0195, -0.0321, -0.0315, -0.0142, -0.0197, -0.0309, -0.0497,
        -0.0090,  0.0005,  0.0533, -0.0178, -0.0215, -0.0016,  0.0044, -0.0139,
        -0.0078,  0.0051,  0.0086, -0.0242,  0.0275], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0493,  0.0447, -0.0515, -0.0059, -0.0175,  0.0181,  0.0301, -0.0200,
         0.0099, -0.0308, -0.0316,  0.0203, -0.0160, -0.0483,  0.0074,  0.0219,
        -0.0544, -0.0112, -0.0019, -0.0175, -0.0084, -0.1762, -0.0056, -0.0517,
         0.0183,  0.0129, -0.0501,  0.0065,  0.0141,  0.0360, -0.0095,  0.0032,
         0.0473,  0.0520,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1947, -0.1985, -0.0566, -0.0094, -0.0139,  0.0190, -0.0187,  0.0091,
        -0.0531, -0.0153, -0.0045, -0.0036,  0.0506, -0.0100,  0.0044, -0.0115,
        -0.0126, -0.0501, -0.0064, -0.0047, -0.0395,  0.0805, -0.0067,  0.0127,
         0.0594,  0.0545,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1748, -0.1487, -0.0006, -0.0248,  0.0486, -0.0463, -0.0873, -0.0445,
        -0.0507,  0.0099, -0.0272, -0.0292, -0.0051, -0.0160, -0.0385, -0.0239,
        -0.0521,  0.0095, -0.0203, -0.0120, -0.0117, -0.0170,  0.1014,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.5864e-02, -5.5577e-02,  2.8131e-02, -2.7422e-02, -1.7908e-02,
        -1.0224e-04,  2.6958e-02, -2.8299e-02, -3.9389e-02, -1.7117e-02,
        -3.6967e-03,  1.1136e-02, -2.2154e-03, -2.1387e-02, -2.0022e-01,
        -3.9535e-02,  4.6396e-02, -3.7164e-02, -2.2043e-02, -4.4547e-02,
         2.2752e-03, -1.7583e-03, -5.1128e-04,  2.7747e-02, -2.7940e-03,
         1.7413e-02, -1.3500e-02, -2.6472e-02,  1.5030e-02,  7.4341e-03,
        -1.8480e-07,  9.8761e-04,  2.6212e-02,  7.3125e-03,  1.8890e-02,
        -1.0259e-02, -8.3951e-03, -2.6852e-02, -1.1846e-02, -4.9200e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0244, -0.0423, -0.0088, -0.0183, -0.0706, -0.0305,  0.0360, -0.0548,
        -0.0182, -0.0311,  0.0054, -0.0075, -0.0404, -0.0425,  0.0044, -0.0367,
        -0.0138, -0.0373,  0.0784,  0.0030, -0.0187, -0.0318,  0.0048, -0.0199,
        -0.0299, -0.0154,  0.0018, -0.0641, -0.0064,  0.0500, -0.0350, -0.0725,
        -0.0239,  0.0216,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0644, -0.0429, -0.0327, -0.0275, -0.0255, -0.0582, -0.0188,  0.0032,
         0.0024,  0.0384, -0.0031, -0.0067,  0.0843, -0.0026, -0.0824, -0.0213,
        -0.0017,  0.0115, -0.0219,  0.0135, -0.0056,  0.0121, -0.0110,  0.0167,
        -0.0246, -0.0833, -0.0507, -0.0244, -0.0135, -0.0159, -0.0116, -0.0005,
         0.0059, -0.0163, -0.0104,  0.0031, -0.0062, -0.0153, -0.0111, -0.0053,
        -0.0104, -0.0133, -0.0098,  0.0010, -0.0072,  0.0005, -0.0068, -0.0061,
        -0.0102, -0.0119,  0.0014,  0.0150], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0377, -0.1217,  0.0389,  0.0169, -0.0612, -0.0017, -0.0110, -0.0502,
         0.0552, -0.0086, -0.0088,  0.0051, -0.0149, -0.0572, -0.0516, -0.0377,
        -0.0967,  0.0102, -0.0021,  0.0418, -0.0166,  0.0009, -0.0391, -0.0825,
         0.0101,  0.0225,  0.0002, -0.0279, -0.0417,  0.0126, -0.0165,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0562,  0.0840, -0.0047, -0.0130, -0.0126, -0.0135,  0.0284,  0.0367,
         0.0238,  0.0102, -0.0036,  0.1000, -0.0942,  0.0191,  0.0150,  0.0683,
        -0.1533, -0.0132,  0.0452, -0.0205, -0.0393,  0.0388,  0.0447, -0.0035,
        -0.0584,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0030, -0.2063,  0.1164,  0.0005, -0.0358, -0.1241, -0.1398, -0.0202,
         0.1232, -0.1030, -0.0054, -0.1224,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2711, -0.0971, -0.0150, -0.0154, -0.0162, -0.0624,  0.0287,  0.0108,
        -0.0006, -0.0108, -0.0290,  0.0368,  0.0012, -0.0312,  0.0033, -0.0281,
        -0.0085, -0.0024,  0.0271, -0.0060,  0.0233, -0.0288,  0.0274, -0.0635,
         0.0155,  0.0354, -0.0019,  0.0865, -0.0161,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0572, -0.1294, -0.0038,  0.0771,  0.0329, -0.0737, -0.1269, -0.0202,
         0.1808, -0.2005, -0.0235, -0.0740,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0852, -0.2109,  0.1065, -0.1105, -0.0268, -0.0246, -0.1515, -0.0156,
        -0.0621, -0.0268, -0.0820,  0.0975,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.4863e-01, -5.6667e-02, -9.0339e-02, -4.3489e-02, -3.9216e-02,
         8.5932e-02,  1.1978e-02, -4.0428e-02, -6.9035e-02,  8.5238e-03,
        -2.7020e-03, -5.5276e-02,  1.1959e-04,  1.5459e-02,  2.5313e-02,
        -2.9919e-02, -3.8699e-02, -1.9740e-02,  8.9029e-02,  2.9502e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0289, -0.0909, -0.0345, -0.0295, -0.1656, -0.0499, -0.0571,  0.0163,
        -0.0042, -0.0076, -0.0099, -0.0244, -0.0263, -0.0182,  0.0087,  0.0039,
        -0.0099,  0.0201,  0.0010, -0.0131, -0.0011,  0.0002, -0.0066, -0.0181,
        -0.0129,  0.0405,  0.0037, -0.0074, -0.0005,  0.0027,  0.0221,  0.0106,
        -0.0259, -0.0725,  0.0042,  0.0440,  0.1070,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1017, -0.4600,  0.0207,  0.0104,  0.0781, -0.0113,  0.0159,  0.2310,
        -0.0708,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1545, -0.0439, -0.0980, -0.0605, -0.0068, -0.0364, -0.0379, -0.0393,
        -0.0437, -0.0213,  0.0210,  0.0329, -0.0268, -0.0552, -0.0433, -0.0069,
        -0.0346,  0.0112, -0.0207, -0.0331, -0.0376,  0.0012,  0.0073,  0.0276,
        -0.0420,  0.0107, -0.0454,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0403, -0.0923, -0.0545, -0.0389, -0.0151, -0.0255, -0.0230,  0.0003,
        -0.0013,  0.0006,  0.0068, -0.0210, -0.0276, -0.0357,  0.0108,  0.0085,
        -0.0178,  0.0019, -0.0254,  0.0548,  0.0309, -0.0125, -0.0362, -0.0067,
        -0.0271,  0.0167, -0.0002, -0.0207, -0.1563,  0.0209,  0.0071,  0.0186,
         0.0715, -0.0725,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.1185, -0.0382,  0.0162, -0.0079, -0.0291, -0.0075,  0.0054, -0.0341,
        -0.0219, -0.0239, -0.0373,  0.0100,  0.0070, -0.0031, -0.0019,  0.0041,
        -0.0710,  0.0440,  0.0020,  0.0119,  0.0079, -0.0532, -0.0291,  0.0141,
        -0.0384, -0.0148, -0.0220,  0.0014,  0.0145, -0.0009,  0.0106, -0.0229,
        -0.0861,  0.0072, -0.0199, -0.0034, -0.0424,  0.0004, -0.0144, -0.0010,
        -0.0082, -0.0761, -0.0106, -0.0021,  0.0038,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.3242e-02, -1.7094e-01, -5.5003e-03,  4.9310e-03, -1.0285e-01,
        -5.9616e-02, -5.6839e-02,  4.9265e-02, -4.8042e-02, -5.5307e-02,
        -1.2085e-01, -4.8009e-05, -6.2032e-02,  9.7777e-02, -3.2064e-02,
        -9.0708e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.3873, -0.0458,  0.0307,  0.0410, -0.0141, -0.0087, -0.0290,  0.0186,
        -0.0369, -0.0441, -0.0259, -0.0282, -0.0454, -0.0211, -0.0237,  0.0041,
        -0.0102, -0.0040,  0.0038,  0.0006,  0.0053,  0.0078, -0.0073, -0.0241,
         0.0268,  0.0371,  0.0683,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0601,  0.1340,  0.0054, -0.0324, -0.0390,  0.0035, -0.0056, -0.0148,
        -0.0020, -0.0338, -0.0108, -0.0171, -0.0223, -0.0053,  0.0007, -0.0138,
        -0.0046,  0.0360, -0.0005, -0.0142,  0.0019, -0.0241, -0.0279, -0.0792,
         0.0161,  0.0039, -0.0179, -0.0396,  0.0037, -0.0218,  0.0166,  0.0087,
        -0.0639,  0.0087, -0.0034,  0.0128, -0.0108, -0.0458, -0.0186,  0.0287,
        -0.0042, -0.0311, -0.0246, -0.0212,  0.0035,  0.0058], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2400,  0.0039,  0.0385,  0.0157, -0.0065, -0.0268,  0.0043, -0.0009,
         0.0133, -0.0013,  0.0034, -0.0360, -0.0098, -0.0141, -0.0252, -0.0198,
        -0.0167, -0.0365,  0.0408, -0.1961, -0.0385, -0.0071, -0.0166, -0.0073,
        -0.0113, -0.0137, -0.0058,  0.0515,  0.0449,  0.0538,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.3597,  0.0225, -0.0265,  0.0198, -0.0501, -0.0789, -0.1569, -0.1987,
        -0.0364, -0.0105,  0.0400,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0101, -0.1866, -0.0596, -0.0290, -0.0425, -0.0141, -0.0393, -0.1218,
         0.0211, -0.1403,  0.0115,  0.0079, -0.0095, -0.0800,  0.0134, -0.0300,
        -0.0149,  0.0161,  0.0447, -0.1076,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.4261, -0.0240, -0.1302, -0.0256, -0.0927, -0.0146, -0.0238, -0.1063,
        -0.0696,  0.0213, -0.0658,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0412, -0.1949, -0.0691, -0.1223, -0.0003, -0.0083,  0.0088, -0.0722,
        -0.0371,  0.0082, -0.0875, -0.0556, -0.0434, -0.0131, -0.0205, -0.0469,
         0.0191, -0.0117, -0.0284,  0.0617, -0.0497,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1140, -0.0557, -0.0205, -0.0366,  0.0421, -0.3019, -0.0384, -0.0548,
        -0.0290,  0.0545, -0.0281, -0.0602,  0.0159, -0.0158, -0.0467,  0.0012,
         0.0413, -0.0434,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0949, -0.0640, -0.1083, -0.0514,  0.0390, -0.0099, -0.0102, -0.0226,
        -0.0027, -0.0010,  0.0399,  0.0231,  0.0070, -0.0161, -0.0582,  0.0496,
        -0.0112, -0.0467, -0.0052, -0.0551,  0.0064, -0.0247, -0.0713, -0.0140,
        -0.0155, -0.0177, -0.0114, -0.0099, -0.0027, -0.0463,  0.0024, -0.0135,
        -0.0016, -0.0007, -0.0051,  0.0043, -0.0098, -0.0234, -0.0031,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1908, -0.1415,  0.0619, -0.0416,  0.0149, -0.0088, -0.0109, -0.0414,
         0.0159,  0.0331,  0.0184, -0.0458, -0.0056,  0.0121, -0.0161, -0.0108,
        -0.0258,  0.0136,  0.0075, -0.0219, -0.0217, -0.0095,  0.0096,  0.0062,
         0.0047, -0.0078, -0.0192,  0.0155, -0.0492, -0.0761,  0.0056, -0.0166,
         0.0065,  0.0135,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.0796, -0.0278,  0.0217, -0.0859, -0.2996, -0.0581, -0.2412, -0.0544,
         0.1318,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1373, -0.0971,  0.0153,  0.0188, -0.0316, -0.0072, -0.0045, -0.0317,
        -0.0065, -0.0170, -0.0214, -0.0059, -0.0046, -0.0032,  0.0143, -0.0133,
        -0.0292, -0.1253, -0.0385,  0.0026,  0.0278, -0.0131,  0.0200,  0.0301,
        -0.0308, -0.0234,  0.0214, -0.0154, -0.0006,  0.0038, -0.0009, -0.0185,
         0.0046, -0.0055, -0.0255, -0.0294,  0.0108, -0.0164, -0.0023, -0.0743,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0974, -0.0087, -0.0866, -0.0358, -0.0268, -0.0072, -0.0220,  0.0235,
        -0.0130, -0.0463,  0.0092, -0.0022, -0.0057, -0.0025,  0.0134,  0.0106,
         0.0042, -0.0491, -0.0234, -0.0065,  0.0078, -0.0069,  0.0177,  0.0007,
        -0.0069, -0.0046, -0.0384,  0.0425, -0.0326,  0.0019,  0.0364, -0.0618,
        -0.0002, -0.0076, -0.0179, -0.0324, -0.0237, -0.0351, -0.0020,  0.0066,
         0.0236,  0.0032, -0.0115, -0.0057, -0.0140, -0.0053, -0.0063, -0.0011,
        -0.0054, -0.0283, -0.0043,  0.0136,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0923, -0.1175, -0.0439,  0.0544, -0.0101, -0.0326, -0.0227,  0.0247,
         0.0338,  0.0181,  0.0064,  0.0532, -0.0406, -0.0153, -0.0464, -0.0371,
        -0.0220, -0.0154, -0.0502, -0.0450,  0.1117,  0.1066,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.6125e-01, -3.6267e-01, -2.2940e-02, -7.0036e-02, -2.1642e-02,
         5.0406e-04, -2.3405e-02, -2.5155e-02,  3.4402e-02,  2.4356e-02,
         3.5362e-04,  7.8835e-03, -3.3690e-02, -7.6942e-03,  1.5549e-03,
         6.3755e-03, -2.1159e-03, -3.6248e-03,  2.9349e-02, -4.2837e-02,
         1.1816e-01,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.8789e-02, -1.8611e-02,  2.4002e-02, -3.5895e-02,  1.6638e-02,
         1.6043e-04, -2.8364e-02, -1.7330e-02,  4.7263e-03, -1.7072e-02,
        -4.8896e-03, -8.2954e-03, -8.8754e-03,  1.8861e-02,  2.1117e-03,
        -1.0921e-02, -6.4889e-03, -2.2541e-02, -2.5092e-02,  3.5459e-03,
         2.6645e-03, -7.3430e-04,  5.8804e-03,  6.7689e-03,  3.4842e-03,
         2.9259e-03, -8.8026e-03,  9.7071e-05, -2.6532e-02,  3.1556e-03,
        -4.9674e-03, -2.3699e-02, -7.6889e-03,  1.0508e-03, -6.4679e-03,
         4.9573e-04, -9.3420e-03,  2.1020e-02,  2.5645e-02, -2.7603e-02,
        -2.0846e-01, -1.6794e-02, -3.9581e-02, -2.7376e-02, -1.0142e-02,
        -1.0663e-02, -1.7053e-02, -4.5122e-02,  1.6999e-02, -1.3885e-03,
        -2.5825e-02, -6.7711e-03,  2.7262e-02, -8.0896e-03, -3.6237e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0810, -0.0432, -0.0627, -0.0253, -0.0021, -0.0195,  0.0189,  0.0185,
        -0.0195,  0.0128,  0.0093,  0.0189, -0.0396, -0.0139, -0.0288,  0.0052,
         0.0097, -0.0020, -0.0151,  0.0158, -0.0223, -0.0230, -0.0194, -0.0021,
         0.0112, -0.0303, -0.0234, -0.0537, -0.0180,  0.0110,  0.0310, -0.0121,
        -0.0239, -0.0534,  0.0317, -0.0008, -0.0144,  0.0044, -0.0067,  0.0096,
        -0.0270, -0.0020, -0.0327,  0.0118,  0.0623,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0721, -0.0662, -0.1286, -0.0528,  0.0151, -0.0248, -0.0490,  0.0398,
         0.0220,  0.0728,  0.0090, -0.0660,  0.0194,  0.0047, -0.0227,  0.0293,
         0.0128, -0.0977,  0.0045, -0.0079, -0.1126, -0.0174, -0.0321, -0.0205,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0261,  0.0041, -0.0731,  0.0042,  0.0156, -0.0682, -0.0161, -0.0507,
         0.0105,  0.0208, -0.0079,  0.0059, -0.0560, -0.0406, -0.0236,  0.0021,
        -0.0242, -0.0255,  0.0013, -0.0297, -0.0187,  0.0131, -0.0128, -0.0835,
         0.0022, -0.0162, -0.0506,  0.0027, -0.0159, -0.0247,  0.0241, -0.0957,
        -0.1334,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0192, -0.2916, -0.0624, -0.0891, -0.0381, -0.1506, -0.0568,  0.0436,
        -0.0799,  0.0101, -0.0357,  0.0062,  0.0272, -0.0028, -0.0011, -0.0201,
        -0.0080,  0.0221,  0.0354,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0520, -0.0927, -0.0463,  0.0417, -0.0830, -0.0037, -0.0519, -0.0041,
        -0.0463,  0.0250, -0.0055, -0.0317, -0.0098, -0.0235,  0.0322,  0.0115,
         0.0029,  0.0012, -0.0405, -0.0312, -0.0134, -0.0975, -0.0217, -0.0529,
         0.1779,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1235, -0.0190, -0.0371,  0.0021, -0.0408, -0.1065, -0.3066, -0.0356,
        -0.0162,  0.0069,  0.0042, -0.0644, -0.0363, -0.0528, -0.0396,  0.0270,
        -0.0814,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0173, -0.0200, -0.0148, -0.0292, -0.1189, -0.0464, -0.0147, -0.0631,
         0.0201, -0.0460,  0.0014, -0.0128, -0.0355,  0.0050, -0.0045,  0.0273,
        -0.0240, -0.0549,  0.0024, -0.0004,  0.0069, -0.0285, -0.0738, -0.0012,
         0.0137, -0.0362,  0.0189, -0.0562,  0.0069, -0.0111, -0.0350, -0.1529,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0464, -0.1204, -0.0271, -0.0101,  0.0249, -0.0231, -0.0363, -0.1033,
        -0.0164,  0.0049, -0.0094, -0.0072,  0.1520, -0.0070, -0.0085,  0.0104,
        -0.0532, -0.0171, -0.0003,  0.0282,  0.0104, -0.0132, -0.0274, -0.1136,
         0.0507, -0.0028,  0.0145,  0.0611,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0061, -0.1654, -0.0021, -0.0112, -0.0199, -0.0342, -0.0940, -0.0271,
        -0.0453, -0.0491, -0.0093, -0.0266, -0.0143,  0.0003, -0.0241, -0.0253,
        -0.0267, -0.0178, -0.0097, -0.0170, -0.0259, -0.0314,  0.0320, -0.0274,
        -0.0305,  0.0175,  0.0385, -0.0952,  0.0762,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1293,  0.0447, -0.0452,  0.0208,  0.0016,  0.0635,  0.1045,  0.0371,
        -0.0025,  0.0236,  0.0434,  0.0248,  0.0308,  0.0180,  0.0022, -0.0243,
        -0.0146, -0.0253,  0.0012, -0.0119,  0.0107, -0.0136, -0.0222, -0.0065,
         0.0057,  0.0055, -0.0216, -0.0069,  0.0305,  0.0041,  0.0200,  0.0311,
         0.1524,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2801, -0.0519, -0.0705,  0.0532, -0.0031, -0.0824, -0.0153, -0.0138,
        -0.0112, -0.0567, -0.0069, -0.0113, -0.0099, -0.0237, -0.0399,  0.0138,
        -0.0354, -0.0096, -0.0566, -0.0093, -0.0856, -0.0157, -0.0206,  0.0233,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0303, -0.1067,  0.0168, -0.0407, -0.0356, -0.1019, -0.0825, -0.0578,
        -0.0518, -0.0329, -0.0199,  0.0240, -0.0393,  0.0051,  0.0549,  0.0297,
        -0.0013, -0.0305, -0.0194, -0.1636, -0.0128, -0.0295, -0.0130,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.5157e-03, -4.9154e-02, -2.0653e-02,  3.4968e-03, -1.4630e-02,
        -5.0652e-02, -1.9002e-02,  4.2992e-02,  8.2540e-05, -2.1836e-02,
         8.4690e-02, -1.6595e-02, -2.5139e-02, -1.9741e-02,  4.1265e-03,
         7.1289e-04, -2.9978e-02, -2.0628e-02, -5.9707e-02, -3.8028e-02,
        -2.1636e-02, -2.9667e-02,  1.2860e-02, -6.0590e-02, -2.7867e-02,
        -1.7554e-02, -6.3630e-03, -3.9851e-02, -4.9933e-03, -1.0841e-02,
        -2.1907e-02, -8.7855e-03, -1.3745e-02,  4.5115e-02, -1.3546e-02,
        -3.0702e-02, -8.3845e-03, -3.5598e-02, -6.4635e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.8082e-02, -1.4252e-01, -7.6320e-02, -2.3523e-02, -6.6075e-02,
         1.3117e-05, -9.8143e-03,  2.7947e-02, -6.0936e-03,  3.7501e-02,
        -2.5156e-02, -7.7878e-03,  1.3167e-02, -3.0103e-02, -1.4815e-02,
        -3.3388e-02, -8.0475e-02, -4.9718e-03, -6.3962e-03, -7.5841e-03,
        -9.6361e-03, -5.5452e-03, -5.8400e-03,  4.3668e-03, -3.4439e-02,
         1.8031e-02, -1.8660e-02,  5.0700e-03, -2.3147e-02, -4.7214e-03,
        -2.9237e-02,  2.5686e-04,  1.5368e-02, -3.1704e-03, -8.2149e-03,
        -4.7527e-03, -1.2674e-02, -3.1929e-02, -1.2036e-02, -1.9954e-02,
        -4.6312e-04, -2.5929e-02, -4.4829e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0503, -0.1247,  0.0246,  0.0112,  0.0006, -0.0033, -0.0071, -0.0085,
        -0.0573,  0.0591,  0.0064, -0.0020, -0.0271, -0.0045,  0.0002,  0.0051,
        -0.0282, -0.0406,  0.0066,  0.0047,  0.0169,  0.0015, -0.0058, -0.0042,
        -0.0077, -0.0127,  0.0127,  0.0025,  0.0057, -0.0047, -0.0100,  0.0323,
         0.0335, -0.0032, -0.0173,  0.0327, -0.0091, -0.0020, -0.0211, -0.0055,
        -0.0252, -0.0277,  0.0002, -0.0273, -0.0019, -0.0012,  0.0003,  0.0101,
         0.0148,  0.0059,  0.0358,  0.0092, -0.0279, -0.0056,  0.0242,  0.0555,
         0.0142], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0177, -0.2734, -0.0830, -0.0047, -0.0189, -0.0681, -0.0742, -0.0571,
         0.0175, -0.0326, -0.1197, -0.0812, -0.1071,  0.0449,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1123, -0.0621, -0.0921, -0.0118,  0.0130, -0.0530, -0.0337,  0.0019,
        -0.0089, -0.0066, -0.0087, -0.1333, -0.0806,  0.0086, -0.0195,  0.0194,
        -0.0116, -0.0287,  0.0164, -0.0036, -0.0611, -0.0508, -0.0111, -0.0022,
         0.0144,  0.0325,  0.0023, -0.0226,  0.0518, -0.0254,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2306, -0.1259,  0.1175,  0.1797,  0.1050,  0.0019,  0.1855, -0.0540,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0606, -0.0282, -0.0102, -0.0482, -0.0293, -0.0918, -0.0119,  0.0011,
         0.0163, -0.0172, -0.0039, -0.0436, -0.0240, -0.0679, -0.0168,  0.0062,
         0.0218, -0.0075, -0.0270, -0.0210,  0.0240,  0.0171, -0.0346, -0.0439,
        -0.0478, -0.0200, -0.0619, -0.0287, -0.0379, -0.0446,  0.0354, -0.0451,
        -0.0044,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0153, -0.0403, -0.0384,  0.0258, -0.0235, -0.0053,  0.0079, -0.0267,
        -0.0050, -0.0307,  0.0015,  0.0006, -0.0121, -0.0232,  0.0231, -0.0004,
        -0.0033,  0.0055, -0.0064,  0.0109,  0.0082, -0.0015, -0.0033,  0.0063,
         0.0084, -0.0127, -0.0028, -0.0374, -0.0020,  0.0040, -0.0020, -0.0047,
        -0.0097, -0.0665,  0.0110, -0.0127, -0.0135,  0.0194, -0.0054, -0.0191,
         0.0021,  0.0002, -0.0326, -0.0536, -0.0039, -0.0057, -0.0158, -0.0172,
        -0.0171, -0.0079, -0.0210, -0.0180, -0.0078, -0.0031,  0.0030,  0.0025,
         0.0042, -0.0185, -0.0124, -0.0185, -0.0190, -0.0015, -0.0660, -0.0048,
         0.0025, -0.0078,  0.0081,  0.0190, -0.0034,  0.0044, -0.0420],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1696, -0.0171,  0.0035,  0.0226, -0.0394, -0.0077, -0.1575, -0.0185,
        -0.0117,  0.0131, -0.0018, -0.0147, -0.0261,  0.0002,  0.0072, -0.0267,
         0.0013,  0.0072,  0.0200, -0.0414,  0.0088,  0.0117,  0.0441, -0.0189,
         0.0098,  0.0033, -0.0185, -0.0119,  0.0304, -0.0115,  0.0185,  0.0245,
        -0.0150, -0.0061, -0.0500, -0.0304, -0.0027, -0.0043, -0.0093,  0.0094,
         0.0032, -0.0503,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0452, -0.1666, -0.0104,  0.0115, -0.0064,  0.0323, -0.0082, -0.0036,
        -0.0380, -0.0295, -0.0249, -0.0268, -0.0630, -0.0149,  0.0083,  0.0270,
         0.0365, -0.0136, -0.0677, -0.0081, -0.0137, -0.0078, -0.0052, -0.0178,
        -0.0010, -0.0257, -0.0268, -0.0218, -0.0560,  0.0162, -0.0136, -0.0106,
        -0.0051,  0.0142, -0.0087,  0.0766,  0.0367,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0370, -0.3006,  0.0100, -0.0483,  0.0697,  0.0457, -0.0937, -0.0135,
         0.0692, -0.0538, -0.0400, -0.0694, -0.0304,  0.0795,  0.0391,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1805, -0.1751, -0.0307, -0.0478, -0.0337, -0.0211, -0.0027, -0.0005,
        -0.0879, -0.0190, -0.0385,  0.0459, -0.0968,  0.0196, -0.0038,  0.0297,
        -0.0012,  0.0277,  0.1032,  0.0347,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0203, -0.0570, -0.0165, -0.0408, -0.0773, -0.0217, -0.0248, -0.0156,
        -0.0309, -0.0353, -0.0112, -0.0205, -0.0098,  0.0061, -0.0228, -0.0062,
        -0.0156, -0.0187,  0.0463, -0.0249, -0.0063, -0.0036, -0.0003,  0.0111,
        -0.0435, -0.0311, -0.0265,  0.0304, -0.0068, -0.0270, -0.0281,  0.0098,
         0.0017, -0.0043, -0.0059, -0.0144, -0.0152, -0.0090, -0.0258, -0.0495,
         0.0076,  0.0052,  0.1149,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0930, -0.0788,  0.0090, -0.0443,  0.0046, -0.0300, -0.0224, -0.1125,
        -0.0065, -0.0288,  0.0187,  0.0100, -0.0273,  0.0111, -0.0309, -0.0185,
         0.0041, -0.0203, -0.0124, -0.0165, -0.0281, -0.0199, -0.0182, -0.0211,
        -0.0143,  0.0202,  0.0221, -0.0129, -0.0359,  0.0012, -0.0121, -0.0305,
        -0.0015, -0.0102, -0.0277,  0.0064, -0.0298,  0.0882,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0221, -0.1404, -0.1253, -0.2865,  0.0392, -0.0570,  0.0288, -0.0084,
        -0.0075,  0.0060, -0.0079, -0.0107, -0.0128, -0.0055, -0.0098, -0.0156,
        -0.0552, -0.0585,  0.0006, -0.0164,  0.0458, -0.0399,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0837, -0.0595, -0.0394, -0.0216, -0.0285, -0.0101, -0.0273, -0.0207,
        -0.0860, -0.0786, -0.0016,  0.0002, -0.0055,  0.0008, -0.0051, -0.0165,
        -0.0603, -0.0103, -0.0288, -0.0155,  0.0375,  0.0214, -0.0216, -0.0281,
        -0.0217, -0.0070, -0.0158,  0.0077, -0.0046, -0.0156,  0.0391, -0.0011,
         0.0325, -0.0336, -0.0214, -0.0456,  0.0195, -0.0197, -0.0065,  0.0001,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0626, -0.1042, -0.0053, -0.0083, -0.0119, -0.0534, -0.0454,  0.0285,
        -0.0093, -0.0616,  0.0090, -0.0345, -0.0298, -0.0383, -0.0282, -0.0130,
        -0.0467,  0.0149, -0.0837,  0.0038, -0.0067,  0.0050,  0.0150, -0.0454,
        -0.0289, -0.0477, -0.0053, -0.0361, -0.0340, -0.0499,  0.0051,  0.0284,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.2063e-02, -2.9276e-01,  2.5157e-02,  3.6680e-02, -1.3744e-02,
        -1.4358e-02, -2.4953e-03, -3.6383e-02, -9.8676e-02,  1.4899e-04,
         1.0510e-03, -1.7691e-02,  1.7019e-02,  1.9769e-03, -1.0413e-02,
        -4.5749e-03, -2.0755e-03, -4.5352e-03,  6.7290e-03, -1.1404e-02,
        -2.1896e-02,  1.1505e-02,  1.3586e-02, -8.4231e-02, -2.4302e-03,
         4.5450e-02, -3.2272e-02, -1.7247e-02,  2.4376e-02, -5.4648e-03,
         2.0435e-02, -5.2254e-03,  1.9441e-03,  2.2733e-02,  1.1273e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.1221, -0.0914, -0.0413, -0.0969,  0.0570, -0.0281, -0.0508, -0.0816,
        -0.0714, -0.0902, -0.0348, -0.0280,  0.0221, -0.0439, -0.0387,  0.0117,
         0.0005, -0.0219, -0.0146, -0.0113,  0.0417,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0798, -0.3087,  0.0401, -0.0830, -0.0167, -0.0602, -0.0149, -0.0545,
        -0.0895, -0.0414,  0.0117, -0.0025, -0.0285,  0.0476, -0.0161,  0.0181,
        -0.0212,  0.0340, -0.0314,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.3813, -0.0451, -0.0224,  0.0622, -0.0178, -0.0216, -0.0997, -0.0062,
        -0.0357, -0.0940, -0.0839, -0.0037, -0.0073,  0.0021,  0.0094, -0.0550,
        -0.0028, -0.0136, -0.0023, -0.0338,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0381, -0.2463, -0.0420, -0.0319, -0.0778,  0.0121, -0.0840,  0.0234,
         0.0575,  0.0066,  0.0079, -0.0203, -0.0337, -0.0159, -0.0274, -0.0208,
        -0.0475, -0.0472,  0.0129, -0.0046, -0.0309,  0.0400, -0.0712,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0440, -0.1783, -0.0708, -0.0376,  0.0174, -0.0038,  0.0264,  0.0027,
         0.0133,  0.0363,  0.0167, -0.0241, -0.0168, -0.1442,  0.0053,  0.0231,
        -0.0808, -0.0018, -0.0218,  0.0115, -0.0440, -0.0067, -0.0323,  0.0582,
        -0.0822,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0589, -0.2505,  0.0012,  0.0073,  0.0070, -0.0043,  0.0303, -0.0239,
         0.0113,  0.0134, -0.0555, -0.0082, -0.0101, -0.0451,  0.0707, -0.0086,
        -0.0073, -0.0056, -0.0371,  0.0023, -0.0037,  0.0044, -0.0175, -0.0254,
        -0.0121,  0.0096,  0.0331,  0.0143, -0.0065,  0.0005, -0.0301,  0.0168,
        -0.0271, -0.0600,  0.0800,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0190,  0.0308, -0.0725, -0.0599, -0.0515, -0.0426, -0.0289,  0.0159,
        -0.1083,  0.0077, -0.0361, -0.0259, -0.0426, -0.0622, -0.1011, -0.0386,
        -0.0393, -0.0237, -0.0770, -0.0432, -0.0205, -0.0528,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1587,  0.0414,  0.0513, -0.0249,  0.0094,  0.0614, -0.0114, -0.0124,
        -0.0385,  0.0275,  0.1181, -0.0062, -0.0554, -0.0122,  0.1192,  0.0353,
         0.0313, -0.0062,  0.0373,  0.0406,  0.1014,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0285, -0.2232,  0.0044,  0.0551,  0.0128,  0.0278, -0.0542, -0.1033,
        -0.0271, -0.0430, -0.0259,  0.0234, -0.0674,  0.0032, -0.0021, -0.1149,
         0.0547,  0.0906,  0.0384,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0464, -0.0599, -0.0242, -0.0537, -0.0225, -0.0328, -0.0096,  0.0148,
         0.0034, -0.0097,  0.0199, -0.0093, -0.0046, -0.0105, -0.0024,  0.0115,
         0.0187, -0.0215, -0.0216,  0.0038,  0.0032, -0.0044, -0.0130,  0.0139,
        -0.0555, -0.0523,  0.0368, -0.0184,  0.0141, -0.0411, -0.0205,  0.0065,
        -0.0233, -0.0384, -0.0026,  0.0043, -0.0282,  0.0112, -0.0190, -0.0135,
         0.0163, -0.0237, -0.0041,  0.0042, -0.0054,  0.1016, -0.0236],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0307, -0.0582, -0.0388,  0.0050,  0.0022, -0.0203,  0.0048,  0.0706,
        -0.0160, -0.0173,  0.0175,  0.0265,  0.0064, -0.0104,  0.0247, -0.0433,
        -0.0700,  0.0066, -0.0149,  0.0072, -0.0475,  0.0067,  0.0171, -0.0536,
         0.0511, -0.0394, -0.0934,  0.0419,  0.0050,  0.0598, -0.0289,  0.0059,
        -0.0263, -0.0322,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1010, -0.1376,  0.0275,  0.0092,  0.0108, -0.0164,  0.0107, -0.0157,
        -0.0112, -0.0152, -0.0015,  0.0083, -0.0402, -0.0796,  0.0135,  0.0490,
        -0.0115, -0.0271, -0.0718, -0.0641,  0.0116, -0.0107, -0.0003, -0.0195,
         0.0138, -0.0330, -0.0123, -0.0183, -0.0042, -0.0282,  0.0195, -0.0341,
        -0.0299, -0.0427,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0728, -0.1146,  0.0145, -0.0010, -0.0139, -0.0230,  0.0977, -0.0044,
         0.0581, -0.0801, -0.0460,  0.0158, -0.0262, -0.2048, -0.0634, -0.0296,
         0.0429, -0.0913,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1634, -0.0144,  0.0502, -0.0096,  0.0256, -0.0061, -0.0024, -0.0520,
        -0.0035,  0.1437,  0.0230, -0.0199,  0.0540, -0.0068, -0.0177,  0.0138,
        -0.0233, -0.0618, -0.0547,  0.0427,  0.0045,  0.0456,  0.0137,  0.0205,
         0.0350,  0.0095,  0.0825,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1484,  0.0420,  0.0856,  0.0265,  0.0223,  0.0772,  0.0366,  0.0469,
         0.0661,  0.0526,  0.0069, -0.0045,  0.0525,  0.0154,  0.0012, -0.0278,
         0.0070,  0.0017,  0.0111,  0.1391,  0.0273,  0.0015, -0.0007, -0.0197,
         0.0638, -0.0158,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0578, -0.1899,  0.0178, -0.0024, -0.0948, -0.1435, -0.0607,  0.0035,
        -0.0299, -0.0226, -0.0819,  0.0154, -0.0176,  0.0133, -0.0014, -0.0415,
        -0.0128, -0.0307, -0.0082,  0.1500, -0.0043,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0616, -0.0791, -0.0943, -0.2792, -0.0447,  0.0952,  0.0496, -0.0118,
        -0.0180,  0.0773, -0.0425, -0.0031,  0.0125,  0.0114,  0.1199,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1943,  0.0502,  0.0211,  0.0434, -0.1131,  0.0666,  0.0233,  0.0475,
         0.0855, -0.0168,  0.0374,  0.0368,  0.1038,  0.0463, -0.0018, -0.0024,
        -0.0835,  0.0263,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1386, -0.0709, -0.0052, -0.1685, -0.0973,  0.0035,  0.0229, -0.0463,
        -0.0187, -0.0297, -0.0160, -0.0125,  0.0619, -0.0158, -0.0265, -0.0644,
        -0.0397,  0.0643,  0.0974,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0873, -0.0488, -0.0170,  0.0072, -0.0490, -0.0216,  0.0509,  0.0151,
        -0.0144, -0.0040, -0.0112,  0.0109,  0.0018, -0.0018,  0.0023,  0.0004,
        -0.0224,  0.0182, -0.0023, -0.0789, -0.0526, -0.0421,  0.0045,  0.0438,
         0.0208, -0.0252, -0.0173, -0.0099, -0.0014, -0.0122,  0.0016,  0.0059,
        -0.0298, -0.0165, -0.0035, -0.0139, -0.0550,  0.0066, -0.0071,  0.0196,
        -0.0189,  0.0787, -0.0419, -0.0054], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0028, -0.0317,  0.0029, -0.0435, -0.0278, -0.0058, -0.0183,  0.0064,
        -0.0773,  0.0161, -0.0258, -0.0097,  0.0462, -0.0148,  0.0419, -0.0178,
         0.0202, -0.1229, -0.0189, -0.0020, -0.0201, -0.0130, -0.0027, -0.0037,
        -0.0059,  0.0026, -0.0113, -0.0443,  0.0619, -0.0401, -0.0312, -0.0160,
        -0.0160,  0.0869, -0.0914,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0926, -0.2055, -0.0253, -0.0077,  0.0145, -0.0068, -0.0064, -0.0690,
        -0.0499, -0.0271, -0.0131, -0.0354, -0.0482, -0.0224, -0.0527, -0.0734,
         0.0236,  0.0027,  0.0539, -0.0278, -0.0117, -0.0219,  0.0485, -0.0597,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0301, -0.0748, -0.0560, -0.0427, -0.0150, -0.0225,  0.0241, -0.0311,
        -0.0197, -0.0230, -0.0406, -0.0372,  0.0293, -0.0268, -0.0466,  0.0164,
         0.0293, -0.0123, -0.0659,  0.0347, -0.0021, -0.0042, -0.0157, -0.0332,
         0.0046, -0.0256, -0.0399, -0.0097,  0.0005, -0.0073, -0.0575, -0.0197,
        -0.0356, -0.0660,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1415, -0.1473,  0.0148,  0.0362,  0.0183, -0.0173, -0.0198,  0.0054,
         0.0053, -0.0071,  0.0027,  0.0308, -0.0236, -0.0101, -0.0130, -0.1186,
        -0.0121,  0.0168, -0.0160,  0.0693,  0.0109, -0.0036, -0.0403, -0.0031,
        -0.0042, -0.0314,  0.0138,  0.0278, -0.0639,  0.0753,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.1914, -0.0556, -0.0651, -0.1144,  0.0427, -0.0500, -0.0951, -0.0379,
         0.0045, -0.0253,  0.0334, -0.0930, -0.1915,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2258, -0.0461,  0.1271, -0.0298,  0.0286, -0.0518,  0.0077, -0.0017,
        -0.0583,  0.0316, -0.0382, -0.0071,  0.0681, -0.0279, -0.0446,  0.0505,
        -0.0684, -0.0867,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.5556, -0.0211, -0.0265, -0.0726, -0.0384,  0.0583, -0.1701, -0.0574,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0401, -0.0393, -0.0213, -0.0175, -0.0177, -0.0104,  0.0007, -0.0198,
         0.0262,  0.0103, -0.0161, -0.0139, -0.0285,  0.0088, -0.0129, -0.0109,
        -0.0043,  0.0030, -0.0149,  0.0106, -0.0142,  0.0236, -0.0340, -0.0139,
        -0.0229, -0.0150, -0.0726, -0.0148, -0.0954, -0.0092, -0.0107, -0.0379,
         0.0048, -0.0912,  0.0051, -0.0066, -0.0250, -0.0265,  0.0143,  0.0038,
        -0.0170, -0.0318, -0.0085,  0.0178,  0.0227,  0.0135, -0.0202],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0173, -0.0854, -0.2850, -0.1294,  0.1135,  0.0014,  0.1293,  0.0249,
        -0.1479, -0.0386, -0.0272,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1638, -0.1130, -0.0569,  0.0031, -0.0499, -0.0169, -0.0450, -0.0975,
         0.0302, -0.0707, -0.0335, -0.0693, -0.0142, -0.0124,  0.1014, -0.0233,
         0.0507,  0.0481,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2837, -0.1761, -0.0729, -0.0697, -0.0175, -0.0123, -0.0462,  0.0064,
        -0.0337, -0.0103,  0.0024,  0.0582,  0.0979,  0.0980,  0.0147,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0186, -0.1995, -0.0992, -0.0522, -0.0463,  0.0384, -0.1535,  0.0051,
        -0.0127, -0.1469, -0.0121,  0.0826,  0.0352, -0.0404,  0.0574,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1067, -0.1102,  0.0180, -0.0220,  0.0020,  0.0076,  0.0218, -0.0286,
        -0.1187, -0.0568,  0.0230, -0.0326, -0.0747, -0.0288, -0.0229,  0.0576,
        -0.0520,  0.0283,  0.0044,  0.0286, -0.0281, -0.0092,  0.0209, -0.0061,
         0.0181,  0.0722,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1221, -0.3219,  0.0089, -0.0005,  0.0288,  0.0616, -0.0576, -0.1020,
        -0.1667, -0.0202, -0.0028,  0.0244,  0.0266,  0.0393,  0.0165,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1234, -0.1178, -0.0242, -0.0427,  0.0006, -0.0125, -0.0334,  0.0145,
        -0.0004, -0.1411,  0.0359, -0.0447, -0.0403, -0.0205,  0.0676, -0.0090,
        -0.0109,  0.0179, -0.0128,  0.0064, -0.0177,  0.0424, -0.0355,  0.0308,
        -0.0410,  0.0559,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0412, -0.0703, -0.0017, -0.0056, -0.0360, -0.0447, -0.0074, -0.0534,
         0.0086, -0.0620, -0.1588, -0.0128, -0.0014,  0.0460, -0.0456, -0.0274,
         0.0380, -0.0411, -0.0401, -0.0105,  0.0485, -0.0078,  0.0123, -0.0055,
         0.1567, -0.0165,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.1398, -0.1575, -0.0427, -0.0050, -0.0367,  0.0089, -0.0124, -0.0225,
         0.0101, -0.0212,  0.0013, -0.0008, -0.0571, -0.0256, -0.0210,  0.0066,
        -0.0438,  0.0277, -0.0654,  0.0020,  0.0045, -0.0164, -0.0213, -0.0187,
        -0.0137, -0.0295, -0.0323,  0.0085,  0.0154,  0.0223,  0.0037, -0.0296,
        -0.0013, -0.0173,  0.0131, -0.0442,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0342, -0.0580, -0.0710, -0.0429,  0.0024,  0.0106, -0.0147, -0.0487,
        -0.0010,  0.0376, -0.0142,  0.0233, -0.0271,  0.0129,  0.0095, -0.0045,
         0.0218, -0.0117,  0.0293, -0.0506,  0.0399, -0.0283, -0.0162,  0.0028,
        -0.0277, -0.0263,  0.0027, -0.0098, -0.0407,  0.0454, -0.0333, -0.0325,
        -0.0363, -0.0099,  0.0343, -0.0878,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0061,  0.2111,  0.1062,  0.0952, -0.0945,  0.0171,  0.0198,  0.0271,
         0.0541,  0.0454,  0.0652,  0.0006,  0.0239,  0.0202, -0.1066, -0.0026,
         0.0107,  0.0395,  0.0355,  0.0186,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1588, -0.0559, -0.0011, -0.0362,  0.0096, -0.0680, -0.0053, -0.0020,
         0.0086, -0.0282, -0.0003, -0.0317,  0.0003,  0.0264, -0.0280,  0.0073,
         0.0585, -0.0325, -0.0096, -0.0057,  0.0141, -0.0134, -0.0093, -0.0339,
         0.0062, -0.0021, -0.0071,  0.0865, -0.0401,  0.0101, -0.0076, -0.0249,
        -0.0119, -0.0770, -0.0005,  0.0620, -0.0069,  0.0121,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0910, -0.0314, -0.0041, -0.0723,  0.0062, -0.0018, -0.0091, -0.0053,
        -0.0205, -0.0175, -0.0415, -0.0883, -0.0279,  0.0267, -0.0367, -0.0174,
         0.0214, -0.0388, -0.0300, -0.0074, -0.0143,  0.0026,  0.0214,  0.0068,
         0.0194, -0.0062,  0.0096,  0.0011, -0.0589,  0.0306, -0.0022,  0.0257,
         0.0176, -0.0074, -0.0084,  0.0431,  0.0236, -0.0279, -0.0166,  0.0583,
         0.0033], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1504,  0.2937, -0.0155,  0.0192,  0.0360,  0.0523, -0.0411, -0.0395,
         0.0162, -0.0080, -0.0011,  0.0198,  0.0353,  0.0488,  0.0168,  0.0037,
         0.0426,  0.0232, -0.0138,  0.0601, -0.0370,  0.0258,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0532, -0.0728, -0.0244, -0.0132, -0.0361, -0.0191, -0.0050, -0.0608,
        -0.0266, -0.0385,  0.0203, -0.0252, -0.0029, -0.0182,  0.0070, -0.0083,
        -0.0164,  0.0189, -0.0205, -0.0169, -0.1252, -0.0051, -0.0413, -0.0121,
         0.0031, -0.0078,  0.0091, -0.0335, -0.0974, -0.0046, -0.0091, -0.0159,
        -0.0336,  0.0958, -0.0020,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0664, -0.0537, -0.0496, -0.0613, -0.1509,  0.0748, -0.0270,  0.0166,
        -0.0841,  0.0090, -0.0380, -0.0443,  0.0002, -0.0038, -0.0171, -0.0022,
         0.0062, -0.0139, -0.0523, -0.0385, -0.0156, -0.0397, -0.0011, -0.0191,
        -0.0267, -0.0346, -0.0355, -0.0081, -0.0097,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0371, -0.1139, -0.0156, -0.0175, -0.0248, -0.0381, -0.0214,  0.0896,
        -0.0128, -0.0178, -0.0255, -0.0358,  0.0100,  0.0084,  0.0018, -0.0066,
        -0.0247,  0.0032, -0.0443, -0.1168, -0.0558, -0.0789,  0.0150, -0.1787,
         0.0061,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0667,  0.0023, -0.0515,  0.0073, -0.0798, -0.0594, -0.0420, -0.0020,
        -0.0585, -0.0525, -0.0427, -0.0316, -0.0280,  0.1086, -0.0771, -0.1278,
         0.0478, -0.0297, -0.0150,  0.0235, -0.0153,  0.0052, -0.0259,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0284,  0.0005, -0.0125, -0.0077, -0.0730, -0.0067,  0.0409, -0.0621,
        -0.0364, -0.0178, -0.0302,  0.0129, -0.0291, -0.0317, -0.0104, -0.0633,
        -0.0171, -0.0192, -0.0620, -0.0319,  0.0044, -0.0453, -0.0010,  0.0147,
         0.0056,  0.0229, -0.1159, -0.0039, -0.0007, -0.0133, -0.0399, -0.0117,
         0.0020, -0.0283,  0.0028,  0.0287, -0.0113,  0.0339, -0.0200,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1012, -0.1581,  0.0469, -0.1524, -0.2622, -0.0810,  0.0261, -0.0290,
        -0.0600,  0.0389, -0.0032, -0.0409,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0784, -0.0463, -0.1633, -0.0149, -0.0130, -0.0268, -0.0236, -0.0138,
        -0.0138, -0.0220, -0.0182,  0.0029, -0.0148, -0.0140, -0.0204, -0.0373,
        -0.0268, -0.0328, -0.0058, -0.0315, -0.0181, -0.0043, -0.0093, -0.0014,
         0.0118, -0.0089,  0.0156,  0.0155,  0.0055, -0.0012, -0.0235, -0.0193,
        -0.0091, -0.0056, -0.0278, -0.0541, -0.0142, -0.0093, -0.0062, -0.0059,
         0.0148,  0.0263,  0.0467,  0.0231,  0.0022,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2149, -0.3607,  0.0674, -0.0695, -0.0724,  0.0108, -0.0223, -0.0021,
        -0.0480, -0.0382, -0.0547,  0.0390,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1534, -0.4324,  0.2090,  0.0537,  0.0458,  0.0146, -0.0390,  0.0520,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.8951e-02,  1.6902e-03,  1.6430e-02,  5.1889e-02,  5.8437e-03,
        -3.1151e-02, -4.7868e-03, -1.3133e-02, -1.6885e-02, -1.0914e-01,
        -9.7822e-03, -2.8635e-02,  9.1893e-03, -8.7747e-03,  6.7956e-03,
         4.4915e-03, -6.4577e-03,  8.6548e-03, -7.3841e-03, -7.1066e-04,
        -4.6349e-03,  2.0521e-03,  5.3742e-03, -1.4254e-02, -1.3058e-02,
        -4.8327e-03, -3.0654e-03,  2.2542e-04, -1.7589e-02, -1.0964e-02,
        -5.0203e-03, -5.1076e-03, -4.1768e-03, -8.4845e-03,  1.0151e-02,
        -1.4574e-02, -1.9276e-02,  1.1626e-02, -1.0041e-02, -3.1087e-03,
         6.5240e-03, -9.1420e-04,  2.8021e-02, -9.2550e-03,  2.2955e-02,
        -3.0560e-02, -2.4083e-02, -2.1517e-02, -2.2632e-02, -6.3056e-02,
        -2.0506e-02, -1.1972e-02, -1.5449e-02, -1.8456e-02, -9.6821e-03,
        -3.7995e-03, -3.0498e-02,  2.1595e-03,  5.0497e-05,  3.5129e-02,
        -2.4385e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0293, -0.0766, -0.0079, -0.0034, -0.0853,  0.0964, -0.0596, -0.0044,
        -0.0266, -0.0057, -0.0537, -0.0496, -0.0030, -0.0268, -0.0533, -0.0077,
         0.0104, -0.0002, -0.0030, -0.0272, -0.0401, -0.0112, -0.0161,  0.0186,
        -0.0480, -0.0150, -0.0344, -0.0121, -0.0282, -0.0435, -0.0126, -0.0637,
        -0.0262,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1516, -0.0531, -0.0456, -0.0220, -0.0054,  0.0112, -0.0238,  0.0264,
         0.0163, -0.0170,  0.0199, -0.0140,  0.0036,  0.0065, -0.0104, -0.0175,
         0.0037, -0.0470, -0.1773, -0.0371,  0.1424, -0.0644,  0.0522, -0.0318,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1684, -0.0649, -0.0239, -0.0762,  0.0587,  0.0117,  0.0299, -0.0212,
         0.0143, -0.1027, -0.0131, -0.0469, -0.0411, -0.0407, -0.0578,  0.0201,
         0.0243, -0.0389,  0.0021,  0.0332,  0.0089,  0.0250, -0.0565,  0.0194,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2026, -0.0915, -0.0475, -0.0024, -0.0057,  0.0112, -0.0062, -0.0097,
        -0.0010,  0.0311, -0.0016,  0.0003, -0.0148,  0.0006, -0.0024, -0.0330,
        -0.0098,  0.0314,  0.0202,  0.0198, -0.0005, -0.0110,  0.0116, -0.1113,
        -0.0013,  0.0215,  0.0412, -0.0382,  0.0562,  0.0027, -0.0310, -0.0087,
        -0.0146,  0.1074,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1616, -0.0715, -0.0075, -0.0112, -0.0108,  0.0435, -0.0277, -0.0056,
         0.0096, -0.0266,  0.0425, -0.0006, -0.0200, -0.1522,  0.0275, -0.1156,
        -0.0245, -0.0056, -0.0023, -0.0105,  0.0374, -0.0232, -0.0155, -0.0362,
         0.1108,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1128, -0.0955, -0.1173, -0.0589, -0.0247,  0.0071, -0.0447, -0.0227,
        -0.0140, -0.3080, -0.0425,  0.0049, -0.0213, -0.0206,  0.0178,  0.0872,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0297, -0.0679, -0.0306, -0.0728, -0.0173, -0.0646, -0.2483, -0.0358,
         0.0037, -0.0346,  0.0224,  0.0892, -0.0413, -0.0431, -0.0211,  0.0649,
         0.0003, -0.0263, -0.0274, -0.0344, -0.0207, -0.0033,  0.0004,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1577, -0.2011, -0.0015,  0.0028, -0.0833,  0.0097, -0.0308, -0.1488,
        -0.0671, -0.0515,  0.0369,  0.0053, -0.1112, -0.0923,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.1884, -0.0591,  0.0018, -0.0353, -0.0343,  0.0677, -0.0427,  0.0283,
        -0.0108,  0.0103, -0.0622, -0.0354, -0.0361,  0.0003, -0.0303,  0.0185,
         0.0232, -0.0876, -0.0555, -0.0135, -0.0230, -0.0282,  0.0081,  0.0996,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0975, -0.0499, -0.0887, -0.1029, -0.1142, -0.1288, -0.1065, -0.0887,
        -0.1034,  0.0538, -0.0657,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0460, -0.0363,  0.0347,  0.0912, -0.0297, -0.0831, -0.1168,  0.0455,
        -0.0877,  0.0265, -0.0075,  0.0424,  0.0065,  0.0527, -0.0532,  0.0064,
        -0.0253, -0.1281, -0.0043,  0.0750, -0.0013,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0213, -0.0494, -0.0763,  0.1349,  0.0616, -0.0284, -0.1287, -0.1949,
        -0.0015, -0.0326, -0.0096, -0.1328,  0.0915,  0.0364,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2215,  0.0696,  0.0425, -0.0205,  0.0310,  0.0352, -0.0130,  0.0149,
         0.0530,  0.0021,  0.0512,  0.1898,  0.0930,  0.0254,  0.0143,  0.0080,
        -0.0041,  0.0492,  0.0272,  0.0078,  0.0267,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0653, -0.0544, -0.0353, -0.1022, -0.0981, -0.1432, -0.0442, -0.1638,
        -0.0003, -0.0989, -0.0789,  0.0109, -0.1044,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.6646e-03, -1.4567e-02,  8.5950e-04,  1.0191e-02, -1.3728e-02,
        -7.8042e-03, -6.6412e-03, -4.7576e-02,  9.0323e-03, -1.6202e-02,
        -2.6715e-02, -1.3246e-02,  2.4751e-02, -9.7940e-03,  1.3927e-03,
        -1.2909e-02,  4.7606e-05,  3.2993e-02, -2.0166e-02, -2.7568e-02,
        -9.4873e-03, -2.1547e-02, -8.8475e-02, -9.5683e-03, -9.4970e-02,
         3.4987e-02, -2.6512e-02, -8.6297e-02, -1.8180e-02, -3.0910e-02,
        -2.1380e-02, -2.8892e-02,  5.4345e-02,  2.3098e-02, -2.2540e-02,
        -3.1800e-02, -1.8174e-02,  8.4622e-04, -7.3482e-04,  1.8467e-03,
         3.8081e-03,  1.2132e-02,  8.2250e-03, -4.4226e-02,  8.1696e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0123, -0.0185, -0.0164, -0.0153, -0.0277,  0.0152, -0.0276, -0.0092,
        -0.0242, -0.0154, -0.0268, -0.0598, -0.0010,  0.1536, -0.0689, -0.0178,
        -0.1186,  0.0262, -0.0079, -0.0057, -0.0203, -0.0516,  0.0020, -0.0351,
         0.0074,  0.0009, -0.0443,  0.0077, -0.0210,  0.0176, -0.0292, -0.0056,
         0.0422,  0.0470,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1858, -0.1690,  0.0082, -0.0252, -0.0318, -0.0272, -0.0090, -0.0315,
        -0.0642, -0.0288, -0.0204, -0.0342,  0.0217, -0.0197,  0.0032, -0.0043,
        -0.0289,  0.0093, -0.0125, -0.0337,  0.0046, -0.0338, -0.0253, -0.0586,
        -0.1053,  0.0035,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0358, -0.2810,  0.0824, -0.0076,  0.0140, -0.0662, -0.0898, -0.0361,
         0.0151,  0.0269, -0.0422, -0.0105, -0.0127, -0.0293, -0.0686,  0.0060,
        -0.0835,  0.0158, -0.0099, -0.0099, -0.0061,  0.0476, -0.0030,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1111, -0.0522, -0.0657, -0.0347, -0.0090,  0.0014,  0.0071, -0.0247,
        -0.0301,  0.0133, -0.0175,  0.0054, -0.0157,  0.0211, -0.1140,  0.0109,
        -0.0058, -0.0288, -0.0185, -0.0119, -0.0130,  0.0018,  0.0217,  0.0203,
        -0.0163, -0.0184, -0.0186, -0.0630,  0.0046, -0.0203,  0.0039, -0.0015,
         0.0116, -0.0083, -0.0157, -0.0127, -0.0118, -0.0200, -0.0947,  0.0226,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0159, -0.0775, -0.0312, -0.0123, -0.0287, -0.0171,  0.0180, -0.0589,
        -0.0619, -0.0282, -0.0190, -0.0089, -0.0143, -0.0374, -0.0140, -0.0286,
        -0.0090, -0.0040,  0.0649, -0.0052,  0.0132,  0.0092, -0.0047,  0.0265,
        -0.0103, -0.0174,  0.0049, -0.0233,  0.0014,  0.0194, -0.0776, -0.0322,
         0.1121, -0.0925,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0604, -0.0189,  0.0013, -0.0153, -0.0060, -0.0096, -0.0212,  0.0023,
        -0.0054,  0.0323, -0.0131, -0.0202,  0.0400, -0.0140, -0.0953, -0.0560,
         0.0151, -0.0129, -0.0278, -0.0023, -0.0057, -0.0095,  0.0109,  0.0096,
        -0.0185, -0.1118, -0.0492, -0.0134, -0.0182,  0.0129, -0.0300, -0.0017,
        -0.0045,  0.0089, -0.0034,  0.0150, -0.0220, -0.0109, -0.0008, -0.0123,
        -0.0078, -0.0203, -0.0120, -0.0196, -0.0129, -0.0018, -0.0234, -0.0057,
        -0.0169, -0.0122, -0.0087,  0.0201], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0907, -0.0995, -0.0189,  0.0302, -0.0824, -0.0099, -0.0180, -0.0228,
         0.0042, -0.0454, -0.0253, -0.0110, -0.0591, -0.0574, -0.0315, -0.0357,
        -0.0511,  0.0624, -0.0164,  0.0044, -0.0443,  0.0024, -0.0288, -0.0226,
        -0.0218, -0.0091,  0.0324, -0.0370, -0.0088, -0.0109, -0.0054,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0358, -0.0133,  0.0425,  0.0058, -0.0489, -0.0088,  0.0042, -0.0002,
        -0.0294,  0.0063, -0.0087, -0.0582, -0.0427,  0.0273,  0.0105, -0.0219,
         0.1550,  0.0359, -0.0392, -0.0146, -0.0519, -0.0528, -0.0419, -0.0811,
         0.1631,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0193, -0.1375, -0.1115,  0.0016, -0.1062, -0.0977, -0.2487, -0.0134,
        -0.0806,  0.0412,  0.0143, -0.1282,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0470, -0.1578, -0.0161, -0.0621,  0.0177,  0.0111,  0.0324, -0.0049,
        -0.0017, -0.0406,  0.0147,  0.0359, -0.0176, -0.1139, -0.0224, -0.0458,
        -0.0344, -0.0068,  0.0059,  0.0275, -0.0200,  0.0111,  0.0038,  0.0494,
         0.0219, -0.1237, -0.0118,  0.0028, -0.0390,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1231, -0.1404,  0.0483,  0.0270,  0.0667, -0.0335, -0.1666, -0.1206,
         0.1318, -0.0743, -0.0413, -0.0262,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0473, -0.2281,  0.0531, -0.0420, -0.0623, -0.0987, -0.1895, -0.0268,
        -0.0596, -0.0723, -0.1091,  0.0112,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2430,  0.0230, -0.0759,  0.0369, -0.1316,  0.0158, -0.0355, -0.0336,
        -0.0560, -0.0157, -0.0227, -0.0266, -0.0086,  0.0279, -0.0143, -0.0610,
        -0.0774, -0.0181, -0.0690,  0.0075,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0065, -0.1633, -0.0090, -0.0300, -0.1243, -0.0417, -0.0294,  0.0286,
        -0.0373,  0.0035,  0.0019,  0.0047, -0.0273, -0.0018,  0.0130,  0.0135,
        -0.0117,  0.0026, -0.0008, -0.0097, -0.0254, -0.0003,  0.0101,  0.0480,
         0.0049,  0.0373, -0.0049, -0.0187, -0.0075, -0.0211,  0.0719, -0.0079,
        -0.0212, -0.0559, -0.0345,  0.0616,  0.0083,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0504, -0.3739, -0.2483, -0.0131,  0.0322,  0.0165, -0.0227,  0.1354,
        -0.1076,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0190, -0.0679,  0.0798, -0.0485, -0.0304, -0.0159, -0.0519, -0.0422,
        -0.0090,  0.0006,  0.0288,  0.0176, -0.0621, -0.0470, -0.0215,  0.0243,
        -0.0033,  0.0003, -0.0462, -0.0258, -0.0431,  0.0124,  0.0056, -0.0213,
         0.0093, -0.1764,  0.0897,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1500, -0.1055, -0.0069,  0.0021, -0.0098, -0.0424, -0.0134, -0.0047,
         0.0039,  0.0061, -0.0045, -0.0183, -0.0128,  0.0480,  0.0109,  0.0138,
        -0.0548,  0.0772,  0.0080,  0.0113, -0.0105, -0.0223,  0.0062,  0.0361,
         0.0144,  0.0161, -0.0265, -0.0399, -0.0811, -0.0107, -0.0109, -0.0021,
         0.0711, -0.0475,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.1802, -0.0091,  0.0034,  0.0073, -0.0334, -0.0050,  0.0076, -0.0163,
        -0.0352, -0.0377, -0.0245, -0.0754, -0.0320,  0.0052,  0.0031,  0.0127,
         0.0110,  0.0169, -0.0113,  0.0069,  0.0268, -0.0004, -0.0272,  0.0107,
        -0.0175, -0.0213, -0.0087,  0.0039, -0.0198,  0.0180,  0.0113, -0.0288,
        -0.0716,  0.0312,  0.0016, -0.0056, -0.0323, -0.0178,  0.0072,  0.0055,
        -0.0095, -0.0576, -0.0182, -0.0062, -0.0072,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0779, -0.0012,  0.0133, -0.0451, -0.1495, -0.0618, -0.2488, -0.0074,
        -0.0929, -0.1092,  0.0322, -0.0029, -0.0726, -0.0331, -0.0164, -0.0356,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1116,  0.0670, -0.0310, -0.0576, -0.0242,  0.0338,  0.0512,  0.0378,
        -0.0477, -0.0684,  0.0208, -0.0296, -0.0323, -0.0186, -0.0392,  0.0266,
        -0.0387,  0.0028,  0.0094, -0.0123, -0.0365, -0.0214, -0.0233,  0.0104,
        -0.0388,  0.0037, -0.1051,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.9849e-02, -1.5984e-01, -2.9310e-02, -1.1155e-02, -1.6757e-02,
         6.1045e-03, -1.8512e-02, -2.1585e-02, -1.7383e-02, -2.5870e-02,
        -2.5276e-02, -7.2865e-03, -2.1427e-02, -6.8090e-03, -1.7893e-02,
         4.2513e-03,  3.2067e-03, -2.3601e-05, -9.0701e-03, -1.8276e-02,
         1.4887e-03, -2.4899e-02, -3.3558e-02, -5.9745e-02, -1.3237e-03,
        -4.4860e-03, -1.3339e-02, -3.5285e-02,  7.4295e-04, -1.7333e-02,
        -8.0168e-03, -1.1396e-04, -2.0281e-02, -8.3449e-03, -1.9126e-02,
        -1.7244e-02, -1.5226e-02, -2.1319e-02, -1.3394e-02,  1.5933e-02,
        -3.6827e-02, -9.2550e-03, -2.9039e-03, -1.2444e-03, -3.9121e-02,
         8.9558e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0337, -0.0078, -0.0280, -0.0620, -0.0178, -0.0245, -0.0188, -0.0356,
         0.0267, -0.0220, -0.0156, -0.0159, -0.0133, -0.0091, -0.0479,  0.0011,
         0.0072, -0.0030, -0.0526, -0.1730, -0.0530, -0.0334,  0.0099, -0.0255,
        -0.0400, -0.0214, -0.0555, -0.0901,  0.0025,  0.0532,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2315,  0.0231, -0.1350,  0.1215,  0.0587, -0.0300,  0.0760,  0.0174,
         0.0375,  0.1797,  0.0896,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0678, -0.1085, -0.0243, -0.0022, -0.0472, -0.0022, -0.0473, -0.0974,
        -0.0364, -0.1064,  0.0178, -0.0300, -0.0239,  0.0229, -0.0196, -0.0944,
        -0.0067,  0.0148, -0.0553, -0.1751,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0102, -0.2765, -0.1020, -0.1586,  0.0240, -0.1353, -0.0876, -0.0586,
        -0.0857, -0.0362,  0.0252,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0310, -0.2360, -0.0045, -0.1120, -0.0860,  0.0439,  0.0013, -0.0329,
        -0.0599, -0.0497,  0.0016,  0.0338, -0.0098,  0.0233, -0.0335, -0.0812,
        -0.0340, -0.0067, -0.0737, -0.0144, -0.0309,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0264,  0.0870,  0.0440, -0.0465,  0.0605,  0.1841, -0.1256,  0.0975,
         0.0288, -0.0799,  0.0267,  0.0466, -0.0094,  0.0151, -0.0136, -0.0551,
         0.0483, -0.0048,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0754, -0.0459, -0.1363, -0.0422,  0.0101, -0.0229,  0.0126,  0.0071,
        -0.0169, -0.0023, -0.0063,  0.0041, -0.0347, -0.0467, -0.0634,  0.0075,
        -0.0163, -0.0322, -0.0365,  0.0269, -0.0219, -0.0217, -0.0736, -0.0008,
        -0.0335, -0.0027, -0.0362, -0.0134, -0.0042, -0.0832, -0.0185, -0.0043,
        -0.0013,  0.0063, -0.0022, -0.0091, -0.0068, -0.0011,  0.0129,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.7673e-02, -2.5668e-01, -1.5713e-02, -2.2575e-02,  9.3176e-05,
        -1.1377e-02,  7.6608e-03, -1.8687e-03,  5.8612e-02,  1.9810e-02,
         1.3638e-02, -2.0568e-02, -4.1611e-02, -3.0438e-02,  1.2899e-02,
        -1.5123e-02, -3.8140e-02,  7.7787e-03, -7.5995e-03, -9.3316e-03,
        -4.9965e-02, -5.0894e-03, -1.9812e-03,  3.6704e-03,  3.2795e-03,
        -1.7984e-03, -1.5310e-02,  1.9717e-02, -4.9417e-02, -7.2410e-02,
         8.1226e-03,  1.3787e-02, -8.8329e-02, -7.9290e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.1623, -0.2218, -0.0244, -0.1431, -0.1186, -0.0061,  0.0520, -0.2483,
        -0.0233,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0312,  0.0913, -0.0086,  0.0135,  0.0564,  0.0614, -0.0002, -0.0096,
         0.0255,  0.0362,  0.0330,  0.0094,  0.0057,  0.0120, -0.0129,  0.0280,
        -0.0005,  0.1748, -0.0093, -0.0432, -0.0045, -0.0064, -0.0191, -0.0083,
         0.0076,  0.0110,  0.0054, -0.0059, -0.0287, -0.0016,  0.0287,  0.0377,
         0.0009, -0.0135,  0.0080, -0.0009,  0.0037,  0.0413, -0.0142, -0.0895,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.1727e-03, -3.3411e-02, -8.7972e-02, -6.4147e-03, -3.5889e-02,
        -2.5339e-02,  1.3178e-02, -1.8055e-02, -2.1446e-02, -7.1427e-02,
         6.5499e-05, -7.3175e-03,  5.2329e-03, -5.3601e-03,  3.2789e-03,
         1.0088e-02,  2.2669e-02,  7.7366e-03, -2.1876e-02, -7.9103e-03,
         8.3238e-03, -3.3119e-02, -4.8710e-03, -6.1889e-03, -5.9402e-02,
        -1.0944e-02, -3.2088e-02,  2.4124e-02, -6.8909e-02, -4.1197e-02,
         1.0250e-02, -4.1955e-02, -1.1691e-02, -3.4456e-03, -1.1771e-02,
        -1.2655e-02, -1.7670e-02, -4.4008e-02,  1.1179e-02,  8.1567e-04,
        -1.7722e-03,  2.9283e-03, -3.0403e-02, -1.4584e-02, -5.4642e-03,
         3.9513e-03,  5.3523e-03, -9.5473e-03,  4.4103e-03, -1.8006e-02,
        -3.4894e-03,  3.2645e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1227, -0.1120, -0.0592, -0.0419, -0.0640, -0.0680, -0.0543, -0.0107,
        -0.0505,  0.0042, -0.0157, -0.0233, -0.0183, -0.0482, -0.0471, -0.0576,
        -0.0128, -0.0405, -0.0618,  0.0242, -0.0146, -0.0483,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1074,  0.4231, -0.0352,  0.0089, -0.0360,  0.0024,  0.0564,  0.0112,
         0.1165,  0.0121, -0.0069,  0.0044, -0.0025,  0.0217, -0.0023,  0.0140,
         0.0535, -0.0232, -0.0162,  0.0259,  0.0202,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.0260e-02, -1.8508e-02,  1.6648e-02, -3.6492e-02, -4.6066e-02,
         2.9719e-03,  1.2460e-02,  1.8094e-02, -2.7171e-02, -5.5201e-03,
        -1.6016e-02, -3.4748e-03, -7.5258e-04,  1.0183e-02, -4.6994e-03,
         5.6468e-03,  1.5193e-02, -9.1530e-03, -2.6758e-02, -1.5066e-02,
         1.2461e-02,  3.6950e-05, -1.1314e-02, -8.3239e-03, -3.1039e-03,
        -1.4736e-04, -4.2271e-02, -1.6882e-02,  2.6200e-03,  8.5740e-03,
        -5.2974e-02, -2.3033e-02, -8.7260e-03, -1.5099e-02, -2.4237e-02,
        -1.0647e-02, -4.2293e-03,  1.7300e-02, -8.4929e-03, -4.0950e-02,
        -9.1642e-02, -2.8439e-04, -5.1378e-02,  8.0369e-03, -2.0641e-02,
         2.9406e-03, -2.4511e-02, -2.4976e-02,  9.6607e-03,  3.4051e-02,
        -2.0455e-02,  8.0919e-03, -8.2342e-03,  5.3833e-02, -8.7083e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0209, -0.0419, -0.0079, -0.0081, -0.0154, -0.0242,  0.0092, -0.0184,
         0.0587,  0.0023,  0.0256,  0.0042, -0.0371,  0.0045,  0.0039,  0.0003,
         0.0454, -0.0061, -0.0102,  0.0098, -0.0347, -0.0133, -0.0132, -0.0039,
         0.0349,  0.0062, -0.0381, -0.0718, -0.0266, -0.0178,  0.0064,  0.0051,
        -0.0193, -0.0879, -0.0012, -0.0135, -0.0275, -0.0258, -0.0025, -0.0151,
        -0.0101,  0.0075, -0.0422, -0.0320, -0.0892,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1602, -0.0946, -0.0674, -0.0824, -0.0013, -0.0137, -0.0453,  0.0078,
        -0.0545, -0.0642, -0.0182, -0.0142,  0.0327, -0.0362, -0.0240,  0.0167,
        -0.0136, -0.0507,  0.0257, -0.0165, -0.0816, -0.0509,  0.0259, -0.0016,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1160, -0.0447,  0.0123, -0.0025, -0.0212, -0.0373,  0.0138, -0.0404,
         0.0336, -0.0014, -0.0040, -0.0038,  0.0091, -0.0800, -0.0367, -0.0034,
        -0.0337, -0.0143, -0.0031, -0.0053, -0.0255, -0.0036, -0.0193, -0.0858,
        -0.0210, -0.0169, -0.1088, -0.0318, -0.0157, -0.0790, -0.0134,  0.0033,
        -0.0593,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.9930e-02, -2.3337e-01, -4.7689e-02, -8.3953e-02,  3.4865e-02,
        -5.8275e-02,  1.2528e-03,  6.1368e-02, -1.5932e-01, -5.3574e-02,
         1.6205e-04, -4.0806e-02, -1.4149e-02, -1.2915e-02, -7.3568e-02,
         3.5370e-03, -7.7911e-03,  4.6746e-02,  3.6731e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1472,  0.1009,  0.0331, -0.0563, -0.0553,  0.0415, -0.0757, -0.0135,
        -0.0745,  0.0081, -0.0606, -0.0135, -0.0430, -0.0063, -0.0012,  0.0220,
        -0.0020, -0.0269,  0.0072,  0.0010, -0.0158, -0.0716, -0.0093,  0.0235,
         0.0899,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.3065, -0.0219, -0.0135,  0.0125, -0.0649,  0.0147, -0.2019,  0.1058,
         0.0365, -0.0302, -0.0465,  0.0120, -0.0122, -0.0123, -0.0331, -0.0388,
         0.0369,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.2017,  0.0204, -0.0122,  0.0094, -0.0203, -0.0414,  0.0045, -0.0452,
        -0.0633, -0.0642, -0.0114, -0.0246,  0.0032, -0.0183, -0.0189, -0.0238,
        -0.0267, -0.0179,  0.0111, -0.0192, -0.0093, -0.0225, -0.1427,  0.0012,
        -0.0117, -0.0315,  0.0158,  0.0317, -0.0038,  0.0289,  0.0232, -0.0201,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0236, -0.1094, -0.0254, -0.0181, -0.0387,  0.0019, -0.0204, -0.0821,
        -0.0193, -0.0226, -0.0075, -0.0266,  0.0384, -0.0341,  0.0184, -0.0255,
        -0.0356, -0.0020, -0.0118, -0.0211, -0.0097, -0.0199, -0.0303, -0.0861,
        -0.0524, -0.0241, -0.0587, -0.1363,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0514, -0.0914,  0.0373,  0.0788,  0.0215, -0.0315, -0.1346, -0.0704,
         0.0318,  0.0065, -0.0111, -0.0180,  0.0236, -0.0161, -0.0142, -0.0333,
         0.0019, -0.0702,  0.0049, -0.0327, -0.0155,  0.0285,  0.0029, -0.0152,
        -0.0330, -0.0271,  0.0098, -0.0030, -0.0837,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0192, -0.0633, -0.0080, -0.0251, -0.1320, -0.0544, -0.0080,  0.0028,
        -0.0651,  0.0157, -0.0102, -0.0092, -0.0108, -0.1038, -0.1105, -0.0154,
        -0.0263, -0.0817, -0.0272,  0.0341,  0.0004, -0.0022, -0.0052,  0.0008,
        -0.0141, -0.0167, -0.0157, -0.0161,  0.0089, -0.0011, -0.0328, -0.0246,
         0.0387,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0518, -0.1053, -0.0505, -0.0908, -0.0399, -0.0598, -0.0006, -0.0615,
        -0.0183, -0.0488, -0.0248, -0.0137, -0.0110, -0.0196, -0.0131, -0.0189,
        -0.0435,  0.0280, -0.0753,  0.0287, -0.0775, -0.0341, -0.0161,  0.0685,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0215, -0.0661, -0.0815, -0.0270,  0.0075, -0.0706, -0.1018, -0.0555,
         0.0065, -0.0314,  0.0162, -0.0436, -0.0491, -0.0482, -0.0173,  0.0244,
         0.0002, -0.0251, -0.0293, -0.1734, -0.0536,  0.0171, -0.0331,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1409, -0.0553,  0.0017, -0.0097, -0.0275, -0.0252, -0.0196, -0.0098,
         0.0040, -0.0201, -0.0006, -0.0363, -0.0320, -0.0065, -0.0084, -0.1428,
        -0.0364, -0.0178, -0.0032,  0.0126, -0.0270, -0.0282,  0.0197,  0.0093,
        -0.0100, -0.0197, -0.0163, -0.0060, -0.0024, -0.0300, -0.0474, -0.0110,
        -0.0088,  0.0026,  0.0045, -0.0119, -0.0082,  0.1022,  0.0244,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0268, -0.1862, -0.0181,  0.0173, -0.0671, -0.0045, -0.0067,  0.0129,
        -0.0544, -0.0109, -0.0242,  0.0091, -0.0078, -0.0338,  0.0080, -0.0184,
        -0.0049,  0.0057,  0.0186,  0.0023, -0.0234, -0.0097, -0.0208,  0.0037,
        -0.0144, -0.0160, -0.0143, -0.0031, -0.0232, -0.0060, -0.0316, -0.0164,
         0.0239, -0.0005,  0.0095, -0.0300, -0.0128, -0.0114, -0.0140, -0.0068,
        -0.0224, -0.0677, -0.0806,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0433, -0.1845, -0.0012,  0.0098, -0.0255, -0.0015, -0.0621, -0.0107,
         0.0063,  0.0163, -0.0154,  0.0047, -0.0075,  0.0144, -0.0034, -0.0081,
        -0.0230, -0.0403, -0.0060,  0.0034,  0.0190,  0.0113, -0.0173, -0.0132,
        -0.0116, -0.0094,  0.0046, -0.0059,  0.0007, -0.0123, -0.0138,  0.0072,
        -0.0587, -0.0088,  0.0186,  0.0101,  0.0136,  0.0235, -0.0184, -0.0082,
        -0.0054, -0.0274, -0.0132,  0.0171,  0.0023, -0.0123, -0.0084, -0.0059,
         0.0101, -0.0162,  0.0179,  0.0256, -0.0090, -0.0256, -0.0002,  0.0192,
         0.0110], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2031, -0.1237, -0.0381, -0.0787, -0.0180,  0.0323, -0.0449, -0.1400,
         0.0471, -0.0845, -0.0441, -0.0563, -0.0218, -0.0673,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2112, -0.0712, -0.0654,  0.0664,  0.0035, -0.0121, -0.0453, -0.0144,
        -0.0102,  0.0065, -0.0138, -0.0817, -0.0158, -0.0516, -0.0398,  0.0198,
         0.0025, -0.0221,  0.0382, -0.0032, -0.0862, -0.0342, -0.0054,  0.0050,
        -0.0073, -0.0055, -0.0093, -0.0064,  0.0299,  0.0162,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0330,  0.1813, -0.3057,  0.1656,  0.0847, -0.1498,  0.0509, -0.0289,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.1577, -0.0615, -0.0053,  0.0007, -0.0217,  0.0238, -0.0005,  0.0041,
         0.0011, -0.0129, -0.0265, -0.0151, -0.0401, -0.1048,  0.0006,  0.0136,
         0.0014, -0.0083, -0.0142, -0.0298,  0.0157, -0.0285, -0.0340, -0.0212,
        -0.0221, -0.0294, -0.0335,  0.0235, -0.0498, -0.0926,  0.0137,  0.0775,
         0.0150,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.3716e-01, -3.0557e-02, -4.5539e-02, -1.6087e-02, -2.1843e-03,
        -1.1328e-02,  1.4916e-03, -6.9246e-03, -1.3906e-03, -1.1828e-02,
        -1.0861e-02,  1.4247e-02, -1.0943e-02, -1.4125e-02,  1.8327e-03,
         4.5693e-02, -6.1010e-04, -3.1710e-03, -1.2500e-02, -6.4021e-03,
        -3.1686e-03, -1.1944e-02,  7.6046e-03, -5.0416e-03, -4.0888e-03,
         6.6008e-03, -2.2828e-03, -1.7185e-02,  1.1293e-03,  7.9554e-03,
        -8.8310e-04, -2.7578e-03, -1.4029e-02, -6.3244e-02,  1.8530e-02,
        -8.8787e-03, -1.3860e-02, -1.1934e-04,  1.0454e-03,  3.9315e-03,
         3.2459e-03, -9.2426e-04,  1.0126e-03, -1.5760e-02,  8.9385e-03,
         2.4117e-02, -1.7687e-02,  1.2444e-02, -1.4852e-02, -1.1466e-02,
        -7.3936e-03, -1.2425e-02, -8.9786e-03, -2.1089e-04, -2.8096e-03,
        -5.0441e-03, -7.7212e-03,  1.4076e-02, -2.1580e-02, -1.8508e-02,
        -2.1949e-02, -2.0753e-03, -1.1003e-01, -4.9184e-03, -3.9072e-03,
        -3.2123e-03,  1.1676e-03, -1.5905e-02, -1.8465e-02,  2.3041e-02,
         6.9801e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0191, -0.0401, -0.0028,  0.0021, -0.0483,  0.0055, -0.1244, -0.0378,
        -0.0106,  0.0185,  0.0285, -0.0024, -0.0119, -0.0125, -0.0262, -0.0241,
        -0.0071,  0.0173, -0.0373, -0.0548,  0.0048, -0.0225,  0.0123, -0.0399,
         0.0235, -0.0251, -0.0395,  0.0208,  0.0150,  0.0124, -0.0021, -0.0323,
        -0.0149, -0.0305, -0.0349, -0.0302, -0.0085, -0.0197, -0.0064,  0.0469,
         0.0249,  0.0020,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0141, -0.2181,  0.0108,  0.0037, -0.0058, -0.0166,  0.0221,  0.0334,
        -0.0263, -0.0850, -0.0419, -0.0246, -0.0237,  0.0089,  0.0110,  0.0057,
         0.0212, -0.0210,  0.0095,  0.0473, -0.0082, -0.0104,  0.0064, -0.0032,
         0.0175,  0.0040, -0.0048, -0.0132, -0.0320,  0.0341,  0.0062, -0.0088,
         0.0392,  0.0099, -0.0171,  0.0734,  0.0610,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0201, -0.1762,  0.1252, -0.0990, -0.1244,  0.0572,  0.0347, -0.0036,
         0.0665, -0.0409, -0.0058, -0.0433, -0.0225, -0.0963,  0.0843,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0058,  0.1453, -0.0083, -0.0312,  0.0073,  0.0308, -0.0208,  0.0085,
         0.1491,  0.0173, -0.0562, -0.0359,  0.0566,  0.0100, -0.0979,  0.0582,
         0.1197,  0.0545,  0.0424,  0.0442,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0016, -0.0558, -0.0321,  0.0033, -0.0109, -0.0542, -0.0099, -0.0175,
        -0.0018, -0.0082, -0.0053, -0.0154, -0.0209, -0.0143, -0.0212, -0.0147,
         0.0072, -0.0099,  0.0184, -0.0137,  0.0265,  0.0177,  0.0063,  0.0167,
        -0.0216, -0.0258, -0.0840, -0.0102, -0.0006, -0.0183, -0.0158, -0.0474,
        -0.0094, -0.0220, -0.0194, -0.0154, -0.0006, -0.0174, -0.0067,  0.0011,
         0.0321,  0.1331, -0.1156,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0391, -0.0590,  0.0644, -0.0576, -0.0156, -0.0230, -0.0245, -0.0864,
         0.0234, -0.0704, -0.0201,  0.0014, -0.0133, -0.0354, -0.0791, -0.0007,
        -0.0117, -0.0086,  0.0074, -0.0111, -0.0099, -0.0101,  0.0111, -0.0392,
        -0.0036, -0.0225, -0.0413,  0.0197,  0.0030, -0.0163, -0.0047, -0.0229,
         0.0017, -0.0152, -0.0243, -0.0093, -0.0116, -0.0815,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0193, -0.1749, -0.0942, -0.1057, -0.0125, -0.0730, -0.0879,  0.0572,
        -0.0166, -0.0225, -0.0109, -0.0547, -0.0366, -0.0174, -0.0006,  0.0120,
         0.0016, -0.0566, -0.0055, -0.0259,  0.0950,  0.0195,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0286, -0.0583,  0.0096,  0.0467,  0.0121,  0.0131, -0.0037, -0.0145,
        -0.0365, -0.0430, -0.0082, -0.0038, -0.0121,  0.0382, -0.0142, -0.0185,
        -0.0916, -0.0045, -0.0142, -0.0417, -0.0076, -0.0254, -0.0201, -0.0346,
        -0.0062, -0.0037, -0.0148, -0.0185, -0.0264, -0.0117, -0.0479,  0.0044,
        -0.0163, -0.0604, -0.0132, -0.1033,  0.0086, -0.0342, -0.0108,  0.0188,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0919, -0.0597,  0.0271, -0.0644, -0.0131,  0.0200,  0.0290, -0.0026,
        -0.0189, -0.0252,  0.0300,  0.0139,  0.0085, -0.0274, -0.0448, -0.0195,
        -0.0170,  0.0092, -0.1007, -0.0213, -0.0170, -0.0010, -0.0483,  0.0051,
         0.0630, -0.0973,  0.0166, -0.0574, -0.0147, -0.0140, -0.0193,  0.0021,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1355, -0.1998, -0.0049,  0.0116,  0.0201, -0.0282, -0.0475, -0.0363,
        -0.0676,  0.0146,  0.0124, -0.0104, -0.0064,  0.0276, -0.0045, -0.0244,
        -0.0132,  0.0043,  0.0140, -0.0104,  0.0163,  0.0250,  0.0480, -0.0324,
         0.0069,  0.0272, -0.0050, -0.0152,  0.0005, -0.0194,  0.0085, -0.0064,
         0.0254, -0.0401, -0.0302,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.1440, -0.0795, -0.0245, -0.1440,  0.0131,  0.0319, -0.0622, -0.0753,
         0.0691, -0.0559, -0.0038, -0.0066, -0.0242, -0.0574, -0.0190, -0.0331,
        -0.0421, -0.0301, -0.0062,  0.0067,  0.0712,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1181,  0.2847,  0.0497,  0.0840,  0.0565,  0.0601,  0.0114,  0.0405,
         0.1223, -0.0431, -0.0248, -0.0213, -0.0071,  0.0009,  0.0227, -0.0100,
         0.0131, -0.0228, -0.0068,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0407, -0.2162, -0.1170,  0.0238, -0.0159, -0.0539, -0.0031,  0.0037,
         0.0231, -0.0297, -0.0998, -0.0106, -0.0027, -0.0992, -0.0227, -0.0236,
         0.0188, -0.0211, -0.0348, -0.1398,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1111, -0.2149, -0.0135, -0.0439, -0.0234, -0.0196,  0.0614, -0.0454,
        -0.0502, -0.0232, -0.0170, -0.0238,  0.0167, -0.0313, -0.0972, -0.0107,
        -0.0632,  0.0079, -0.0190, -0.0528,  0.0119,  0.0004,  0.0415,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1259, -0.0674, -0.0220, -0.0164, -0.0192, -0.0479, -0.0391,  0.0005,
        -0.0571, -0.0240, -0.0469, -0.0495, -0.0249, -0.0784, -0.0100,  0.0733,
         0.0063,  0.0341, -0.0187,  0.0117, -0.0214, -0.0125, -0.0551, -0.0343,
        -0.1035,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0072, -0.1766, -0.0204,  0.0019, -0.0439, -0.0149,  0.0437, -0.0237,
         0.0389,  0.0620, -0.0496, -0.0011, -0.0150, -0.0283, -0.0009,  0.0416,
        -0.0469,  0.0046, -0.0055, -0.0306, -0.0116, -0.0278, -0.0170,  0.0036,
         0.0068, -0.0415, -0.0130,  0.0047,  0.0173,  0.0196,  0.0298, -0.0028,
         0.0204, -0.0406,  0.0864,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1716,  0.0206, -0.0266, -0.0309, -0.0321, -0.0189, -0.0514,  0.0273,
        -0.0093,  0.0016,  0.0473,  0.0058,  0.0552, -0.0163, -0.0478, -0.1612,
        -0.0177, -0.0817, -0.0814, -0.0362,  0.0353, -0.0237,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0974,  0.0011, -0.0011,  0.0358,  0.0381, -0.0633, -0.0087,  0.0343,
        -0.0322, -0.0251, -0.0915,  0.0037,  0.0020, -0.0150, -0.0074,  0.0171,
        -0.0727,  0.0102,  0.2642,  0.0590, -0.1202,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1213, -0.1463, -0.0374,  0.0478,  0.0290,  0.0209, -0.0093, -0.1147,
        -0.0495,  0.0368,  0.0263,  0.0206, -0.0819, -0.0639, -0.0024, -0.0255,
        -0.0350,  0.0461, -0.0852,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1443, -0.0137, -0.0697, -0.0442,  0.0076, -0.0535,  0.0180,  0.0098,
         0.0022, -0.0006, -0.0036, -0.0105, -0.0120, -0.0225, -0.0038,  0.0012,
         0.0119, -0.0305,  0.0071, -0.0177,  0.0061,  0.0199, -0.0049, -0.0060,
        -0.0074, -0.0241, -0.0012, -0.0062, -0.0306, -0.1337, -0.0233,  0.0005,
        -0.0233, -0.0545,  0.0049, -0.0158,  0.0152, -0.0010,  0.0071, -0.0043,
         0.0129, -0.0055,  0.0073,  0.0077, -0.0012,  0.0496,  0.0417],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.2559e-02, -1.0020e-01, -6.3760e-02, -8.2591e-03,  7.7196e-03,
        -4.8561e-02,  1.7167e-02, -1.3731e-05, -1.4851e-02, -7.2355e-03,
         1.6588e-02,  1.2343e-02,  3.4798e-02, -2.8962e-02,  3.5577e-02,
        -7.4332e-02, -2.9292e-02, -4.6695e-02,  5.2969e-02, -2.0262e-03,
         1.2014e-03, -2.5065e-02, -1.3752e-02, -1.5180e-01, -1.9021e-02,
        -6.0490e-04,  6.8653e-03, -5.1468e-03,  1.3621e-02,  1.2562e-02,
         3.4089e-04,  3.4928e-02, -3.9157e-03,  3.7275e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0450, -0.2634,  0.0105,  0.0209,  0.0242, -0.0153,  0.0109, -0.0147,
        -0.0139, -0.0149,  0.0070,  0.0008, -0.0348, -0.0847,  0.0290,  0.0064,
         0.0159,  0.0305, -0.0520, -0.0447,  0.0041,  0.0050, -0.0067, -0.0093,
        -0.0019, -0.0037, -0.0185, -0.0137,  0.0138, -0.0398,  0.0105, -0.0606,
        -0.0454,  0.0273,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.1089, -0.1579, -0.0552, -0.0148, -0.0056, -0.0225, -0.0095, -0.1122,
        -0.0590, -0.0573, -0.0648, -0.0195, -0.0425, -0.0664, -0.0578, -0.0170,
        -0.0243, -0.1048,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2466, -0.0238,  0.0026, -0.0199, -0.0147,  0.0091, -0.0107,  0.0098,
        -0.0078, -0.1188,  0.0452, -0.0327, -0.0359, -0.0471, -0.0639, -0.0139,
        -0.0369, -0.0378, -0.0053, -0.1084,  0.0175, -0.0004, -0.0041,  0.0222,
         0.0288,  0.0236,  0.0127,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0058, -0.0768, -0.0122, -0.0169,  0.0168, -0.0428, -0.0173, -0.1051,
        -0.1261, -0.0466, -0.0306, -0.0367, -0.0447, -0.0552, -0.0076,  0.0162,
        -0.0607, -0.0047, -0.0215, -0.1067, -0.0199, -0.0135, -0.0022,  0.0289,
         0.0106, -0.0739,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.4405e-01, -1.4898e-01, -4.0835e-02, -4.2055e-02, -1.5807e-01,
        -5.2785e-02, -4.2197e-05,  4.3304e-03, -3.9390e-02, -3.9254e-02,
        -4.8566e-02, -1.5216e-02, -8.7785e-03,  8.1675e-03,  1.1829e-02,
        -3.3921e-02, -3.9999e-02, -1.6920e-02, -2.6313e-02,  2.8608e-02,
        -9.1890e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0850, -0.0790, -0.0963, -0.1647, -0.0161,  0.0308, -0.0271, -0.0051,
         0.0127,  0.0817, -0.0557, -0.0083,  0.0827,  0.0914, -0.1633,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0772,  0.0431,  0.0234,  0.0385, -0.0927,  0.0873, -0.0960, -0.0131,
        -0.0287, -0.0028,  0.0662,  0.0539,  0.0765,  0.0619, -0.0280, -0.0317,
        -0.1052,  0.0740,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0133, -0.1726, -0.0842, -0.1029, -0.0522, -0.0022, -0.0104, -0.0280,
         0.0729, -0.0338, -0.0417, -0.0155, -0.0232, -0.0304, -0.0355, -0.0193,
        -0.0471, -0.0729, -0.1419,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.9510e-02, -1.0017e-01, -1.4270e-03,  9.6587e-03, -8.0656e-03,
        -2.0383e-02,  1.2747e-02, -1.1439e-02, -1.1911e-02,  1.3772e-02,
        -2.5174e-03, -3.6477e-03, -1.5502e-02, -2.5456e-03, -2.2758e-03,
        -1.8128e-02, -3.9599e-02,  7.1650e-06,  3.3221e-02,  6.5581e-02,
        -5.9485e-02,  6.1527e-03, -3.1713e-02, -2.4150e-02, -3.1223e-02,
        -3.1384e-02, -2.8732e-02,  2.6881e-02, -1.9092e-02, -1.7086e-02,
         2.6384e-03, -4.0706e-02, -4.0875e-02, -3.3197e-03, -3.0855e-03,
         1.6632e-02, -7.2497e-02, -4.3875e-03, -2.3587e-02,  5.5627e-03,
         5.5169e-03,  4.9219e-02,  3.6246e-02, -1.7720e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0475, -0.0443, -0.0411, -0.0759, -0.0510, -0.0244, -0.0170, -0.0034,
        -0.0711, -0.0010, -0.0297,  0.0212,  0.0115,  0.0008,  0.0277,  0.0039,
         0.0624, -0.1144,  0.0018,  0.0015, -0.0384,  0.0126,  0.0052,  0.0028,
        -0.0220, -0.0134,  0.0006, -0.0399,  0.0445, -0.0056, -0.0184, -0.0186,
        -0.0010, -0.1197,  0.0056,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.1266e-03, -4.1804e-01,  1.6068e-02, -2.7165e-02,  1.0816e-02,
        -2.2404e-02, -6.3759e-02, -8.1702e-02, -2.9582e-02, -1.9213e-02,
        -2.4662e-04, -5.3807e-02, -6.1170e-02,  2.9888e-05, -2.5610e-03,
        -4.0720e-02,  2.8297e-02,  7.1996e-03,  2.2988e-03, -3.0254e-02,
        -2.4996e-02, -7.9762e-03, -4.1237e-02, -4.3288e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0236, -0.1150, -0.0141, -0.0108, -0.0011, -0.0214, -0.0349, -0.0206,
        -0.0493,  0.0062, -0.0167, -0.0272,  0.0210, -0.0142, -0.0671,  0.0007,
         0.0308,  0.0160, -0.1376, -0.0075, -0.0312, -0.0249, -0.0021, -0.0238,
        -0.0395, -0.0325, -0.0169, -0.0166,  0.0038, -0.0095, -0.0309, -0.0179,
         0.0263, -0.0880,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0676,  0.1292,  0.1750,  0.1018, -0.0281, -0.0216, -0.0056,  0.0038,
        -0.0291, -0.0191,  0.0067, -0.0189, -0.0004,  0.0172,  0.0006, -0.0490,
        -0.0234, -0.0229, -0.0019,  0.0274, -0.0003, -0.0153, -0.0240, -0.0134,
        -0.0181,  0.0186,  0.0188, -0.0209,  0.0730,  0.0483,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.1493, -0.1082, -0.0831, -0.0526, -0.0010, -0.0413, -0.0319, -0.0386,
        -0.0991, -0.2780, -0.0086, -0.0090,  0.0992,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0860, -0.0572,  0.1121, -0.0041, -0.0281, -0.0340, -0.0319, -0.0668,
        -0.1687, -0.0094, -0.0596, -0.0323, -0.0201, -0.0135, -0.1054, -0.1010,
         0.0191,  0.0506,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.3417, -0.0889, -0.0034, -0.0490, -0.1559, -0.1544, -0.1815, -0.0254,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0281, -0.0668, -0.0109, -0.0197, -0.0114, -0.0271, -0.0160, -0.0105,
        -0.0017, -0.0103, -0.0347, -0.0384, -0.0115,  0.0152, -0.0083, -0.0041,
         0.0007,  0.0043, -0.0419,  0.0146, -0.0570, -0.0176, -0.0456, -0.0324,
        -0.0097, -0.0309, -0.0287,  0.0322, -0.0869,  0.0088,  0.0014, -0.0057,
         0.0062, -0.0857,  0.0318, -0.0171, -0.0159, -0.0224,  0.0047, -0.0111,
        -0.0122, -0.0036, -0.0069,  0.0066,  0.0041, -0.0304, -0.0083],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0175, -0.1564, -0.0789, -0.2952,  0.0047, -0.0870,  0.0307,  0.0979,
        -0.0178,  0.0453, -0.1687,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2035, -0.0554, -0.0188, -0.0347, -0.0726, -0.0420, -0.0375, -0.1038,
        -0.1034, -0.0290, -0.0397, -0.0506, -0.0422, -0.0045,  0.0584, -0.0043,
        -0.0476, -0.0519,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1675, -0.1949, -0.1005,  0.0395,  0.0637,  0.0598, -0.0661, -0.0038,
        -0.0221,  0.0325,  0.0008,  0.0162, -0.0576,  0.1096,  0.0653,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0585, -0.1865,  0.0255, -0.0569, -0.0284, -0.0681, -0.1649,  0.0576,
        -0.0251, -0.0323, -0.1533, -0.0132,  0.1075, -0.0135, -0.0087,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1528, -0.0746, -0.0473,  0.0107,  0.0353,  0.0468, -0.0310,  0.0069,
        -0.0557, -0.0533,  0.0054, -0.0340, -0.0696, -0.0126,  0.0064,  0.0175,
        -0.0466,  0.0421, -0.0028,  0.0287,  0.0024, -0.0730,  0.0125,  0.0750,
         0.0259, -0.0311,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1544, -0.0072, -0.0857, -0.0780, -0.0513,  0.0097, -0.0181, -0.0407,
        -0.2532, -0.0368, -0.0425, -0.0581, -0.0621, -0.0384,  0.0638,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5575e-02, -7.2501e-02, -4.2496e-02, -3.1107e-02, -1.6236e-02,
        -9.4899e-03, -1.5449e-02, -7.5085e-03, -2.7298e-02, -2.1041e-01,
        -2.2669e-02, -2.6550e-02, -2.2046e-02, -2.7357e-02,  2.9390e-02,
         3.6121e-03,  2.2849e-03,  1.3605e-02, -6.8710e-02, -1.9050e-01,
        -1.4644e-04, -1.4016e-02,  2.6878e-02, -1.7865e-02,  4.1929e-02,
        -4.4369e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0345,  0.0851,  0.0193, -0.0103, -0.0243, -0.0555, -0.0283, -0.0412,
        -0.0015, -0.0429, -0.1451, -0.0023, -0.0086, -0.0437, -0.0340, -0.0481,
        -0.0018, -0.0373,  0.0158, -0.0408,  0.0304,  0.0093, -0.0313, -0.0160,
         0.1428, -0.0499,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.1494, -0.1292,  0.0268, -0.0131, -0.0209,  0.0069, -0.0144, -0.0409,
         0.0278, -0.0259,  0.0098, -0.0207, -0.1000,  0.0092, -0.0406,  0.0255,
        -0.0230,  0.0081, -0.0298,  0.0051, -0.0191, -0.0161,  0.0205, -0.0058,
        -0.0783,  0.0018, -0.0210, -0.0017, -0.0191,  0.0112,  0.0094, -0.0024,
         0.0118,  0.0297, -0.0154, -0.0098,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.1240e-02, -1.3938e-01, -6.5264e-02, -8.6430e-02, -5.3814e-02,
        -6.1506e-03, -2.9730e-02,  6.0692e-03, -1.1501e-02, -3.7404e-02,
        -9.3164e-03,  1.2409e-04, -1.8603e-02,  1.2458e-02, -5.9675e-03,
        -6.5499e-03,  1.3223e-02, -1.6305e-02, -2.4887e-03,  1.4944e-03,
        -1.7890e-02,  2.5073e-02, -3.5787e-02,  5.4274e-04, -1.8854e-02,
        -9.8521e-02, -2.1528e-02, -1.1981e-02, -7.6252e-02, -1.6016e-02,
        -1.9539e-02, -3.2605e-02, -4.0601e-02,  2.0297e-03,  3.2298e-02,
        -6.9690e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2187,  0.1041,  0.0376,  0.0434, -0.0478,  0.0335,  0.0433,  0.0149,
        -0.0042, -0.0146,  0.0099,  0.0059,  0.0120, -0.0009, -0.0177, -0.0180,
         0.0592,  0.0928,  0.0466, -0.1750,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.8200e-02, -1.0264e-01,  5.7616e-02, -1.2257e-02,  2.3023e-02,
        -1.2918e-02, -2.1866e-02, -1.5143e-02, -3.5109e-03, -1.8614e-02,
        -8.6967e-03, -1.9045e-02,  1.5867e-02, -2.9557e-02, -7.8375e-02,
        -5.1125e-03,  3.8197e-02, -3.6676e-02,  6.0315e-05, -6.7600e-03,
        -4.8040e-02, -1.7829e-02, -3.7795e-03, -2.7916e-02, -2.2956e-02,
         1.5775e-02, -1.5766e-02, -5.4390e-03, -3.0488e-02, -1.7356e-02,
        -1.4540e-02, -1.4878e-02, -1.0994e-02, -7.8460e-02,  1.1702e-02,
        -5.3850e-02, -2.2384e-02, -2.3709e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0505, -0.0150, -0.0268, -0.0361,  0.0010, -0.0061,  0.0165,  0.0086,
        -0.0233, -0.0703, -0.0278, -0.0917, -0.0022, -0.0078, -0.0604, -0.0351,
        -0.0088, -0.0315,  0.0019, -0.0114, -0.0022, -0.0092, -0.0091,  0.0255,
        -0.0099,  0.0144,  0.0042, -0.0197, -0.0108, -0.0252, -0.0312,  0.0149,
         0.0291,  0.0009, -0.0279,  0.0365, -0.0560, -0.0382,  0.0037, -0.0289,
         0.0697], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.8907e-02, -4.1800e-01,  5.5175e-02,  1.4491e-02, -5.4617e-02,
        -4.7995e-02, -1.1892e-01,  1.1649e-02,  5.0426e-03,  6.2103e-03,
        -3.1018e-02,  5.2508e-03, -3.0985e-03,  3.8614e-03, -5.7876e-03,
        -1.3837e-02, -1.8013e-02, -2.7203e-03, -3.6699e-04, -4.5464e-02,
        -6.8650e-02,  1.0929e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0864, -0.0274, -0.0153, -0.0085, -0.0179, -0.0294, -0.0082, -0.0013,
        -0.0409,  0.0241,  0.0115, -0.0416, -0.0245, -0.0176,  0.0169, -0.0138,
         0.0011, -0.0195,  0.0050, -0.0301, -0.1171,  0.0289, -0.0311, -0.0118,
         0.0012,  0.0018,  0.0078, -0.0310, -0.1087, -0.0322,  0.0045, -0.0124,
         0.0131,  0.0560,  0.1012,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0030, -0.0979, -0.0998,  0.0471, -0.1127,  0.0423,  0.0209,  0.0283,
        -0.0214, -0.0559, -0.0410, -0.0235, -0.0016,  0.0097,  0.0213, -0.0059,
         0.0311, -0.0078, -0.0666, -0.0185, -0.0519,  0.0042,  0.0002, -0.0380,
        -0.0164, -0.0250, -0.0268, -0.0321,  0.0491,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0402, -0.1032, -0.0290, -0.0131, -0.0359, -0.0552, -0.0155, -0.0023,
        -0.0291,  0.0498,  0.0069, -0.0219, -0.0190, -0.0269, -0.0505, -0.0443,
        -0.1526, -0.0534, -0.0187,  0.0146, -0.0843, -0.0783, -0.0333, -0.0080,
         0.0140,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0891,  0.0215, -0.0493, -0.0902, -0.1240,  0.0125, -0.0861, -0.0477,
        -0.0455, -0.0516, -0.0011,  0.0338, -0.0136,  0.0717, -0.0457, -0.0258,
        -0.0085, -0.0187,  0.0269, -0.0041, -0.0087,  0.1047, -0.0191,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1320,  0.0204, -0.0075,  0.0057,  0.0118, -0.0250, -0.0359, -0.0419,
        -0.0273, -0.0009,  0.0020, -0.0083, -0.0312, -0.0595,  0.0380, -0.0411,
         0.0391, -0.0075, -0.0190,  0.0079,  0.0080, -0.0549,  0.0584,  0.0064,
        -0.0327, -0.0046,  0.0016,  0.0040, -0.0306, -0.0085, -0.0014, -0.0296,
        -0.0242, -0.0139,  0.0119,  0.0218,  0.0103,  0.0991, -0.0161,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1412, -0.0206,  0.0661, -0.0500, -0.2621, -0.0931, -0.0831, -0.0783,
        -0.0442, -0.0732, -0.0651,  0.0229,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0034, -0.0902, -0.2211, -0.0214, -0.0057,  0.0259, -0.0187, -0.0209,
         0.0126, -0.0117, -0.0062, -0.0254, -0.0320, -0.0042, -0.0269, -0.0207,
        -0.0026, -0.0224,  0.0028, -0.0108, -0.0096, -0.0293, -0.0042, -0.0110,
         0.0240,  0.0118, -0.0029, -0.0053, -0.0367,  0.0101, -0.0063, -0.0123,
        -0.0058, -0.0054, -0.0402, -0.0331, -0.0341, -0.0024, -0.0124, -0.0094,
         0.0084, -0.0061, -0.0176,  0.0359, -0.0399,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1923, -0.2183, -0.0250, -0.0199,  0.0928, -0.0170,  0.1855, -0.0466,
        -0.0461,  0.0007, -0.0994, -0.0564,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1541, -0.5183, -0.1255, -0.0942, -0.0094, -0.0748, -0.0082,  0.0155,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.1548e-02, -9.3466e-02, -1.5688e-02,  1.2989e-02, -1.4418e-04,
        -1.7763e-03,  1.2941e-02,  1.0717e-02,  2.9067e-02, -7.2989e-02,
        -1.3776e-02,  4.5790e-03, -1.5541e-02, -7.2664e-03, -7.4587e-03,
         2.1514e-02,  2.7043e-02, -3.8495e-02, -7.9022e-03, -1.4550e-02,
        -2.4568e-03,  9.1742e-03, -9.8322e-03, -2.4137e-02, -1.0497e-02,
        -4.6761e-04, -5.8000e-03, -4.5566e-03, -4.4994e-03, -3.9194e-03,
         9.3770e-03,  3.4389e-03,  7.0950e-03,  9.0174e-03, -1.5531e-03,
        -1.2614e-02,  9.4832e-05,  1.8004e-03,  2.1679e-02,  4.4648e-03,
        -1.4327e-03,  2.8813e-03,  5.6593e-03, -2.9758e-04,  1.5737e-03,
         1.1336e-02, -5.7557e-02, -3.5181e-02, -5.2917e-02, -5.4361e-02,
        -1.5942e-02,  1.3484e-03, -1.5402e-02,  1.5982e-02, -2.1664e-02,
         8.4293e-03, -1.4121e-02,  9.5873e-03,  1.1828e-02, -2.5900e-02,
         1.0673e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0684, -0.0722,  0.0362, -0.0238, -0.1022, -0.0038, -0.0344, -0.0026,
        -0.0108,  0.0195,  0.0257, -0.0323, -0.0129, -0.0187, -0.0534,  0.0237,
        -0.0319, -0.0075, -0.0292, -0.0318, -0.0587, -0.0578, -0.0466,  0.0304,
        -0.0135, -0.0235, -0.0193,  0.0076, -0.0227, -0.0343, -0.0119,  0.0319,
        -0.0010,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.5555e-01, -5.0641e-02, -3.0288e-02, -2.9185e-02,  5.1933e-03,
         1.0509e-02, -3.1144e-02, -1.3175e-02, -1.2902e-02,  2.2579e-02,
         4.9262e-02, -2.0894e-02, -1.5090e-02, -1.2442e-02, -4.9967e-05,
        -1.5322e-02, -3.0347e-02, -5.0698e-02, -2.5788e-01, -3.9768e-02,
         2.8540e-02, -1.9680e-02, -4.5732e-02, -5.3131e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0170, -0.0705, -0.0539, -0.0350, -0.0299, -0.0985, -0.0471, -0.0219,
        -0.0086, -0.0534, -0.0140, -0.0975, -0.1415, -0.0316,  0.0206, -0.0321,
         0.0103, -0.0173, -0.0461, -0.0060, -0.0281, -0.0387,  0.0547, -0.0256,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1130, -0.0728, -0.0383, -0.0080,  0.0360,  0.0022, -0.0426, -0.0398,
         0.0233, -0.0352,  0.0089, -0.0271,  0.0044, -0.0066, -0.0258, -0.0401,
        -0.0250,  0.0328,  0.0085, -0.0180, -0.0293, -0.0294, -0.0597, -0.0616,
         0.0005,  0.0136, -0.0379,  0.0075,  0.0861, -0.0067, -0.0377, -0.0005,
        -0.0048,  0.0161,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.3880e-02, -1.3426e-01,  2.9847e-02,  5.0767e-03, -4.0834e-02,
         2.9001e-02,  9.9996e-04, -1.4682e-02, -8.1901e-03, -1.0925e-01,
        -2.8869e-02, -1.6072e-02, -8.7061e-03, -2.1499e-01, -7.9102e-02,
        -6.1677e-02, -1.7178e-02, -1.1644e-04, -2.9841e-02, -1.2119e-02,
         1.2447e-02, -1.5373e-02,  4.6331e-02, -5.0923e-02,  2.3041e-04,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0777,  0.0593,  0.0539, -0.0006,  0.0418,  0.0339,  0.0723,  0.0318,
         0.0780,  0.3349,  0.1035, -0.0026,  0.0227, -0.0140, -0.0577, -0.0153,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0100,  0.1175,  0.0334,  0.0417,  0.0319,  0.0355,  0.2783,  0.0441,
        -0.0429,  0.0162,  0.0086, -0.0262,  0.0401,  0.0214,  0.0379, -0.0168,
         0.0190,  0.0465,  0.0448,  0.0009, -0.0233, -0.0519, -0.0109,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0262,  0.1118,  0.0098, -0.0044,  0.1226,  0.0563,  0.0259,  0.1569,
         0.0789,  0.1021, -0.0007, -0.0143, -0.1587,  0.1314,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.1933, -0.0436, -0.0385, -0.1306, -0.0264,  0.0226, -0.0043,  0.0085,
        -0.0122, -0.0182, -0.0386, -0.0268, -0.0654, -0.0030, -0.0065,  0.0390,
         0.0251, -0.0935, -0.0540, -0.0524, -0.0415, -0.0174,  0.0037, -0.0348,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0355,  0.2102,  0.1041, -0.0467,  0.0104, -0.2001, -0.0008, -0.0194,
        -0.0406,  0.2050,  0.1272,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2110, -0.1071, -0.0485, -0.0913, -0.0232, -0.0389, -0.0451,  0.0322,
        -0.0343,  0.0324,  0.0082,  0.0047, -0.0153, -0.0182, -0.0627, -0.0254,
        -0.0742, -0.0667, -0.0243,  0.0305, -0.0058,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2067, -0.0681, -0.0428, -0.1021,  0.0109, -0.0697, -0.0983, -0.1029,
        -0.0414,  0.0090, -0.0115,  0.0137,  0.0681, -0.1549,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1024, -0.0178, -0.0219,  0.0080, -0.0105, -0.0032, -0.0506,  0.0461,
         0.0749, -0.0227, -0.0821, -0.0923, -0.1209, -0.0163, -0.0132, -0.0048,
         0.0134, -0.0901, -0.0767,  0.0013, -0.1308,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.3426e-01, -7.9229e-02, -3.3109e-02, -1.9345e-02, -7.2997e-02,
        -1.1397e-01, -8.7328e-05, -1.1124e-01, -1.0276e-01, -3.6457e-02,
        -4.8529e-02, -3.5628e-02, -1.1239e-01,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.9861e-01, -2.7275e-02,  5.6394e-03,  4.5476e-04,  1.4560e-03,
        -3.9576e-03, -1.7011e-02, -3.2748e-02, -3.0885e-03, -5.0184e-02,
        -3.0130e-02,  1.3828e-02, -1.7417e-03, -1.6568e-02, -2.0560e-02,
        -2.4268e-02,  2.8957e-03,  2.0019e-02, -9.0922e-03,  3.8925e-03,
         7.6463e-03, -1.9621e-04, -4.9917e-02,  1.6866e-03, -1.2228e-01,
         2.8991e-03, -3.5759e-02, -4.8137e-02, -1.3320e-02,  1.2688e-06,
        -7.2535e-03, -2.9033e-02, -2.0885e-02,  1.5621e-02,  4.2737e-02,
        -1.7265e-03, -1.5559e-02, -7.5083e-04,  1.7286e-03, -1.2259e-02,
        -1.7691e-02,  1.6434e-02,  1.4890e-02,  9.2854e-03,  2.8881e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1651, -0.0357, -0.0199, -0.0181, -0.0066, -0.0425, -0.0181, -0.0112,
        -0.0063, -0.0138, -0.0321,  0.0079,  0.0123, -0.0319, -0.0271,  0.0286,
        -0.0126,  0.0046, -0.0096, -0.0232, -0.0025, -0.1041,  0.0040, -0.0313,
        -0.0018, -0.0015,  0.0119, -0.0259, -0.0849, -0.0399, -0.0018,  0.0620,
        -0.0561, -0.0448,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0437, -0.1776, -0.0262, -0.0111, -0.0029, -0.0061,  0.0208, -0.0098,
        -0.0599, -0.0137, -0.0144, -0.0052,  0.0025,  0.0154,  0.0136,  0.1073,
         0.0228, -0.0205,  0.0089, -0.0530, -0.1225, -0.0329, -0.0529, -0.0028,
         0.0028, -0.1503,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 6.4764e-02, -1.1251e-01, -5.8139e-02, -1.1249e-01,  1.5903e-01,
        -1.5577e-01, -9.1297e-03, -2.6711e-02, -3.8401e-02, -2.4692e-02,
         1.0833e-02, -5.5645e-06,  1.1194e-03, -2.6220e-02,  6.4942e-03,
        -7.8587e-05, -1.9212e-02,  2.0788e-02, -3.3561e-02,  1.0820e-02,
        -3.6001e-02,  3.9808e-02, -3.3416e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.7590e-02, -8.1886e-02,  4.0857e-02, -2.2387e-02, -7.2969e-03,
         5.5074e-03,  8.4712e-03, -1.5066e-02, -6.0509e-02, -2.2127e-02,
         1.9096e-02,  2.4064e-02,  1.4614e-02,  1.3793e-02, -1.4965e-01,
         9.2311e-05, -3.9546e-02, -2.9055e-02, -2.3685e-02, -9.9666e-03,
         4.7171e-03, -1.7517e-02,  2.4689e-03,  6.6750e-02, -2.0067e-03,
         1.4219e-02, -7.8245e-02, -3.8231e-02, -4.0443e-03, -1.0005e-04,
        -1.5637e-02,  3.0666e-03,  2.0234e-02,  2.1557e-04,  1.1983e-02,
         2.0035e-02,  2.0659e-02,  1.4395e-02,  4.6142e-03,  5.6058e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.5588e-01, -5.4665e-02,  1.1202e-02, -3.2043e-02, -2.9873e-02,
        -3.3723e-02, -2.9414e-03, -3.6637e-02,  7.2948e-04, -2.5789e-02,
        -1.1768e-02, -6.5498e-03, -7.9062e-03, -4.5396e-02, -2.3317e-02,
        -2.8976e-02, -2.6769e-03,  2.3302e-02,  9.4210e-03,  1.7184e-04,
        -3.2583e-02, -1.1663e-02, -3.2424e-02, -3.5194e-02, -1.5343e-02,
        -1.2725e-02,  6.3066e-05, -3.0010e-02, -1.2204e-02,  3.1851e-02,
        -7.4914e-02,  2.9788e-02, -1.1954e-01,  1.8739e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.1550, -0.0199, -0.0242,  0.0064,  0.0023,  0.0171, -0.0038,  0.0154,
         0.0056,  0.0318, -0.0071,  0.0093,  0.0444, -0.0184, -0.0807,  0.0039,
        -0.0111,  0.0033, -0.0286,  0.0112, -0.0169, -0.0045, -0.0146,  0.0020,
        -0.0355, -0.0971, -0.0372,  0.0103,  0.0135, -0.0199,  0.0126, -0.0139,
        -0.0034, -0.0088, -0.0048, -0.0023, -0.0090, -0.0136, -0.0102, -0.0181,
        -0.0220,  0.0056, -0.0196, -0.0141, -0.0114, -0.0002,  0.0029, -0.0055,
        -0.0038, -0.0175,  0.0052,  0.0444], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1344, -0.0036,  0.0276,  0.0319, -0.0731, -0.0287, -0.0148, -0.0056,
         0.0141, -0.0239, -0.0106,  0.0055, -0.0385, -0.0562, -0.0502, -0.0166,
        -0.0947,  0.0433, -0.0094,  0.1017, -0.0174, -0.0088, -0.0186, -0.0555,
        -0.0119, -0.0142,  0.0345, -0.0003, -0.0025,  0.0414,  0.0103,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0584, -0.0334, -0.0357, -0.0183, -0.0255,  0.0100, -0.0740,  0.0021,
        -0.0320,  0.0521, -0.0521, -0.0280, -0.0308,  0.0045,  0.0408, -0.0643,
         0.0844,  0.0150, -0.0862, -0.0056, -0.0337, -0.0222, -0.0914,  0.0616,
        -0.0378,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2243,  0.0545, -0.0155, -0.0330, -0.0990, -0.0807, -0.1551,  0.0408,
         0.1213, -0.0757, -0.0255, -0.0746,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0482, -0.2178, -0.0154, -0.0802, -0.0345, -0.0060, -0.0161, -0.0015,
        -0.0222, -0.0438,  0.0102,  0.0290, -0.0467, -0.0406, -0.0666, -0.0347,
        -0.0060,  0.0012, -0.0106,  0.0163, -0.0166, -0.0175,  0.0253,  0.0338,
        -0.0047, -0.0368,  0.0211, -0.0004, -0.0959,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0477, -0.3084,  0.0220, -0.1340,  0.0085, -0.0617, -0.1800,  0.0596,
        -0.0697, -0.0213,  0.0304, -0.0567,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1323, -0.0912, -0.1353, -0.0919, -0.0559, -0.0235, -0.1839,  0.0004,
        -0.1012, -0.0712, -0.0885, -0.0250,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0925,  0.0179,  0.1156, -0.1013,  0.0958,  0.0696,  0.0297,  0.0013,
         0.0638,  0.0149, -0.0095,  0.0521,  0.0030,  0.0449,  0.0549,  0.0188,
         0.0234,  0.0041, -0.0959,  0.0908,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.5670e-02, -1.9960e-01,  3.8523e-04,  2.7067e-03, -1.3278e-01,
        -3.0105e-02,  3.9017e-02,  4.4116e-02, -6.3346e-03, -3.0473e-02,
         2.5369e-02,  8.5058e-03, -1.5864e-02, -2.1144e-02, -1.9411e-04,
         1.2156e-02, -8.9187e-03,  9.5888e-03, -1.4082e-02, -9.3977e-03,
        -3.2729e-02, -5.5572e-03, -1.4573e-02,  1.6812e-02, -5.2010e-02,
         2.3241e-02,  8.0712e-06, -2.9058e-03, -1.5978e-02,  3.7507e-02,
        -2.3323e-03,  1.1567e-02, -1.6392e-02, -4.8546e-02,  3.5025e-03,
         4.3536e-02,  6.3984e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2881, -0.2929, -0.0512, -0.1622, -0.0788,  0.0140,  0.0228,  0.0008,
        -0.0893,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0669,  0.0115,  0.3462,  0.0034, -0.0419, -0.0037,  0.0442, -0.0072,
        -0.0113, -0.0103, -0.0090, -0.0269, -0.0065, -0.0082, -0.0303, -0.0198,
         0.0062, -0.0675,  0.0405,  0.0197,  0.0026, -0.0144, -0.0264, -0.0150,
         0.0165,  0.1125,  0.0313,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0256, -0.1828, -0.0198, -0.0155, -0.0266, -0.0327,  0.0100, -0.0184,
        -0.0076,  0.0061, -0.0125, -0.0280, -0.0242,  0.0304, -0.0221,  0.0264,
        -0.0325,  0.0081, -0.0034,  0.0291,  0.0750,  0.0046, -0.0114,  0.0239,
        -0.0177,  0.0165, -0.0133, -0.0235, -0.1967, -0.0048, -0.0111,  0.0036,
        -0.0082, -0.0282,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.1276, -0.0011,  0.0036,  0.0211, -0.0191, -0.0128, -0.0140, -0.0063,
        -0.0268,  0.0113, -0.0539, -0.0769, -0.0323, -0.0119, -0.0243, -0.0062,
        -0.0237, -0.0103, -0.0137, -0.0019,  0.0135, -0.0189, -0.0360,  0.0003,
        -0.0113, -0.0124, -0.0037, -0.0235,  0.0074, -0.0071, -0.0119, -0.0344,
        -0.0493,  0.0161,  0.0130, -0.0174, -0.0220,  0.0055,  0.0080,  0.0076,
        -0.0186, -0.0619,  0.0082,  0.0111,  0.0820,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0582,  0.0500,  0.0478, -0.0608, -0.0380, -0.0618, -0.2371, -0.0042,
        -0.1017, -0.0744,  0.0722, -0.0239, -0.0633, -0.0159, -0.0118, -0.0789,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.3516, -0.0458, -0.0504, -0.0337, -0.0168, -0.0262,  0.0079,  0.0339,
        -0.0241, -0.0409, -0.0346, -0.0021, -0.0770, -0.0175, -0.0032,  0.0203,
        -0.0086, -0.0040,  0.0117, -0.0022, -0.0098,  0.0276, -0.0088, -0.0336,
         0.0191,  0.0107, -0.0778,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0089,  0.0660, -0.0444, -0.0143, -0.0477, -0.0195, -0.0354, -0.0076,
        -0.0138, -0.0152, -0.0142, -0.0130, -0.0168, -0.0078,  0.0079, -0.0098,
        -0.0132,  0.0104, -0.0153, -0.0305,  0.0111, -0.0078, -0.0146, -0.0787,
         0.0118,  0.0101, -0.0275, -0.0040,  0.0104,  0.0092,  0.0050, -0.0362,
        -0.0152, -0.0035, -0.0376, -0.0206, -0.0264, -0.0456, -0.0488, -0.0383,
        -0.0215, -0.0371, -0.0048,  0.0066, -0.0445, -0.0116], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0883,  0.0272, -0.0143, -0.0135, -0.0161, -0.0155, -0.0204, -0.0204,
         0.0111, -0.0017, -0.0068,  0.0003, -0.0145,  0.0061, -0.0040,  0.0247,
         0.0052, -0.0144, -0.0675, -0.2287, -0.0659, -0.0236,  0.0197,  0.0169,
        -0.0008,  0.0235, -0.0443, -0.0512, -0.0344,  0.1189,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0724, -0.0521, -0.1421, -0.0425, -0.1093, -0.1105, -0.1156,  0.0948,
         0.2120, -0.0280, -0.0207,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0139, -0.1622,  0.0351,  0.0306, -0.0475,  0.0170, -0.0523, -0.1418,
         0.0274, -0.0885,  0.0530, -0.0329, -0.0326, -0.0276,  0.0384,  0.0170,
        -0.0354, -0.0310,  0.0067, -0.1091,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.3079,  0.1640,  0.0677,  0.0575, -0.0643,  0.0499,  0.0823,  0.0037,
         0.0125,  0.1666, -0.0235,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0080, -0.1078, -0.0705, -0.0916, -0.0291,  0.1076, -0.0686, -0.0677,
        -0.0860, -0.0047,  0.0131, -0.0351, -0.0308,  0.0670, -0.0533, -0.0843,
        -0.0018, -0.0384,  0.0080, -0.0173,  0.0095,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0426, -0.1251, -0.1274,  0.0133, -0.1019, -0.2818,  0.0710, -0.0174,
         0.0033, -0.0440,  0.0120, -0.0218,  0.0086, -0.0242, -0.0008, -0.0535,
         0.0142,  0.0372,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0643, -0.0948, -0.1013, -0.0168,  0.0081, -0.0192,  0.0167,  0.0255,
        -0.0069, -0.0146,  0.0208, -0.0055,  0.0232, -0.0037, -0.0543, -0.0116,
        -0.0244, -0.0689, -0.0265, -0.0038, -0.0058, -0.0264, -0.0823, -0.0061,
        -0.0077, -0.0279, -0.0288, -0.0146, -0.0043, -0.0341, -0.0148, -0.0093,
        -0.0384, -0.0143, -0.0092, -0.0039, -0.0420,  0.0187,  0.0005,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0120, -0.2373, -0.0280, -0.0411, -0.0153, -0.0205,  0.0010, -0.0114,
         0.0970,  0.0139,  0.0269, -0.0640, -0.0009, -0.0272, -0.0295,  0.0047,
        -0.0222,  0.0127, -0.0025, -0.0198, -0.0192, -0.0136,  0.0067,  0.0044,
         0.0289, -0.0301, -0.0181, -0.0031, -0.0474, -0.0563,  0.0018,  0.0054,
        -0.0659, -0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.1274, -0.2361,  0.0700, -0.0350, -0.0260, -0.1206, -0.0085,  0.3244,
        -0.0521,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0910, -0.1046, -0.0248,  0.0442,  0.0373,  0.0085, -0.0207,  0.0006,
        -0.0011, -0.0035, -0.0218, -0.0066, -0.0003, -0.0180, -0.0208, -0.0239,
        -0.0057, -0.2038, -0.0285, -0.0279, -0.0192, -0.0016,  0.0182, -0.0092,
        -0.0144, -0.0099, -0.0041, -0.0195, -0.0325, -0.0264, -0.0054, -0.0232,
        -0.0155, -0.0089,  0.0063, -0.0087,  0.0030, -0.0295, -0.0387,  0.0121,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0862, -0.0106, -0.1055, -0.0028, -0.0064, -0.0234,  0.0090,  0.0090,
        -0.0030, -0.0346,  0.0018,  0.0031,  0.0100,  0.0131,  0.0192,  0.0197,
         0.0086,  0.0212, -0.0238, -0.0075, -0.0099, -0.0233,  0.0146, -0.0141,
         0.0197, -0.0241, -0.0254,  0.0152, -0.0585, -0.0289, -0.0244, -0.0222,
         0.0030, -0.0165, -0.0132, -0.0495, -0.0220, -0.0271, -0.0027,  0.0156,
        -0.0166,  0.0121, -0.0165,  0.0029,  0.0162,  0.0066,  0.0081,  0.0112,
        -0.0069, -0.0355, -0.0186,  0.0003,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.8106e-02, -1.5775e-01, -3.6370e-02, -4.1600e-02, -1.2319e-02,
         6.0966e-02, -3.9371e-02, -1.4095e-02, -3.2534e-02,  3.8234e-05,
         8.8343e-03,  1.7729e-02, -2.2757e-02,  8.6714e-04, -4.1877e-02,
        -5.9450e-02, -7.0271e-02, -3.5149e-02, -4.4452e-02, -4.4612e-02,
         4.5011e-02,  1.1584e-01,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0732,  0.2859, -0.0432, -0.0067, -0.0170,  0.0546,  0.0931,  0.0353,
        -0.0225,  0.0154,  0.0661,  0.0056, -0.0080, -0.0368,  0.0033,  0.0445,
         0.0427, -0.0061, -0.0539, -0.0473,  0.0389,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.6752e-02, -4.3253e-02, -3.8249e-04, -4.1331e-02,  3.2273e-02,
        -1.7870e-03, -2.5497e-02,  1.2315e-02, -2.1100e-02, -7.6161e-03,
        -8.1250e-03, -1.7087e-02, -7.6881e-03,  1.3351e-02,  1.9083e-02,
         2.8930e-03,  5.0806e-03,  6.8913e-03, -1.5839e-02, -6.2991e-03,
        -5.7551e-03,  1.8977e-02,  9.6927e-04, -4.5436e-03, -1.8653e-02,
         1.7127e-02,  2.6991e-03, -2.3424e-02, -6.2914e-05,  1.0454e-02,
        -6.1600e-03, -1.4585e-02, -6.5541e-03,  1.2813e-03, -2.2540e-02,
        -2.6400e-03, -4.4833e-04, -3.1137e-03, -9.9209e-03, -1.4411e-02,
        -2.4947e-01, -2.1930e-02, -9.1869e-03,  1.1348e-03, -3.7605e-02,
        -1.7514e-03, -2.6553e-02, -2.9791e-02, -9.5332e-03,  1.1859e-02,
        -9.3693e-03,  1.4070e-02, -5.0054e-03, -1.2715e-02, -1.1060e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0849, -0.0486,  0.0460, -0.0308, -0.0174, -0.0225, -0.0151, -0.0034,
         0.0099,  0.0059,  0.0107,  0.0046, -0.0218, -0.0092,  0.0085, -0.0112,
         0.0064, -0.0036, -0.0204, -0.0216, -0.0205, -0.0149, -0.0113,  0.0112,
         0.0133,  0.0339, -0.0375, -0.0738, -0.0206, -0.0185, -0.0101, -0.0101,
        -0.0293, -0.0938, -0.0176, -0.0387, -0.0052, -0.0205, -0.0036, -0.0058,
        -0.0105,  0.0068,  0.0556,  0.0160, -0.0187,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0125,  0.2034,  0.0529,  0.0798, -0.0466,  0.0209,  0.0347,  0.0114,
         0.0364, -0.0159, -0.0394, -0.0013,  0.0131,  0.0442,  0.0421,  0.0247,
         0.0178,  0.0737,  0.0267,  0.0253,  0.0807,  0.0659, -0.0140,  0.0167,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0002,  0.0619,  0.0296, -0.0169,  0.0617,  0.0401,  0.0380,  0.0189,
         0.0291, -0.0106,  0.0081,  0.0178, -0.0030,  0.0220,  0.0179, -0.0231,
         0.1030,  0.0134, -0.0048,  0.0392,  0.0323, -0.0340,  0.0502,  0.0207,
        -0.0001,  0.0316,  0.0794,  0.0547,  0.0042,  0.0766, -0.0079,  0.0377,
         0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1418, -0.2025, -0.0835, -0.0760, -0.0308, -0.0554, -0.0095,  0.0018,
        -0.1193, -0.0386,  0.0125, -0.0055, -0.0296, -0.0045, -0.0014, -0.0311,
        -0.0059, -0.0873, -0.0631,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1858, -0.2151, -0.0523, -0.0199, -0.0065,  0.0211, -0.0391, -0.0196,
        -0.0939, -0.0156,  0.0194, -0.0109,  0.0016, -0.0066,  0.0086, -0.0031,
         0.0078,  0.0127, -0.0031, -0.0324, -0.0356, -0.0732, -0.0286,  0.0184,
         0.0691,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0790, -0.1497, -0.0085, -0.0687, -0.1125, -0.0544, -0.1780, -0.0154,
        -0.0675, -0.0602,  0.0288, -0.0211, -0.0162, -0.0328, -0.0525,  0.0101,
         0.0445,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0025, -0.0017,  0.0070,  0.0170,  0.0318,  0.0107, -0.0211,  0.0176,
        -0.0763,  0.0707,  0.0108,  0.0138,  0.0184, -0.0278,  0.0504,  0.0291,
         0.0457,  0.0372, -0.0143,  0.0260,  0.0199,  0.0505,  0.1402,  0.0023,
         0.0730,  0.0155,  0.0085,  0.0311,  0.0623, -0.0334, -0.0119,  0.0214,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0389, -0.1448, -0.0697, -0.0259, -0.0326, -0.0240, -0.0532, -0.0579,
        -0.0187, -0.0261,  0.0091, -0.0073,  0.0675, -0.0289, -0.0014, -0.0058,
        -0.0276, -0.0091, -0.0243, -0.0025, -0.0299, -0.0074, -0.0253, -0.1081,
         0.0154, -0.0269, -0.0850,  0.0265,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0481, -0.1382,  0.0323, -0.0146,  0.0218, -0.0346, -0.0959, -0.0322,
        -0.0240, -0.0349, -0.0110, -0.0111, -0.0073, -0.0264, -0.0126, -0.0627,
        -0.0199, -0.0278, -0.0048, -0.0427, -0.0402,  0.0100,  0.0169, -0.0357,
        -0.0376,  0.0321, -0.0444, -0.0117, -0.0687,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0638,  0.0122, -0.0131, -0.0058, -0.1159, -0.0229,  0.0139, -0.0790,
         0.0121,  0.0329,  0.0661, -0.0186, -0.0228, -0.0985, -0.0600, -0.0089,
        -0.0229, -0.0818, -0.0337, -0.0073,  0.0419,  0.0018,  0.0014, -0.0005,
        -0.0210, -0.0027, -0.0028, -0.0171, -0.0093,  0.0038, -0.0016,  0.0201,
        -0.0837,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0927, -0.0994, -0.0283, -0.0124, -0.0299, -0.0498,  0.0086, -0.0130,
        -0.0298, -0.0571, -0.0413, -0.0014, -0.0176, -0.0058, -0.0431, -0.0041,
        -0.0059, -0.0329, -0.1009,  0.0166, -0.1072, -0.0246, -0.0576,  0.1200,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1232, -0.0976, -0.0680, -0.0142, -0.0522, -0.0911, -0.0539, -0.0524,
        -0.0252, -0.0434, -0.0216, -0.0039, -0.0447, -0.0298,  0.0013, -0.0029,
        -0.0075, -0.0294, -0.0119, -0.1659, -0.0433, -0.0089, -0.0078,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0013, -0.0761, -0.0228,  0.0243, -0.0119, -0.0641, -0.0216,  0.0035,
         0.0013, -0.0234,  0.0071, -0.0316, -0.0291, -0.0436, -0.0229, -0.0281,
        -0.0293, -0.0468, -0.0287, -0.0420, -0.0024, -0.0430, -0.0128,  0.0094,
        -0.0315, -0.0090, -0.0017, -0.0119, -0.0448, -0.0429, -0.0551, -0.0154,
        -0.0177,  0.0285, -0.0157, -0.0215, -0.0211, -0.0427, -0.0134,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0367, -0.1182, -0.0576,  0.0021, -0.0940, -0.0177,  0.0162,  0.0102,
        -0.0227, -0.0087, -0.0117, -0.0117, -0.0072, -0.0037,  0.0092, -0.0217,
        -0.0231, -0.0018,  0.0107,  0.0043, -0.0486, -0.0057, -0.0081, -0.0162,
        -0.0052, -0.0275, -0.0305, -0.0152, -0.0143, -0.0195, -0.0277, -0.0052,
        -0.0321, -0.0159, -0.0267, -0.0020, -0.0236, -0.0562, -0.0124, -0.0091,
        -0.0210, -0.0450, -0.0431,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0435, -0.1532,  0.0130,  0.0085, -0.0231,  0.0085, -0.0153, -0.0054,
        -0.0291, -0.0104, -0.0115,  0.0044, -0.0204, -0.0111,  0.0021, -0.0078,
        -0.0126, -0.0227,  0.0004, -0.0030,  0.0060, -0.0132, -0.0140, -0.0197,
        -0.0137, -0.0103, -0.0216, -0.0073,  0.0338, -0.0329,  0.0307,  0.0186,
        -0.0338,  0.0337,  0.0123, -0.0071,  0.0240,  0.0148, -0.0290, -0.0016,
        -0.0233, -0.0331, -0.0026,  0.0178,  0.0171, -0.0200, -0.0010, -0.0072,
         0.0124, -0.0031, -0.0085, -0.0176, -0.0027, -0.0082, -0.0208, -0.0115,
         0.0088], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0720, -0.3000,  0.0035, -0.0668, -0.0509, -0.0746,  0.0006, -0.1214,
        -0.1034,  0.0485, -0.0492,  0.0235, -0.0086, -0.0770,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0615, -0.1201, -0.0582, -0.0225, -0.0095, -0.0254, -0.0099, -0.0151,
        -0.0261, -0.0173, -0.0233, -0.1311, -0.0261, -0.0174, -0.0306, -0.0422,
         0.0041, -0.0301, -0.0007, -0.0250, -0.0847, -0.0283, -0.0210, -0.0051,
        -0.0139,  0.0149, -0.0004, -0.0016,  0.0497,  0.0841,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1735,  0.1982,  0.0627,  0.3395,  0.0897, -0.0392,  0.0331, -0.0642,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([-0.1352,  0.0472,  0.0320,  0.0178,  0.0128,  0.0866, -0.0099, -0.0091,
        -0.0145,  0.0033,  0.0221,  0.0145,  0.0203,  0.0553,  0.0121,  0.0029,
         0.0213,  0.0026, -0.0106,  0.0065, -0.0239,  0.0124,  0.0272,  0.0261,
         0.0464,  0.0301,  0.0195, -0.0149,  0.0281,  0.0667, -0.0345, -0.0725,
        -0.0612,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.2118e-02, -8.5721e-02, -4.8138e-02, -1.1410e-02, -4.3534e-03,
        -6.1182e-03,  2.3340e-03, -1.0338e-02, -1.8965e-02, -1.7048e-02,
         1.4214e-03,  5.8146e-04, -1.1782e-02, -1.5393e-02,  4.5773e-04,
         1.0167e-03, -1.1265e-02, -1.1131e-02, -8.4584e-03, -1.8774e-03,
         1.0721e-03,  8.0800e-04, -1.5197e-02, -1.2804e-02,  1.4934e-02,
        -1.5919e-02, -5.3133e-03, -2.9384e-02,  2.7494e-03,  2.0098e-05,
        -5.4767e-03, -7.6296e-03, -1.8675e-02, -7.6186e-02, -1.1085e-02,
         3.5877e-04, -1.6452e-02, -1.0685e-02, -7.7110e-03, -8.7276e-04,
         3.6230e-03, -8.9832e-03, -6.1102e-03,  1.8770e-03, -9.5347e-03,
         3.0648e-02, -2.8641e-02, -4.6744e-03, -3.3966e-02, -3.1135e-02,
        -1.0354e-02, -1.7489e-02, -3.2303e-03, -3.0316e-03,  2.3335e-03,
         1.2035e-02, -8.2927e-03, -9.3808e-04, -2.2006e-02, -3.2344e-02,
        -1.9378e-02, -1.3292e-03, -4.6200e-02, -3.1657e-02, -4.7606e-03,
        -1.0803e-02, -2.6565e-03,  8.4808e-04, -2.0807e-02,  1.5713e-02,
         2.1339e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.3705e-02, -4.3083e-02, -2.7854e-02,  2.8662e-02, -4.0566e-02,
        -3.4167e-03, -2.1610e-01, -2.3295e-02, -1.2523e-02,  9.1352e-03,
         8.9429e-03,  1.5887e-03, -1.0894e-02,  8.9026e-03, -1.0557e-02,
        -2.6749e-02, -2.7665e-03, -9.8568e-03, -2.0970e-02, -2.7609e-02,
         4.2572e-02, -9.4675e-04,  5.8850e-02, -4.7615e-02,  2.1370e-03,
        -4.9513e-03, -2.3264e-02,  2.3262e-05,  3.3667e-02, -2.5390e-02,
        -4.2390e-03, -1.7254e-02, -1.1077e-02, -1.4131e-02, -1.1003e-02,
        -1.7198e-02, -2.1046e-02, -5.3600e-03, -1.3061e-02,  1.1537e-02,
        -4.2997e-02, -4.5044e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0158,  0.1307,  0.0185, -0.0004,  0.0300, -0.0230,  0.0097,  0.0123,
        -0.0300,  0.0303,  0.0669,  0.0119,  0.0558,  0.0042,  0.0061, -0.0179,
        -0.0093, -0.0003,  0.0463,  0.0151, -0.0002,  0.0052,  0.0197, -0.0070,
         0.0596,  0.0206,  0.0232,  0.0237,  0.0536,  0.0113,  0.0137, -0.0066,
         0.0045,  0.0345,  0.0042, -0.1374,  0.0401,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0272, -0.4598, -0.0035, -0.1206, -0.0606,  0.0066, -0.0299, -0.0020,
         0.0612, -0.0053, -0.0298, -0.0672, -0.0820, -0.0339, -0.0104,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0872,  0.1675, -0.0291,  0.0450,  0.0636,  0.0487, -0.0019, -0.0062,
         0.1868,  0.0058,  0.0040, -0.0529,  0.0965, -0.0026, -0.0248, -0.0019,
         0.0813, -0.0064, -0.0132,  0.0747,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0900,  0.0253,  0.0210,  0.0254, -0.1720, -0.0251, -0.0132, -0.0010,
         0.0032, -0.0129, -0.0034, -0.0154, -0.0123, -0.0164, -0.0119, -0.0163,
        -0.0076, -0.0113, -0.0192, -0.0139, -0.0167,  0.0143, -0.0269, -0.0088,
        -0.0131, -0.0050, -0.0866, -0.0075,  0.0261, -0.0044, -0.0127,  0.0003,
         0.0185,  0.0021,  0.0088, -0.0267,  0.0108, -0.0035, -0.0142, -0.0047,
        -0.0153,  0.0784,  0.0778,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0299,  0.0359, -0.0133, -0.0512, -0.0079, -0.0270, -0.0311, -0.1345,
        -0.0276,  0.0071, -0.0030, -0.0085, -0.0192, -0.0143, -0.0556, -0.0078,
        -0.0004,  0.0140, -0.0031, -0.0102, -0.0407, -0.0125, -0.0149, -0.0213,
        -0.0276, -0.0051, -0.0256, -0.0365, -0.0609, -0.0216, -0.0042, -0.0352,
        -0.0276, -0.0110, -0.0215, -0.0210, -0.0526, -0.0587,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0410, -0.1471, -0.0034, -0.1696, -0.0090, -0.0565, -0.0309,  0.0796,
         0.0271, -0.1472, -0.0124, -0.0095, -0.0356, -0.0031,  0.0116,  0.0582,
         0.0118, -0.0097,  0.0179, -0.0032,  0.0502, -0.0652,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0388, -0.0205,  0.0395,  0.0075, -0.0025,  0.0204, -0.0131, -0.0229,
         0.0269, -0.0457,  0.0088,  0.0049, -0.0188,  0.0060, -0.0146, -0.0198,
        -0.1470, -0.0127, -0.0454, -0.0150, -0.0144,  0.0266, -0.0495, -0.0236,
        -0.0383, -0.0145, -0.0049,  0.0354, -0.0012, -0.0076,  0.0074, -0.0753,
         0.0017, -0.0101, -0.0102, -0.0300, -0.0057,  0.0172, -0.0185, -0.0769,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0548,  0.0545, -0.0168, -0.0755, -0.0225, -0.0165,  0.0149,  0.0196,
        -0.0396, -0.0428,  0.0107, -0.0002, -0.0252, -0.0525, -0.0770, -0.0498,
         0.0423, -0.0178, -0.0849, -0.0228, -0.0262, -0.0042, -0.0104, -0.0172,
         0.0027, -0.0934, -0.0082, -0.0368, -0.0342, -0.0020, -0.0202, -0.0038,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.8248e-02,  3.0952e-01, -1.6691e-02,  9.2700e-03,  2.3799e-02,
         1.9369e-02,  1.3057e-02,  3.4405e-02,  7.1908e-02,  2.7896e-02,
         7.4813e-03,  1.6796e-02,  7.3955e-03,  1.1250e-02,  8.4572e-03,
         2.3022e-02,  1.7453e-02,  1.7742e-02,  2.8532e-03,  8.4096e-03,
         6.9892e-03,  3.3786e-02,  1.7290e-02,  3.7728e-02, -5.8223e-03,
         1.2127e-02,  2.4147e-02,  3.2047e-02, -1.2438e-02,  1.3341e-02,
        -1.5263e-02,  2.2007e-02,  1.3413e-03,  4.0338e-02,  3.0846e-04,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.1273, -0.0594, -0.0881, -0.0854, -0.0024, -0.0211, -0.0643, -0.0936,
         0.0297, -0.0507,  0.0344, -0.0327, -0.0097, -0.0485, -0.0087,  0.0198,
        -0.0117, -0.0187, -0.0303, -0.0657, -0.0979,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0652,  0.2283,  0.0082,  0.0708,  0.0850,  0.0530, -0.0147,  0.0708,
         0.0875, -0.0283, -0.0116, -0.0690, -0.0150, -0.0196,  0.0522, -0.0091,
         0.0309,  0.0541, -0.0268,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0770, -0.1740, -0.0418, -0.0283,  0.0053, -0.0356,  0.0062,  0.0003,
         0.0076, -0.0699, -0.1151,  0.0051, -0.0363, -0.1144, -0.0123, -0.0084,
        -0.0191, -0.0267,  0.0746,  0.1419,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0289, -0.2740,  0.0292, -0.1065, -0.0233,  0.0021,  0.0740, -0.0171,
         0.0047, -0.0298, -0.0336, -0.0324,  0.0218, -0.0357, -0.0301, -0.0442,
        -0.0766, -0.0323,  0.0025, -0.0432,  0.0084, -0.0411,  0.0085,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0020, -0.0862, -0.0008,  0.0086,  0.0076, -0.0400, -0.0169, -0.0141,
         0.0092, -0.1492, -0.0743, -0.0305, -0.0281, -0.1511, -0.0027,  0.0304,
         0.0508, -0.0166, -0.0227,  0.0285, -0.0356,  0.0094, -0.0442,  0.0549,
        -0.0856,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0830,  0.2582,  0.0254,  0.0303,  0.0020,  0.0312,  0.0111,  0.0445,
        -0.0207, -0.0166,  0.0360, -0.0033,  0.0039,  0.0057, -0.0266,  0.0110,
        -0.0037, -0.0087, -0.0361,  0.0074,  0.0148,  0.0229,  0.0420,  0.0303,
         0.0236,  0.0230,  0.0123, -0.0065,  0.0417, -0.0107,  0.0184, -0.0055,
         0.0066,  0.0310,  0.0455,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0977,  0.0596,  0.0688, -0.0445, -0.0097,  0.0015, -0.0348, -0.0110,
        -0.0832, -0.0116, -0.0787, -0.0184,  0.0120, -0.0271, -0.0433, -0.1266,
        -0.0341, -0.0278, -0.0433, -0.0428, -0.0925,  0.0309,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1945,  0.0415,  0.0028,  0.0083, -0.0026,  0.0448,  0.0095, -0.0158,
        -0.0825,  0.0533,  0.0690, -0.0039,  0.0525,  0.0826, -0.0977,  0.0227,
         0.0397,  0.0160,  0.0381, -0.0127,  0.1094,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1768, -0.2114, -0.0379,  0.0614, -0.0312, -0.0438, -0.0155, -0.1155,
        -0.0453, -0.0213,  0.0212, -0.0063,  0.0183, -0.0553,  0.0302, -0.0488,
         0.0393, -0.0027,  0.0179,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0364, -0.0203, -0.0757,  0.0229, -0.0425, -0.0608,  0.0296,  0.0068,
        -0.0039,  0.0121,  0.0106, -0.0103,  0.0006, -0.0215, -0.0017,  0.0061,
        -0.0087, -0.0358, -0.0114, -0.0089,  0.0058, -0.0414,  0.0086, -0.0054,
         0.0487,  0.0323,  0.0104, -0.0497, -0.0004, -0.0580,  0.0079,  0.0133,
        -0.0355, -0.0783, -0.0309, -0.0007,  0.0117,  0.0237,  0.0157, -0.0149,
         0.0038, -0.0086, -0.0077,  0.0048, -0.0025, -0.0509, -0.0016],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0600, -0.1272, -0.0641, -0.0045,  0.0030, -0.0194, -0.0149,  0.0063,
        -0.0286,  0.0382,  0.0037, -0.0147, -0.0123,  0.0169, -0.0299, -0.0218,
        -0.0218, -0.0163, -0.0328,  0.0206, -0.0177, -0.0205, -0.0268, -0.1695,
         0.0037,  0.0101, -0.0573,  0.0278, -0.0190, -0.0455,  0.0194, -0.0006,
         0.0245, -0.0007,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0045, -0.3124, -0.0122,  0.0062, -0.0305, -0.0191, -0.0155, -0.0067,
        -0.0179, -0.0109,  0.0261,  0.0051, -0.0308, -0.0627,  0.0473,  0.0080,
        -0.0029, -0.0026, -0.0737, -0.0285,  0.0117, -0.0472, -0.0138, -0.0320,
        -0.0047, -0.0137, -0.0142, -0.0220,  0.0098, -0.0372, -0.0238, -0.0312,
        -0.0134,  0.0019,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0145, -0.2665, -0.1811, -0.0042, -0.0441, -0.0262,  0.0551,  0.0874,
        -0.0450, -0.0408, -0.0199,  0.0220, -0.0189, -0.0280, -0.0159, -0.0863,
        -0.0288,  0.0154,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0817, -0.1282, -0.0430,  0.0276, -0.0139,  0.0160, -0.0143, -0.0237,
         0.0039, -0.1251, -0.0170,  0.0013, -0.0258, -0.0352, -0.0246,  0.0015,
        -0.0069,  0.0377,  0.0215, -0.1200,  0.0307,  0.0052, -0.0318, -0.0125,
        -0.0214,  0.0632, -0.0662,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0163,  0.0176, -0.0021,  0.0255, -0.0062, -0.1210,  0.0596,  0.0649,
        -0.1583,  0.0364,  0.0590,  0.0582, -0.0664,  0.0166, -0.0072, -0.0020,
         0.0013, -0.0041, -0.0362, -0.1185, -0.0047, -0.0064,  0.0064,  0.0496,
        -0.0088, -0.0466,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0803,  0.2037,  0.0198,  0.0363,  0.1172, -0.0058,  0.0022, -0.0145,
         0.0454,  0.0450,  0.0602, -0.0140,  0.0241,  0.0673, -0.0103,  0.0206,
         0.0116, -0.0095,  0.0327, -0.1524, -0.0272,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0699, -0.1419, -0.1153, -0.0892, -0.0320, -0.0957,  0.0477, -0.0622,
        -0.0266,  0.0522, -0.0295, -0.0501, -0.0202, -0.0916,  0.0759,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1423, -0.0130, -0.0542,  0.0837, -0.0550,  0.0534, -0.0443, -0.0460,
         0.0060, -0.0127, -0.0104,  0.1570,  0.0877,  0.0318, -0.0602, -0.0108,
        -0.0649, -0.0666,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1462, -0.1892, -0.0711, -0.0916, -0.0449, -0.0144,  0.0324, -0.0481,
         0.0099, -0.0372, -0.0475, -0.0239,  0.0255, -0.0755, -0.0151, -0.0320,
         0.0307,  0.0087, -0.0559,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.1402e-01, -4.4830e-02, -2.5294e-02,  1.3063e-03, -1.0018e-02,
        -2.3801e-02,  4.9599e-02, -6.2297e-03, -1.3594e-02, -1.1371e-02,
        -2.8047e-03,  2.0750e-03, -9.5728e-03, -5.3934e-03,  1.9698e-02,
        -1.2197e-02, -3.1489e-02, -1.0600e-02, -9.6912e-03,  3.5540e-02,
        -7.3622e-02,  3.8705e-02, -4.0648e-02, -2.3358e-02,  3.2412e-03,
        -8.5324e-03, -2.9156e-02,  1.4223e-02, -4.8932e-02, -2.7008e-02,
         1.5005e-05, -3.4741e-02, -3.1344e-02, -2.1286e-02, -9.9563e-03,
         4.5888e-03, -2.2980e-02,  7.1736e-03, -4.1058e-03,  4.7656e-02,
         1.3355e-02,  4.6069e-02, -3.5725e-03, -6.6071e-03], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2133,  0.0552, -0.0134, -0.0058,  0.0021, -0.0291, -0.0473, -0.0036,
        -0.0510,  0.0483, -0.0225, -0.0076, -0.0063, -0.0144,  0.0079, -0.0359,
        -0.0573, -0.0359,  0.0099, -0.0088, -0.0127, -0.0132,  0.0185,  0.0028,
        -0.0008, -0.0064,  0.0041, -0.0347,  0.0412, -0.0243, -0.0473,  0.0338,
        -0.0161, -0.0182, -0.0500,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0404, -0.1787, -0.0091, -0.0016, -0.0358, -0.0776, -0.0294,  0.0026,
        -0.0102,  0.0078, -0.0025, -0.0083, -0.0783, -0.0104, -0.0425, -0.0151,
        -0.0415, -0.0226, -0.0540,  0.0260, -0.0119,  0.1003,  0.1797,  0.0140,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0697, -0.1086, -0.0397, -0.0297, -0.0184, -0.0366,  0.0218,  0.0015,
        -0.0155, -0.0108, -0.0020,  0.0204,  0.0129, -0.0407, -0.0835,  0.0256,
         0.0187,  0.0005, -0.0742, -0.0052,  0.0333, -0.0279, -0.0036, -0.0108,
        -0.0205, -0.0432, -0.0292, -0.0253,  0.0020,  0.0517, -0.0430,  0.0023,
        -0.0498,  0.0215,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1653,  0.0919, -0.0188,  0.0083,  0.0184,  0.0358,  0.0760,  0.0058,
         0.0386, -0.0055,  0.0277,  0.0143,  0.0242,  0.0250,  0.0315,  0.1618,
         0.0082, -0.0247,  0.0495,  0.0093, -0.0064,  0.0007, -0.0210,  0.0219,
         0.0172, -0.0040,  0.0029, -0.0140, -0.0347, -0.0368,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0617, -0.2680, -0.1011, -0.1598, -0.0346, -0.0580, -0.0438, -0.0200,
         0.0046, -0.0589, -0.0705, -0.0988, -0.0201,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0481,  0.0262,  0.0629, -0.0146, -0.0330, -0.0529,  0.0305, -0.0353,
        -0.1453,  0.0332, -0.0621, -0.0690, -0.0119, -0.0779, -0.1077, -0.0960,
        -0.0184,  0.0750,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1861, -0.1061, -0.0615, -0.2179, -0.1959,  0.0072, -0.0526, -0.1727,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0161, -0.1035, -0.0261,  0.0115, -0.0057, -0.0157,  0.0076,  0.0107,
         0.0356, -0.0040,  0.0075, -0.0004, -0.0009, -0.0210, -0.0005,  0.0001,
         0.0002,  0.0089, -0.0091,  0.0036,  0.0125, -0.0071, -0.0149, -0.0080,
         0.0037, -0.0026,  0.0076,  0.0324, -0.1001, -0.0200,  0.0366, -0.0484,
        -0.0328, -0.0865, -0.0197, -0.0499, -0.0248, -0.0090, -0.0245, -0.0040,
        -0.0152, -0.0295,  0.0041,  0.0037, -0.0076,  0.0604,  0.0454],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0755, -0.1102, -0.0844, -0.1316, -0.0550, -0.1809,  0.0133, -0.0424,
        -0.1880,  0.0552,  0.0634,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0346, -0.1236, -0.0710, -0.0186, -0.0047, -0.0154, -0.0250, -0.1086,
        -0.1051, -0.0476, -0.0400, -0.0736, -0.1023, -0.0908,  0.0601, -0.0134,
         0.0409,  0.0248,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0875, -0.1514,  0.0730, -0.1296, -0.0070,  0.0815, -0.0055, -0.0355,
        -0.0074, -0.0083, -0.0165,  0.0360, -0.0572, -0.0772, -0.2265,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0668,  0.1356,  0.0100, -0.0063, -0.0101,  0.0110,  0.2674,  0.0097,
         0.0540,  0.0190,  0.1691, -0.0056,  0.0828, -0.0618,  0.0907,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1035, -0.0348,  0.0491, -0.0399, -0.0310,  0.0206,  0.0257, -0.0044,
        -0.0575, -0.0641, -0.0236, -0.0291, -0.0803,  0.0181, -0.0109,  0.0304,
        -0.0765, -0.0170,  0.0176,  0.0459, -0.0411, -0.0917, -0.0077,  0.0106,
        -0.0125, -0.0566,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1388, -0.1289, -0.0353, -0.0925, -0.0058, -0.0477, -0.0823, -0.0161,
        -0.2380, -0.0180, -0.0402,  0.0455, -0.0091,  0.0787,  0.0230,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1561,  0.1374,  0.0215,  0.0125,  0.0663, -0.0124,  0.0401, -0.0251,
         0.0110,  0.1948, -0.0033,  0.0126,  0.0246,  0.0020,  0.0236,  0.0077,
         0.0109,  0.0031,  0.0267,  0.0079,  0.0004, -0.0202, -0.0015,  0.0113,
         0.0677, -0.0991,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1015,  0.0346, -0.0019,  0.0017,  0.0151, -0.0236,  0.0227, -0.0238,
         0.0019, -0.0223, -0.0722,  0.0188,  0.0021, -0.0458, -0.0574, -0.0227,
         0.0633, -0.0446,  0.0106, -0.0361,  0.1145,  0.0294,  0.0376,  0.0088,
         0.0316, -0.1551,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0485, -0.2262, -0.0368,  0.0012, -0.0295, -0.0030, -0.0070, -0.0286,
        -0.0195, -0.0274, -0.0056, -0.0166, -0.0790, -0.0093, -0.0275,  0.0238,
        -0.0198,  0.0032, -0.0250, -0.0074,  0.0017, -0.0087,  0.0016, -0.0422,
        -0.0578,  0.0225, -0.0293,  0.0121, -0.0161, -0.0065,  0.0222, -0.0270,
        -0.0341, -0.0143, -0.0394, -0.0195,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0755, -0.0893, -0.0384, -0.0310, -0.0262, -0.0112, -0.0250, -0.0704,
         0.0121,  0.0548, -0.0044,  0.0160,  0.0115,  0.0353, -0.0099, -0.0080,
        -0.0167, -0.0059,  0.0130, -0.0073,  0.0132, -0.0257, -0.0094, -0.0329,
        -0.0288, -0.0851,  0.0124, -0.0035, -0.0547, -0.0062, -0.0214, -0.0231,
        -0.0112,  0.0304,  0.0587,  0.0215,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0736,  0.2157,  0.0722,  0.0385, -0.1374, -0.0778, -0.0003,  0.0028,
        -0.0257, -0.0143,  0.0193,  0.0270,  0.0469, -0.0278, -0.0500,  0.0193,
        -0.0174,  0.0490,  0.0085, -0.0765,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0571, -0.2016, -0.0339,  0.0147, -0.0098, -0.0029, -0.0206, -0.0244,
        -0.0023,  0.0059, -0.0043, -0.0418, -0.0065, -0.0035, -0.0432, -0.0234,
        -0.0009, -0.0192,  0.0151, -0.0019, -0.0526,  0.0153, -0.0117, -0.0163,
        -0.0352,  0.0025, -0.0189, -0.0272,  0.0261, -0.0431, -0.0220, -0.0224,
        -0.1014, -0.0300,  0.0008, -0.0183, -0.0014, -0.0219,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1019, -0.0874, -0.0278, -0.1004, -0.0181, -0.0238, -0.0317, -0.0102,
        -0.0167, -0.0336, -0.0321, -0.1411,  0.0048, -0.0027,  0.0358, -0.0162,
        -0.0010,  0.0045, -0.0160, -0.0130,  0.0120,  0.0028,  0.0243,  0.0170,
        -0.0103,  0.0022, -0.0029, -0.0172, -0.0243, -0.0113, -0.0168, -0.0010,
         0.0151,  0.0118, -0.0151,  0.0153, -0.0070,  0.0318, -0.0058, -0.0241,
         0.0132], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.0709e-02, -4.4329e-01, -4.4151e-02, -3.2001e-02, -1.9083e-02,
        -3.2694e-02, -3.4398e-02, -3.9036e-02, -2.5469e-02,  1.6489e-02,
         2.5582e-03, -1.9605e-02, -1.9497e-02, -5.9235e-02,  1.4478e-02,
        -2.9357e-02,  1.1944e-03, -2.0248e-02,  9.7837e-05,  2.6160e-03,
        -3.0283e-02, -8.3510e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0907, -0.0269,  0.0019,  0.0110,  0.0084, -0.0358, -0.0455,  0.0132,
        -0.0520,  0.0089,  0.0491, -0.0521, -0.0508, -0.0362, -0.0085, -0.0460,
        -0.0298, -0.0011,  0.0096, -0.0303, -0.0833, -0.0119, -0.0031, -0.0161,
        -0.0216,  0.0068, -0.0099, -0.0328, -0.1408, -0.0032, -0.0190,  0.0021,
        -0.0302, -0.0002, -0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2084, -0.0730, -0.0619, -0.0083, -0.1013,  0.0134, -0.0159,  0.0099,
        -0.0300, -0.0098, -0.0677, -0.0089, -0.0038, -0.0057, -0.0050, -0.0175,
         0.0402,  0.0266, -0.0606, -0.0041, -0.0359, -0.0308, -0.0239, -0.0187,
        -0.0083,  0.0075, -0.0650, -0.0218,  0.0161,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0818, -0.1753, -0.0395, -0.0099,  0.0047, -0.0274,  0.0026, -0.0266,
        -0.0248, -0.0131, -0.0088, -0.0602, -0.0141, -0.0574, -0.0639, -0.0664,
        -0.0383, -0.0119, -0.0084,  0.0244, -0.0454, -0.0355, -0.0301,  0.0236,
        -0.1057,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1946, -0.0059,  0.0156, -0.0212, -0.1841, -0.0445, -0.0378, -0.0862,
        -0.0427, -0.0324,  0.0010,  0.0027,  0.0294,  0.0088, -0.0261, -0.1295,
        -0.0349, -0.0146, -0.0113, -0.0134, -0.0215,  0.0284,  0.0132,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0900,  0.0508,  0.0055,  0.0132,  0.0249,  0.0428, -0.0373, -0.0352,
        -0.0272, -0.0045,  0.0181, -0.0375, -0.0153, -0.0761, -0.0080, -0.0328,
        -0.0105, -0.0093, -0.0291,  0.0261,  0.0221, -0.0119,  0.0106, -0.0364,
         0.0274, -0.0101, -0.1031, -0.0302,  0.0122, -0.0137, -0.0408, -0.0052,
        -0.0042, -0.0296,  0.0142,  0.0021,  0.0035, -0.0179,  0.0107,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0556, -0.1268,  0.0341, -0.1282, -0.2205, -0.0510,  0.0161, -0.1242,
        -0.0160, -0.0370, -0.1378, -0.0528,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-4.3365e-02, -3.7962e-02, -1.6405e-01,  6.5948e-03, -1.2533e-02,
        -7.0896e-02, -3.2793e-02, -6.3125e-02, -1.2370e-03, -1.9184e-02,
        -1.0113e-02,  2.5838e-02, -3.5197e-03,  2.5179e-03, -2.0914e-02,
         1.0287e-02, -1.4254e-02, -3.5436e-02, -7.2056e-03, -3.8323e-02,
        -4.0689e-03, -2.2660e-02,  4.0124e-03,  5.1336e-03, -1.3408e-02,
         1.1370e-02, -2.6363e-02,  2.7610e-05,  8.3131e-03,  4.7036e-02,
         1.6527e-02, -1.5901e-02, -8.7500e-03, -6.8114e-03, -1.4380e-02,
        -2.1683e-02, -1.7903e-02,  3.6417e-03, -1.7342e-02, -1.7812e-04,
         4.0692e-03,  5.7874e-03,  2.6256e-02, -7.5378e-03, -7.0687e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1115, -0.3482, -0.0423, -0.0932,  0.0163,  0.0810,  0.0197,  0.0382,
        -0.1668,  0.0337, -0.0252, -0.0239,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2003, -0.0730,  0.4281, -0.0137, -0.0736, -0.1358, -0.0718, -0.0038,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.7901e-02, -2.1191e-01, -1.1503e-02, -7.2422e-03, -6.0934e-03,
         9.6391e-03,  1.1093e-02, -8.5661e-03, -5.2241e-03,  3.5349e-02,
        -1.6324e-03, -7.1510e-04, -5.7105e-03,  4.3845e-03,  2.0891e-03,
         1.7694e-02, -5.2656e-04, -3.7668e-03, -3.4415e-02,  1.2620e-04,
        -1.8829e-03,  3.8135e-03, -9.3716e-03, -1.5185e-02, -3.6376e-03,
        -2.4726e-03, -4.6602e-03, -7.9188e-03, -1.7350e-02,  2.4925e-03,
        -1.1603e-02, -1.6554e-02, -4.0516e-03,  4.3184e-03,  9.3572e-03,
        -3.7577e-03, -1.0943e-02, -3.0310e-03,  3.1523e-03, -1.9452e-03,
        -1.2754e-03, -1.7370e-03, -5.5559e-03,  1.1523e-04, -4.8072e-02,
         6.1226e-03, -1.1349e-01, -1.5290e-02, -2.3968e-02, -7.9091e-02,
        -9.4703e-03,  4.9065e-02, -1.0914e-02,  1.4819e-03, -1.1696e-02,
        -6.2857e-03, -2.3608e-02,  1.2466e-03, -1.2265e-02, -5.7955e-03,
        -1.0375e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0681, -0.0702, -0.0064, -0.0397, -0.0709, -0.0092, -0.0464, -0.0146,
         0.0222, -0.0428, -0.0341, -0.0123, -0.0110, -0.0092, -0.0173, -0.0091,
         0.0493, -0.0118, -0.0148,  0.0276, -0.0342, -0.0061, -0.0496, -0.0289,
        -0.0216, -0.0365, -0.0088, -0.0038, -0.0108, -0.0444, -0.0326, -0.0434,
        -0.0922,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1039,  0.0535,  0.0397,  0.0319,  0.0182,  0.0231,  0.0192,  0.0027,
         0.0162,  0.0166, -0.0116,  0.0117,  0.0141,  0.0161,  0.0304, -0.0603,
         0.0050,  0.0430,  0.1881, -0.0078, -0.0140,  0.0480, -0.0250,  0.1998,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1613, -0.0771, -0.0656, -0.0182, -0.0140, -0.0906, -0.0008, -0.0170,
         0.0206, -0.0684, -0.0413, -0.0349, -0.1077, -0.0095,  0.0237, -0.0237,
         0.0092, -0.0011,  0.0086,  0.0042, -0.0270,  0.0331,  0.0981,  0.0442,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0818, -0.0929, -0.0566, -0.0542,  0.0226,  0.0097,  0.0279, -0.0217,
        -0.0007, -0.0181, -0.0175,  0.0253,  0.0066, -0.0048, -0.0045, -0.0433,
        -0.0272, -0.0134,  0.0023, -0.0645, -0.0089,  0.0110,  0.0659, -0.0485,
        -0.0073, -0.0183, -0.0108, -0.0097,  0.0123, -0.0056, -0.0429, -0.0046,
         0.1335, -0.0251,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0343, -0.0400, -0.0274,  0.0347, -0.0648,  0.0198, -0.0009,  0.0375,
         0.0025, -0.0237,  0.0161, -0.0449,  0.0183, -0.2125, -0.0179, -0.1087,
        -0.0456, -0.0417, -0.0013, -0.0106, -0.0152, -0.0985, -0.0571, -0.0028,
        -0.0231,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0943, -0.0185, -0.0876, -0.0218, -0.0005, -0.0259, -0.0486, -0.0305,
        -0.0268, -0.3155,  0.0981, -0.0186, -0.0439,  0.0356,  0.0856, -0.0483,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0042,  0.0858,  0.0237,  0.0440,  0.0278,  0.0710,  0.2865,  0.0344,
         0.0076,  0.0247,  0.0019,  0.0038,  0.0611,  0.0196,  0.0379,  0.0040,
         0.0407,  0.0341,  0.0373,  0.0179,  0.0110, -0.1042,  0.0167,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0538, -0.0572,  0.0346,  0.1112,  0.1909,  0.1711, -0.0759, -0.0283,
        -0.0459, -0.0224,  0.0236,  0.0403,  0.0298, -0.1151,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.1175, -0.0654, -0.0137, -0.1133, -0.0373,  0.1010, -0.0338,  0.0370,
        -0.0372, -0.0209, -0.0354, -0.0398, -0.0198, -0.0050, -0.0029,  0.0238,
         0.0158, -0.1264, -0.0052, -0.0062, -0.0244, -0.0153, -0.0290,  0.0740,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1295, -0.2313, -0.0022, -0.0162, -0.0398, -0.0615, -0.2695,  0.0238,
        -0.0389,  0.0190, -0.1685,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1327, -0.0375, -0.0137, -0.0536,  0.0008, -0.1180, -0.1496, -0.0257,
        -0.0773, -0.0433, -0.0268, -0.0155, -0.0343,  0.0011, -0.0419, -0.0404,
         0.0020, -0.1102, -0.0539,  0.0150,  0.0065,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.3540, -0.0934, -0.0097,  0.0387, -0.0112, -0.0236, -0.0678, -0.0893,
         0.0567, -0.0650,  0.0170, -0.0670, -0.0179, -0.0886,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2472, -0.0504,  0.0125,  0.0538,  0.0134,  0.0024,  0.0216,  0.0466,
         0.0110, -0.0330, -0.1052, -0.0190, -0.1129, -0.0126, -0.0305, -0.0145,
         0.0508, -0.0046, -0.0029, -0.0477, -0.1074,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0710, -0.1351, -0.0760, -0.1017,  0.0343, -0.0582, -0.0230, -0.2074,
        -0.0839, -0.0357, -0.0924, -0.0062, -0.0749,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0247, -0.0355,  0.0077, -0.0018,  0.0017,  0.0126, -0.0117, -0.0158,
        -0.0331,  0.0026, -0.0267,  0.0074,  0.0154,  0.0112, -0.0136, -0.0093,
        -0.0039,  0.0073,  0.0077,  0.0270, -0.0344, -0.0172,  0.0015,  0.0549,
        -0.1561,  0.0222, -0.0381, -0.0639, -0.0365, -0.0208, -0.0284,  0.0352,
         0.0086,  0.0182,  0.0179, -0.0135, -0.0194, -0.0035,  0.0119,  0.0026,
        -0.0111,  0.0140, -0.0080,  0.0542,  0.0313], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0316, -0.1240, -0.0411, -0.0284,  0.0017,  0.0106, -0.0197, -0.0101,
        -0.0202, -0.0399, -0.0776,  0.0119,  0.0057,  0.0067, -0.0236, -0.0032,
        -0.1228, -0.0005,  0.0287,  0.0033, -0.0280, -0.0266, -0.0259, -0.0531,
        -0.0141, -0.0407, -0.0061, -0.0492, -0.0438, -0.0058, -0.0157, -0.0239,
        -0.0051, -0.0506,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0542, -0.1794, -0.0445, -0.0133, -0.0251, -0.0155, -0.0251, -0.0168,
         0.0485, -0.0290, -0.0103, -0.0080, -0.0051, -0.0009,  0.0043,  0.0228,
         0.0055,  0.0268, -0.0047, -0.0768, -0.1839, -0.0180, -0.0265,  0.0437,
        -0.0135,  0.0978,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0568, -0.2360,  0.0424, -0.0445,  0.0232, -0.0724,  0.0793, -0.0141,
         0.0051, -0.0265, -0.0052, -0.0062,  0.0146,  0.0140,  0.0186, -0.0199,
        -0.0831,  0.0304,  0.0230,  0.0263, -0.0139,  0.1388,  0.0056,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0223, -0.0221,  0.0058, -0.0424,  0.0061, -0.0177, -0.0212, -0.0240,
        -0.0453, -0.0161,  0.0160, -0.0022,  0.0163, -0.0278, -0.1426,  0.0733,
         0.0011, -0.0497, -0.0318, -0.0276,  0.0291, -0.0112, -0.0121,  0.0086,
         0.0055, -0.0025, -0.0277, -0.0934,  0.0244,  0.0161,  0.0150, -0.0042,
         0.0025, -0.0224,  0.0465,  0.0210, -0.0066,  0.0211, -0.0186,  0.0003,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0040, -0.0707, -0.0164, -0.0125, -0.0592, -0.0165, -0.0590, -0.0677,
        -0.0177, -0.0272, -0.0068,  0.0021, -0.0199, -0.0216, -0.0188, -0.0225,
        -0.0242,  0.0282,  0.0216, -0.0551, -0.0345, -0.0170,  0.0025,  0.0135,
        -0.0060, -0.0396,  0.0069, -0.0243, -0.0133, -0.0336, -0.0446,  0.0046,
         0.0378, -0.1501,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0533, -0.0389, -0.0174,  0.0131, -0.0032, -0.0049, -0.0192,  0.0056,
         0.0025,  0.0300, -0.0064, -0.0111, -0.0154,  0.0206, -0.1071, -0.0132,
         0.0136, -0.0089, -0.0404, -0.0059, -0.0162,  0.0013,  0.0122, -0.0027,
        -0.0433, -0.0922, -0.0887, -0.0036, -0.0057, -0.0202,  0.0353, -0.0076,
         0.0279, -0.0136, -0.0166,  0.0006, -0.0158, -0.0272, -0.0123, -0.0236,
        -0.0133, -0.0093, -0.0035, -0.0034, -0.0063, -0.0040,  0.0003,  0.0056,
        -0.0109, -0.0175,  0.0238, -0.0047], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0619, -0.0269, -0.0132,  0.0210,  0.0747,  0.0036,  0.0488, -0.0623,
         0.0181,  0.0350,  0.0234, -0.0173,  0.0021,  0.0581,  0.0555,  0.0383,
         0.0624, -0.0479,  0.0355, -0.0078,  0.0261, -0.0022,  0.0271,  0.0662,
        -0.0021,  0.0002,  0.0160, -0.0046,  0.0211,  0.0711,  0.0495,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0722,  0.0341,  0.0159, -0.0069,  0.0104,  0.0224,  0.0109, -0.0325,
         0.0196, -0.0199,  0.0268,  0.0421,  0.0562, -0.0216,  0.0034,  0.0329,
        -0.1058,  0.0211,  0.1266,  0.1157,  0.0359,  0.0393,  0.0596,  0.0028,
        -0.0654,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1003, -0.1888,  0.0038,  0.0795,  0.0156,  0.0704,  0.2162, -0.0719,
         0.0395,  0.0654, -0.1125,  0.0361,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0750,  0.2784,  0.0084,  0.0299,  0.0439,  0.0157, -0.0063, -0.0040,
        -0.0023,  0.0627,  0.0180,  0.0018,  0.0137,  0.0760,  0.0410,  0.0497,
         0.0243,  0.0278, -0.0052, -0.0103,  0.0115,  0.0236,  0.0179, -0.0168,
         0.0164,  0.0543,  0.0347, -0.0224, -0.0079,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0164, -0.3715,  0.0163, -0.1716,  0.0545, -0.0566, -0.1889,  0.0121,
        -0.0076,  0.0139,  0.0316, -0.0589,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.2385e-02, -3.8733e-01, -1.5001e-02, -1.0111e-01, -1.4722e-02,
        -4.0453e-02, -1.7858e-01, -1.1566e-01, -5.5509e-02, -1.8311e-02,
         1.2370e-04, -5.0806e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1362, -0.0501, -0.1169, -0.0416, -0.0503,  0.0298, -0.0366, -0.0593,
        -0.0669, -0.0123, -0.0310, -0.0970, -0.0039,  0.0416, -0.0150, -0.0602,
        -0.0254, -0.0246,  0.0704,  0.0312,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0805,  0.1799,  0.0158,  0.0084,  0.1765, -0.0027,  0.0549, -0.0411,
         0.0127, -0.0008, -0.0201,  0.0008,  0.0167, -0.0010, -0.0183,  0.0022,
         0.0092,  0.0026,  0.0018,  0.0257,  0.0230, -0.0105,  0.0097, -0.0290,
        -0.0057, -0.0227,  0.0117,  0.0267,  0.0015, -0.0072,  0.0027, -0.0032,
         0.0165,  0.0838,  0.0211, -0.0465, -0.0070,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0508,  0.0478,  0.2990,  0.0335,  0.0895,  0.0117,  0.0965,  0.1032,
        -0.2681,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1523, -0.0230, -0.0002, -0.0652, -0.0296, -0.0918, -0.0758, -0.0214,
        -0.0075, -0.0106, -0.0022, -0.0769, -0.0747, -0.0478,  0.0025,  0.0413,
        -0.0181,  0.0525, -0.0529, -0.0578, -0.0410,  0.0012, -0.0008,  0.0075,
        -0.0230,  0.0143, -0.0081,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0383,  0.0947,  0.0450,  0.0242,  0.0170,  0.0220,  0.0100,  0.0089,
        -0.0128,  0.0759, -0.0480,  0.0197, -0.0045,  0.0023,  0.0025,  0.0154,
         0.0194, -0.0065,  0.0169, -0.0280, -0.0927,  0.0251,  0.0262,  0.0199,
         0.0173,  0.0498,  0.0120,  0.0189,  0.1592, -0.0059,  0.0083,  0.0199,
        -0.0168, -0.0158,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 1.2944e-01,  5.6461e-04, -1.3248e-02,  1.7148e-02, -1.6413e-02,
        -1.0322e-03,  5.5995e-03,  1.3994e-02, -3.3116e-05,  7.4997e-04,
        -5.6959e-03, -8.7593e-02, -3.3225e-02, -1.5485e-03, -2.5012e-02,
        -8.5492e-03, -3.2887e-02,  4.8250e-02, -2.0293e-02,  4.5406e-04,
         1.7347e-02,  1.5699e-02, -2.1141e-02,  2.8903e-03, -2.0582e-02,
        -1.4386e-02, -2.4181e-02, -3.4800e-02, -4.0815e-05, -9.1632e-04,
        -2.4948e-02, -5.3581e-02, -5.3433e-02,  4.2594e-03,  4.1299e-02,
        -2.1527e-02, -3.7638e-02, -1.0449e-02,  9.2919e-03,  4.8147e-03,
        -1.3872e-02, -6.5913e-02,  1.1721e-02, -3.3485e-02,  5.5774e-05,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0889,  0.0139,  0.0441, -0.0119,  0.1464,  0.1096,  0.1285,  0.0115,
         0.0757,  0.1085,  0.0345,  0.0056,  0.0828, -0.0516,  0.0069,  0.0797,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1631,  0.0455, -0.0459, -0.0706, -0.0108, -0.0381, -0.1064,  0.0223,
        -0.0492, -0.0474, -0.0213, -0.0481, -0.0538,  0.0169, -0.0133, -0.0033,
        -0.0078,  0.0175, -0.0091, -0.0030, -0.0026, -0.0157, -0.0301, -0.0596,
         0.0217,  0.0382,  0.0388,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.2992e-01,  1.0469e-03, -2.9483e-02, -2.4829e-06, -6.9217e-03,
         8.6666e-03, -1.3463e-02, -4.4035e-03, -2.0741e-02, -6.2308e-04,
        -1.4203e-02, -2.5057e-02, -2.3288e-02, -2.6687e-03,  2.0535e-03,
         7.0530e-03, -1.2413e-02,  2.3538e-03,  3.2860e-02, -3.4703e-02,
         2.0274e-02, -1.5429e-02, -2.8347e-03, -1.2528e-01, -9.2704e-03,
         2.0728e-03, -3.1025e-03,  7.3651e-03,  6.3009e-04, -1.7865e-02,
         5.9743e-03, -6.2349e-03, -1.2789e-02,  2.8677e-03,  1.6853e-03,
         1.9656e-03, -4.0897e-03, -7.7502e-02, -1.8345e-02,  6.9246e-03,
        -1.4492e-02, -5.2698e-02, -7.6536e-03, -1.4310e-02,  4.9816e-02,
         4.6606e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0514,  0.0412, -0.0300, -0.0458, -0.0178, -0.0731, -0.0364, -0.0144,
         0.0004,  0.0093, -0.0550, -0.0483, -0.0238, -0.0315, -0.0715, -0.0147,
        -0.0553, -0.0214, -0.0162, -0.0671, -0.0345,  0.0253,  0.0083, -0.0132,
        -0.0319, -0.0061, -0.0393,  0.0021, -0.0242,  0.0908,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1029,  0.1945,  0.1506,  0.0892,  0.0151,  0.0599, -0.0112,  0.1283,
         0.0976, -0.0724,  0.0784,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0721, -0.2105, -0.0452,  0.0323, -0.0458, -0.0489, -0.0452, -0.1600,
        -0.0023, -0.0763, -0.0113,  0.0056, -0.0254,  0.0024,  0.0449, -0.0444,
        -0.0028, -0.0286, -0.0151, -0.0807,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0088, -0.2300, -0.0549, -0.1104,  0.0641,  0.0023, -0.2952, -0.0042,
         0.0514,  0.1097,  0.0691,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0231, -0.2134, -0.0510, -0.1074, -0.0338,  0.0138, -0.0162, -0.0449,
        -0.0531, -0.0254, -0.0364, -0.0056, -0.0302,  0.0215, -0.0317, -0.0621,
         0.0651,  0.0133,  0.0136, -0.0621,  0.0762,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0935, -0.0780, -0.0358,  0.0328, -0.0020, -0.4164,  0.0043, -0.0122,
         0.0037,  0.0119, -0.0251,  0.0079,  0.0426,  0.0204,  0.0258,  0.0744,
         0.0810, -0.0325,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.7386e-03, -4.7471e-02, -2.2297e-01,  1.5976e-02, -1.4549e-02,
        -1.8978e-02, -1.6860e-02, -1.5217e-02, -1.1577e-02, -1.2207e-04,
        -1.4483e-02, -1.2106e-03, -6.1119e-03, -3.2596e-02, -5.0612e-02,
        -1.1423e-02, -2.6984e-02, -3.0231e-02, -2.6131e-02,  3.1046e-02,
         7.4028e-03, -3.1705e-02, -7.5321e-02, -1.2535e-02, -2.4272e-02,
         6.8128e-03, -2.6218e-02, -2.9095e-02, -1.9878e-02, -4.7044e-02,
         1.0486e-02, -2.5406e-02, -2.2876e-02,  2.0132e-02, -7.6300e-03,
        -7.4270e-04, -1.5871e-02,  4.6895e-03, -8.5961e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.8852e-02,  2.7146e-01, -6.4116e-02, -4.7339e-02, -3.0782e-02,
         2.9426e-03, -8.3475e-03,  2.0187e-02, -3.7621e-02, -1.4319e-02,
        -1.3909e-02,  9.0372e-03, -1.5875e-02, -2.3310e-03, -3.2547e-02,
        -1.9937e-03, -5.1604e-03, -2.3753e-03, -1.3590e-02,  2.6576e-03,
        -2.6615e-03,  2.7645e-03,  3.7726e-04,  1.2509e-02, -1.1516e-02,
        -7.4116e-03,  4.5778e-03, -8.6615e-03,  2.8633e-02,  9.1278e-02,
         1.5625e-02,  2.3949e-04, -1.0631e-01,  5.1991e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.2562, -0.1486, -0.1442, -0.1053, -0.0975, -0.1316, -0.0552,  0.0068,
        -0.0545,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0201, -0.0868, -0.0468, -0.0086, -0.0276, -0.0053, -0.0090,  0.0094,
         0.0242, -0.0019, -0.0441, -0.0081,  0.0089, -0.0017,  0.0225, -0.0048,
         0.0234, -0.2427, -0.0292,  0.0278,  0.0048,  0.0110,  0.0518, -0.0023,
        -0.0211, -0.0090, -0.0146,  0.0323, -0.0104, -0.0031,  0.0053, -0.0365,
        -0.0022, -0.0249, -0.0135, -0.0076,  0.0088, -0.0104,  0.0501,  0.0270,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.5537e-02, -4.5304e-02, -9.6720e-02, -2.1465e-02, -3.2885e-02,
         7.1959e-03, -1.4596e-02, -5.2817e-03, -8.3763e-03, -6.2001e-02,
        -6.5180e-03,  1.4525e-03, -2.0939e-03,  6.1400e-03, -2.0578e-04,
         1.7958e-02,  5.3266e-03, -2.3219e-02, -3.3044e-02, -1.3059e-02,
        -7.4885e-03, -2.5021e-02,  1.7301e-02, -1.8138e-02,  3.0375e-02,
        -5.1367e-03, -1.2996e-02, -3.8095e-02, -1.1059e-02, -6.7770e-02,
        -3.6045e-02, -2.9689e-02, -4.6601e-03, -1.8235e-02, -8.0801e-03,
        -3.1242e-03, -3.5209e-02, -4.8098e-02, -5.8179e-03, -4.6180e-04,
         1.3140e-02, -5.6293e-03, -1.4121e-02, -6.3842e-04,  6.6466e-03,
        -7.6404e-05,  4.9469e-04,  9.3548e-03,  9.5379e-03, -8.1592e-03,
        -2.1673e-02, -4.9351e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0638, -0.1460, -0.0520, -0.0121, -0.0354,  0.0083, -0.0403,  0.0144,
        -0.0083,  0.0240, -0.0729, -0.0796, -0.0174, -0.0032, -0.0451, -0.0847,
        -0.0421,  0.0838, -0.0489, -0.0416, -0.0214, -0.0546,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.1823e-02, -3.7177e-01, -4.3058e-02, -3.6842e-02, -8.7288e-02,
        -2.1373e-02, -4.1668e-02,  6.3998e-03, -1.2820e-02, -8.9636e-02,
         9.0488e-03, -3.4391e-02,  6.2049e-03, -1.9960e-02, -3.3472e-04,
        -2.8006e-02, -4.6786e-02, -1.2519e-02,  2.7865e-02, -3.6489e-02,
         1.5720e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0589, -0.0505,  0.0087, -0.0058,  0.0312,  0.0030, -0.0323,  0.0029,
        -0.0157,  0.0051, -0.0058,  0.0037, -0.0013,  0.0066, -0.0032, -0.0060,
        -0.0125,  0.0078, -0.0089, -0.0069, -0.0040,  0.0130,  0.0046,  0.0126,
        -0.0108,  0.0143, -0.0167, -0.0205,  0.0028,  0.0054, -0.0174, -0.0169,
        -0.0089, -0.0103, -0.0028, -0.0116, -0.0349, -0.0136, -0.0076, -0.0166,
        -0.1663, -0.0004, -0.0321, -0.0291, -0.0235, -0.0067, -0.0333, -0.0485,
        -0.0018, -0.0026, -0.0270, -0.0232,  0.0036, -0.0523, -0.0277],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1741, -0.0320,  0.0203, -0.0094, -0.0010, -0.0312, -0.0099, -0.0003,
         0.0071,  0.0070,  0.0470,  0.0085,  0.0041, -0.0082,  0.0034,  0.0036,
         0.0065, -0.0199,  0.0138, -0.0029,  0.0133, -0.0181, -0.0084,  0.0027,
         0.0150,  0.0135, -0.0090, -0.1302, -0.0332, -0.0339, -0.0128, -0.0032,
        -0.0644, -0.0721, -0.0163, -0.0060,  0.0035, -0.0293, -0.0022, -0.0115,
        -0.0196,  0.0123,  0.0240, -0.0248, -0.0107,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1064,  0.1738,  0.0615,  0.0972,  0.0286,  0.0074,  0.0127, -0.0155,
         0.0268, -0.0029,  0.0009,  0.0283,  0.0228, -0.0002,  0.0237, -0.0375,
         0.0166,  0.0939, -0.0216,  0.0168,  0.1319, -0.0062, -0.0530,  0.0139,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0942, -0.0575, -0.0235, -0.0248, -0.0224,  0.0104,  0.0257, -0.0192,
        -0.0076, -0.0017, -0.0084,  0.0061, -0.0088, -0.0252, -0.0728, -0.0147,
        -0.0632, -0.0420, -0.0137, -0.0087,  0.0400, -0.0033, -0.0286, -0.0534,
         0.0811, -0.0218, -0.0512, -0.0015, -0.0091, -0.0988, -0.0092, -0.0184,
         0.0330,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0096, -0.3303, -0.0531, -0.0324, -0.0378,  0.0164, -0.0847,  0.0237,
        -0.1219, -0.0154,  0.0194,  0.0033,  0.1079, -0.0578, -0.0230, -0.0187,
        -0.0070, -0.0335, -0.0041,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0473,  0.0508,  0.0707,  0.0677,  0.0507,  0.0106,  0.0613,  0.0297,
         0.1090,  0.0005, -0.0413, -0.0041, -0.0084, -0.0328, -0.0013, -0.0097,
         0.0475,  0.0379,  0.0115, -0.0193,  0.0206,  0.1194,  0.0525, -0.0544,
        -0.0409,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2110,  0.0845,  0.0450,  0.0606,  0.0249,  0.0539,  0.2938, -0.0109,
        -0.0150,  0.0310,  0.0150,  0.0353,  0.0226,  0.0361,  0.0321, -0.0282,
         0.0003,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0937,  0.0041, -0.0332, -0.0131,  0.0032, -0.0009,  0.0275, -0.0657,
         0.0346, -0.0235, -0.0075,  0.0293, -0.0039, -0.0206, -0.0392, -0.0456,
        -0.0209, -0.1199,  0.0343, -0.0736, -0.0277, -0.0607, -0.0687, -0.0084,
        -0.0017, -0.0115, -0.0334, -0.0278,  0.0010,  0.0045,  0.0458, -0.0145,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1222, -0.1309, -0.0588, -0.0388, -0.0308, -0.0087, -0.0161, -0.0405,
        -0.0143, -0.0045,  0.0216,  0.0165,  0.0210, -0.0346, -0.0334,  0.0390,
        -0.0136,  0.0111, -0.0669, -0.0205, -0.0265, -0.0024, -0.0753, -0.1213,
        -0.0097, -0.0112,  0.0059,  0.0039,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0587, -0.1458,  0.0191, -0.0299,  0.0155, -0.0212, -0.0898, -0.0505,
        -0.0164, -0.0401, -0.0223, -0.0717,  0.0068,  0.0169, -0.0434, -0.0206,
        -0.0144, -0.0374, -0.0115,  0.0105, -0.0071, -0.0169,  0.0154,  0.0075,
        -0.0100,  0.0469, -0.0091, -0.1086,  0.0364,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0787,  0.0075, -0.0288, -0.0282, -0.0902, -0.0255,  0.0455,  0.0041,
         0.0175,  0.0275, -0.0118, -0.0242, -0.0551, -0.0842, -0.0281, -0.0384,
        -0.0272, -0.0869, -0.0134, -0.0318,  0.0034,  0.0135, -0.0008, -0.0344,
        -0.0290,  0.0060, -0.0205, -0.0354, -0.0243,  0.0019,  0.0009,  0.0692,
        -0.0062,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0074, -0.1414, -0.0297, -0.0606, -0.0299, -0.0531, -0.0248, -0.0326,
        -0.0210, -0.0274, -0.0282, -0.0188, -0.0265, -0.0206,  0.0006,  0.0031,
        -0.0340, -0.0136, -0.1162, -0.0498, -0.1262,  0.0426,  0.0698, -0.0220,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0809, -0.1195, -0.0493, -0.0087,  0.0226, -0.0870, -0.0658, -0.0586,
        -0.0178, -0.0194, -0.0092, -0.0337, -0.0233, -0.0118,  0.0069, -0.0294,
         0.0067, -0.0231, -0.0038, -0.2070, -0.0537,  0.0025,  0.0592,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0887, -0.0733,  0.0048,  0.0313, -0.0056, -0.0169, -0.0212, -0.0124,
         0.0141, -0.0234,  0.0433, -0.0088, -0.0133,  0.0042, -0.0047,  0.0211,
        -0.0484, -0.0579, -0.0111, -0.0325,  0.0068, -0.0191, -0.0057, -0.0135,
        -0.0420, -0.0080, -0.0009, -0.0123, -0.0044, -0.0314, -0.1766, -0.0008,
        -0.0135,  0.0221, -0.0132, -0.0188, -0.0048,  0.0524,  0.0171,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0469, -0.1260, -0.0006, -0.0212, -0.1014, -0.0086,  0.0243,  0.0127,
        -0.0032,  0.0253, -0.0005, -0.0206,  0.0190,  0.0022,  0.0379, -0.0198,
         0.0092, -0.0048, -0.0060,  0.0164, -0.0213, -0.0164, -0.0166, -0.0272,
        -0.0346, -0.0221, -0.0143, -0.0088, -0.0099, -0.0226, -0.0079, -0.0064,
        -0.0451,  0.0052,  0.0034,  0.0026, -0.0211, -0.0781, -0.0157, -0.0128,
        -0.0207, -0.0547,  0.0258,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.2373e-02, -2.0116e-01, -1.2767e-03, -1.3094e-02,  5.7759e-03,
        -1.5172e-02, -3.4706e-02, -1.3903e-02, -6.4450e-02,  2.5351e-02,
        -4.1239e-03,  2.7054e-03, -1.2284e-02,  2.6216e-03,  3.5749e-03,
         1.6895e-02,  2.8085e-03, -4.2043e-02,  4.5439e-04, -4.5340e-03,
         2.5463e-03, -2.3417e-02, -3.4872e-02, -2.4017e-02, -1.2351e-02,
        -1.7650e-02, -7.9153e-03, -1.2873e-03, -8.9897e-03,  7.9443e-03,
        -1.3801e-04, -5.0529e-03, -2.2257e-02,  4.2552e-03,  3.9757e-03,
         1.4074e-02,  4.9004e-02, -1.7314e-02, -1.7306e-02, -6.9396e-03,
        -9.7146e-03,  6.8358e-04, -7.3019e-03,  2.2704e-02, -3.1014e-03,
        -2.1910e-02, -6.9442e-03, -4.0732e-03,  4.0201e-03, -1.7432e-02,
        -1.3941e-02, -1.0759e-03,  2.1882e-02,  7.1336e-03,  3.2183e-03,
        -1.0528e-02, -4.3726e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0867, -0.3502, -0.0581, -0.0172, -0.0213,  0.0077, -0.0509, -0.1015,
        -0.0477, -0.0341, -0.0327, -0.0361,  0.0540, -0.1019,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1978, -0.0902, -0.0242, -0.0232,  0.0063, -0.0022, -0.0067, -0.0063,
         0.0103,  0.0070, -0.0220, -0.1324, -0.0502,  0.0009, -0.0271, -0.0058,
        -0.0012, -0.0374, -0.0031, -0.0134, -0.1240, -0.0419, -0.0207, -0.0007,
         0.0372,  0.0021,  0.0259,  0.0110,  0.0349,  0.0342,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.5113e-01, -1.4261e-01, -3.9784e-04, -9.6528e-02,  7.0695e-02,
         1.8466e-02, -6.0669e-02, -1.5950e-01,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0972, -0.0395,  0.0058, -0.0136, -0.0088, -0.0899,  0.0147,  0.0208,
         0.0156, -0.0042, -0.0126,  0.0076, -0.0382, -0.0883, -0.0116, -0.0028,
         0.0101, -0.0193, -0.0155, -0.0258,  0.0080, -0.0035, -0.0170, -0.0549,
        -0.0513, -0.0224, -0.0190,  0.0852, -0.0391, -0.0833,  0.0377, -0.0126,
         0.0239,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0652,  0.0275,  0.0208, -0.0053,  0.0106,  0.0332,  0.0041, -0.0161,
         0.0163,  0.0117,  0.0035,  0.0069,  0.0124, -0.0028, -0.0045, -0.0131,
         0.0115,  0.0058,  0.0156,  0.0044,  0.0080, -0.0063,  0.0088,  0.0032,
        -0.0010,  0.0031,  0.0136,  0.0469, -0.0076,  0.0132, -0.0044,  0.0060,
         0.0142,  0.0636,  0.0114,  0.0133,  0.0113,  0.0054,  0.0011, -0.0129,
         0.0015, -0.0020,  0.0029,  0.0271, -0.0148, -0.0216,  0.0059,  0.0026,
         0.0316,  0.0160,  0.0008,  0.0233,  0.0029,  0.0086,  0.0035, -0.0121,
         0.0198,  0.0086,  0.0111,  0.0335,  0.0269, -0.0008,  0.0670,  0.0160,
         0.0005,  0.0010, -0.0020, -0.0258,  0.0043, -0.0277, -0.0314],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0875, -0.0647, -0.0057,  0.0150, -0.0430,  0.0202, -0.2121, -0.0032,
        -0.0119, -0.0145, -0.0043, -0.0098, -0.0156,  0.0047, -0.0263, -0.0065,
         0.0041,  0.0108, -0.0228, -0.0183,  0.0025,  0.0014,  0.0575, -0.0561,
         0.0083, -0.0055, -0.0223, -0.0126,  0.0026,  0.0090, -0.0129, -0.0103,
        -0.0128, -0.0319, -0.0280, -0.0325, -0.0258, -0.0056, -0.0011, -0.0004,
        -0.0484,  0.0116,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0719, -0.2879, -0.0443, -0.0044,  0.0063,  0.0410, -0.0029, -0.0087,
        -0.0123, -0.0231, -0.0327, -0.0077, -0.0426,  0.0163, -0.0183,  0.0091,
        -0.0029, -0.0256, -0.0704, -0.0176, -0.0084, -0.0111, -0.0123, -0.0122,
         0.0044, -0.0200, -0.0158,  0.0033, -0.0319,  0.0161,  0.0113,  0.0023,
         0.0191, -0.0214, -0.0228, -0.0240,  0.0177,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1497, -0.3605,  0.0615, -0.0483, -0.0222,  0.0393, -0.0604, -0.0085,
        -0.0253, -0.0236, -0.0458,  0.0088, -0.0366,  0.0903,  0.0192,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0645,  0.0936,  0.0501,  0.0358,  0.0598,  0.0326,  0.0610,  0.0326,
         0.1144,  0.0054,  0.0473, -0.0359,  0.0790,  0.0347, -0.0467, -0.0684,
         0.0527,  0.0366, -0.0002, -0.0489,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0438, -0.0372,  0.0287,  0.0493, -0.1042, -0.0578, -0.0078, -0.0067,
        -0.0046, -0.0171, -0.0075, -0.0328, -0.0290, -0.0101, -0.0279, -0.0091,
         0.0034, -0.0034,  0.0071, -0.0204, -0.0018, -0.0141, -0.0333,  0.0133,
        -0.0045, -0.0188, -0.0558, -0.0171, -0.0104, -0.0078, -0.0512, -0.0013,
        -0.0149, -0.0295,  0.0035, -0.0259, -0.0073, -0.0061, -0.0089, -0.0290,
         0.0548,  0.0269,  0.0558,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0653, -0.1098,  0.0149, -0.0438, -0.0215, -0.0193, -0.0174, -0.1188,
        -0.0139, -0.0043, -0.0016, -0.0161, -0.0111,  0.0040, -0.0662, -0.0121,
        -0.0107,  0.0094, -0.0032,  0.0203, -0.0219, -0.0143,  0.0014, -0.0045,
        -0.0255, -0.0073,  0.0042,  0.0125, -0.0545, -0.0126,  0.0032, -0.0303,
        -0.0268, -0.0073, -0.0392,  0.0132, -0.0417,  0.0962,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.4191e-02, -1.7756e-01, -1.0193e-01, -1.7255e-01, -1.9421e-02,
        -6.3777e-02, -1.0200e-02,  6.1485e-02,  1.4296e-02, -1.0738e-01,
        -3.3148e-02, -5.1625e-02, -4.8394e-02, -1.3853e-02,  4.3324e-05,
         1.2437e-02, -7.0082e-03, -1.6330e-03, -8.0593e-03, -1.0497e-02,
         3.6271e-02, -4.2429e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0446, -0.0553,  0.0056, -0.0013,  0.0018, -0.0175, -0.0212, -0.0043,
        -0.0521, -0.0721,  0.0008, -0.0022, -0.0249,  0.0020, -0.0181, -0.0275,
        -0.1101, -0.0195, -0.0241, -0.0376,  0.0067, -0.0033, -0.0268, -0.0010,
        -0.0437, -0.0066, -0.0058, -0.0178,  0.0075, -0.0323, -0.0438, -0.0070,
        -0.0185,  0.0091, -0.0085, -0.0880,  0.0191,  0.0163, -0.0669, -0.0284,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0031, -0.0120,  0.0385,  0.0023,  0.0018, -0.0139,  0.0134, -0.0293,
        -0.0181, -0.0618, -0.0128, -0.0116, -0.0230, -0.0551, -0.0502, -0.0610,
        -0.0283, -0.0279, -0.0863, -0.0173, -0.0294, -0.0098, -0.0210, -0.0563,
        -0.0598, -0.0681,  0.0063, -0.0350, -0.0500, -0.0608, -0.0219, -0.0140,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1084, -0.1628,  0.0067,  0.0579, -0.0281, -0.0066, -0.0110, -0.0268,
        -0.0839, -0.0131, -0.0084, -0.0186, -0.0010, -0.0114, -0.0066, -0.0041,
        -0.0259, -0.0135, -0.0089,  0.0047, -0.0129,  0.0064, -0.0032, -0.0520,
        -0.0055,  0.0443, -0.0155, -0.0038,  0.0104,  0.0024,  0.0243,  0.0126,
         0.0103,  0.0870, -0.1008,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0410,  0.0681,  0.0794,  0.0992,  0.0202,  0.1360,  0.0472,  0.0739,
         0.0520,  0.0744, -0.0110,  0.0356,  0.0100,  0.0537,  0.0255, -0.0090,
         0.0355,  0.0299, -0.0158,  0.0406, -0.0420,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0508, -0.3493,  0.0245, -0.1057, -0.0326, -0.0265,  0.0161, -0.0988,
        -0.1136,  0.0039, -0.0006,  0.0151, -0.0065, -0.0117, -0.0137,  0.0185,
        -0.0335,  0.0445, -0.0342,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1242, -0.2337, -0.0586, -0.0007, -0.0140, -0.0285,  0.0072, -0.0550,
         0.0058, -0.0630, -0.1472, -0.0541, -0.0140, -0.1264,  0.0066, -0.0022,
         0.0114, -0.0185,  0.0194,  0.0096,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0930,  0.2164,  0.0524,  0.0564,  0.0352,  0.0085, -0.0175,  0.0042,
         0.0468,  0.0375,  0.0172,  0.0230,  0.0208,  0.0305,  0.0424,  0.0320,
         0.0384,  0.0174,  0.0124,  0.0557,  0.0010, -0.0726, -0.0685,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0304,  0.1463,  0.0557, -0.0037,  0.0321,  0.0117, -0.0116, -0.0081,
         0.0249,  0.1430,  0.0022,  0.0155,  0.0370,  0.1622, -0.0799, -0.0304,
        -0.0125, -0.0342,  0.0484,  0.0269,  0.0305,  0.0209, -0.0048, -0.0042,
        -0.0231,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1450e-01,  2.3103e-01,  3.5236e-02,  9.6824e-03,  4.6197e-02,
         3.7322e-02, -4.2871e-02,  8.0363e-03,  2.0897e-04,  3.4045e-02,
         5.0911e-03,  2.0742e-02,  3.6193e-03, -2.2596e-02, -1.8894e-02,
        -4.3932e-02,  2.3166e-02, -7.5522e-03,  1.3447e-02,  2.0275e-02,
        -4.3318e-03,  1.0178e-02,  2.7031e-02,  1.4016e-02,  1.6145e-02,
         3.6575e-02, -8.3133e-03, -7.0190e-03,  2.1468e-03, -1.1843e-02,
         3.7759e-03, -8.6584e-04,  9.0124e-03, -7.9923e-02, -3.0381e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0702,  0.0296, -0.0002, -0.0495, -0.0074,  0.0159, -0.0417, -0.0051,
        -0.0859,  0.0049,  0.0590, -0.0235,  0.0282, -0.0458, -0.0439, -0.0562,
         0.0316, -0.0700, -0.1517, -0.0835, -0.0528, -0.0434,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.3621e-01,  1.8180e-03,  6.3688e-02, -3.5349e-02, -9.1213e-03,
        -5.0447e-02, -2.5442e-02,  1.8805e-02,  4.4509e-03, -8.0260e-02,
        -9.1203e-02,  3.8648e-02, -1.1614e-02,  5.8704e-05, -1.1804e-02,
        -4.1127e-02, -2.6581e-02, -2.3682e-02, -2.7182e-03, -9.0251e-02,
         3.6720e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0413, -0.3736, -0.0388, -0.0077,  0.0046, -0.0600, -0.0511,  0.0092,
        -0.0385,  0.0304, -0.0341, -0.0386, -0.0389, -0.0357, -0.0079, -0.0363,
         0.0032,  0.1205, -0.0296,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1297,  0.0484, -0.0773, -0.0260,  0.0017, -0.0020, -0.0037,  0.0114,
         0.0220,  0.0025,  0.0013,  0.0207,  0.0055, -0.0310,  0.0054, -0.0123,
        -0.0020, -0.0578, -0.0036, -0.0210,  0.0199, -0.0050,  0.0295, -0.0106,
         0.0219,  0.0082, -0.0067, -0.0394, -0.0046, -0.1276, -0.0240, -0.0179,
        -0.0423, -0.0359,  0.0222,  0.0011, -0.0089, -0.0020,  0.0228, -0.0041,
        -0.0101,  0.0022, -0.0058, -0.0013,  0.0032,  0.0018,  0.0356],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0261, -0.1116, -0.0552, -0.0099, -0.0026, -0.0180,  0.0066,  0.0187,
        -0.0188,  0.0070,  0.0028,  0.0297,  0.0020,  0.0066,  0.0626, -0.0169,
        -0.0612, -0.0496, -0.0275,  0.0213,  0.0135, -0.0285, -0.0441, -0.1156,
        -0.0567,  0.0179,  0.0418,  0.0067, -0.0282,  0.0189, -0.0074,  0.0008,
        -0.0338, -0.0315,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0083, -0.3716,  0.0180,  0.0183, -0.0525, -0.0295, -0.0162,  0.0172,
        -0.0113, -0.0034,  0.0177,  0.0099, -0.0427, -0.0004,  0.0123,  0.0616,
         0.0042, -0.0034, -0.0204, -0.0350, -0.0047,  0.0073,  0.0053,  0.0039,
         0.0011, -0.0201, -0.0382, -0.0223, -0.0029, -0.0086,  0.0068, -0.0341,
         0.0519, -0.0389,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0739, -0.3359, -0.0255,  0.0130, -0.0017, -0.0300,  0.0072,  0.0169,
         0.0154, -0.1107, -0.0688,  0.0377, -0.0399, -0.0979, -0.0330, -0.0391,
        -0.0285, -0.0249,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0215, -0.0497, -0.0814,  0.0010, -0.0105,  0.0386, -0.0205,  0.0077,
        -0.0304, -0.1461,  0.0212,  0.0695, -0.0031,  0.0257,  0.0243, -0.0167,
        -0.0166,  0.0183,  0.0327, -0.1481, -0.0473,  0.0225, -0.0409, -0.0180,
         0.0120,  0.0041, -0.0717,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0478, -0.0448, -0.1011,  0.0071, -0.0327,  0.0039, -0.0039, -0.0539,
        -0.1309,  0.0599,  0.0043, -0.0338, -0.0832, -0.0279, -0.0206,  0.0010,
        -0.0229, -0.0232, -0.0234, -0.2088, -0.0133, -0.0126, -0.0154,  0.0045,
        -0.0171, -0.0020,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2272,  0.1130,  0.0504,  0.0515,  0.1510,  0.0248,  0.0286,  0.0868,
         0.0247,  0.0255,  0.0301,  0.0145,  0.0040,  0.0009,  0.0224, -0.0030,
         0.0241,  0.0225,  0.0079, -0.0484, -0.0386,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0870,  0.1632,  0.1057,  0.2001,  0.0431,  0.0637,  0.0426,  0.0127,
         0.0135, -0.0941,  0.0042,  0.0270,  0.0313,  0.0726, -0.0391,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1402,  0.0509, -0.0106, -0.0934,  0.0272, -0.0165,  0.0762, -0.0269,
         0.0234,  0.0087,  0.0141, -0.0260, -0.1308,  0.0725, -0.0966, -0.0377,
         0.0191,  0.1294,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.9966e-02, -2.6475e-01, -8.7601e-02, -2.0350e-01, -4.3755e-02,
         7.9116e-03,  1.4313e-03, -2.5176e-02, -4.4779e-03, -7.3269e-02,
        -4.8399e-02, -3.7135e-03,  1.0533e-02, -3.7765e-02, -1.0869e-02,
        -8.4362e-05,  1.3196e-03, -3.2423e-02, -1.0306e-01,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0637, -0.0417, -0.0385, -0.0323, -0.0129, -0.0321, -0.0176,  0.0219,
        -0.0142, -0.0149,  0.0019,  0.0196, -0.0124,  0.0026,  0.0114,  0.0019,
        -0.0126,  0.0036,  0.0559,  0.0106, -0.0809,  0.0036, -0.0652, -0.0206,
        -0.0230, -0.0130, -0.0288,  0.0288,  0.0048, -0.0148, -0.0111,  0.0092,
        -0.0337, -0.0117, -0.0299, -0.0088, -0.0334,  0.0054, -0.0119, -0.0372,
        -0.0034,  0.0279,  0.0259, -0.0449], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0072, -0.1292, -0.0349, -0.0076,  0.0281, -0.0381, -0.0076,  0.0175,
        -0.1229, -0.0316, -0.0157,  0.0197,  0.0093, -0.0112,  0.0460, -0.0045,
        -0.0095, -0.1348,  0.0140, -0.0055, -0.0147, -0.0088,  0.0050, -0.0022,
         0.0109, -0.0216,  0.0095, -0.0193,  0.0364, -0.0034, -0.0384, -0.0044,
         0.0010, -0.0174,  0.1120,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0330, -0.3728, -0.0010, -0.0084,  0.0106,  0.0377, -0.0967, -0.0649,
         0.0151,  0.0067, -0.0035, -0.0388, -0.0749,  0.0048, -0.0096, -0.0657,
         0.0209, -0.0052,  0.0045,  0.0309, -0.0151, -0.0048,  0.0272, -0.0471,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0037, -0.1454, -0.0411, -0.0140, -0.0241, -0.0092,  0.0340, -0.0223,
        -0.0707,  0.0129,  0.0008,  0.0183,  0.0172, -0.0391, -0.0266,  0.0270,
         0.0260, -0.0332, -0.0648,  0.0073,  0.0140, -0.0227, -0.0143, -0.0131,
        -0.0096, -0.0428, -0.0251, -0.0363, -0.0018,  0.0213, -0.0813, -0.0218,
        -0.0376,  0.0207,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2408, -0.0413, -0.0168, -0.0398,  0.0092, -0.0096, -0.0605, -0.0044,
        -0.0274, -0.0192, -0.0036, -0.0136, -0.0404, -0.0023, -0.0015, -0.2244,
        -0.0452,  0.0044, -0.0126,  0.0034,  0.0118, -0.0138, -0.0012, -0.0196,
        -0.0327, -0.0158, -0.0261, -0.0319,  0.0210,  0.0058,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0994, -0.2034, -0.1163, -0.1570, -0.0438,  0.0261, -0.0632, -0.0419,
        -0.0316, -0.1332, -0.0072, -0.0622,  0.0146,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0243, -0.0993,  0.0720,  0.0453, -0.0424, -0.0244, -0.0186,  0.0286,
        -0.1647, -0.0457, -0.0621, -0.0613, -0.0245, -0.0402, -0.1123, -0.0246,
         0.0638, -0.0460,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0554, -0.3884, -0.0991, -0.0422, -0.1842, -0.0264, -0.1106, -0.0937,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0589, -0.0316, -0.0210, -0.0015, -0.0112, -0.0016, -0.0053, -0.0130,
         0.0045,  0.0179, -0.0045, -0.0120, -0.0137,  0.0258,  0.0006, -0.0124,
        -0.0024, -0.0024, -0.0111,  0.0036,  0.0246, -0.0134, -0.0057, -0.0334,
        -0.0042, -0.0244, -0.0112, -0.0380, -0.1241, -0.0259,  0.0492, -0.0678,
        -0.0017, -0.1033,  0.0037, -0.0368, -0.0435, -0.0174, -0.0143, -0.0031,
        -0.0017, -0.0099,  0.0002,  0.0273,  0.0151,  0.0447, -0.0004],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2281, -0.1221,  0.0649, -0.2845, -0.0308, -0.0068, -0.0167,  0.0040,
         0.0514, -0.1521,  0.0384,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1110, -0.1394, -0.0510, -0.0780, -0.0839, -0.0207, -0.0119, -0.1153,
        -0.0648, -0.0569, -0.0218, -0.0200, -0.0647, -0.0228,  0.0955,  0.0166,
        -0.0098, -0.0160,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1206,  0.2164,  0.0890, -0.0563,  0.0160, -0.0297,  0.0133,  0.0320,
         0.0265,  0.0338, -0.0404,  0.0192,  0.0999,  0.0030,  0.2039,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0135, -0.3288, -0.0889, -0.0240, -0.0200, -0.0171, -0.1001,  0.0927,
        -0.0167, -0.0313, -0.1128,  0.0022,  0.1021, -0.0245, -0.0254,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2283, -0.0418, -0.0301,  0.0429, -0.0046, -0.0636, -0.0744, -0.0540,
        -0.0224,  0.0051,  0.0064, -0.0259, -0.0953, -0.0146,  0.0138,  0.0342,
         0.0087,  0.0073,  0.0003,  0.0087, -0.0110, -0.0891,  0.0029,  0.0292,
         0.0723, -0.0130,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0538, -0.0267, -0.1217, -0.0771, -0.0592, -0.0660, -0.0199, -0.0353,
        -0.3445, -0.0102, -0.0364, -0.0080, -0.1122, -0.0139, -0.0150,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0337, -0.1492, -0.0330, -0.0205, -0.0547,  0.0326, -0.0567, -0.0035,
         0.0134, -0.1777,  0.0227,  0.0010, -0.0324, -0.0403, -0.0246,  0.0076,
        -0.0013,  0.0190, -0.0248,  0.0132, -0.0109,  0.0129, -0.0182,  0.0125,
         0.0877,  0.0957,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0270,  0.0250,  0.0104, -0.0159, -0.0353, -0.0325, -0.0081, -0.0355,
        -0.0080, -0.0137, -0.1591, -0.0125, -0.0188,  0.0096, -0.0623, -0.0345,
        -0.0493, -0.0290, -0.0280, -0.0512,  0.0261, -0.0189, -0.0095, -0.0058,
         0.1394, -0.1347,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.0269, -0.1847, -0.0586, -0.0079, -0.0217,  0.0195, -0.0420, -0.0371,
        -0.0053, -0.0306, -0.0122, -0.0072, -0.0476,  0.0087, -0.0135, -0.0109,
        -0.0132,  0.0103, -0.0397, -0.0130, -0.0131, -0.0173, -0.0223, -0.0406,
        -0.0543,  0.0307, -0.0449,  0.0050, -0.0065, -0.0192,  0.0055, -0.0171,
        -0.0055,  0.0144, -0.0595,  0.0335,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0721, -0.1198, -0.0390, -0.0266, -0.0201, -0.0219, -0.0371, -0.0655,
        -0.0023,  0.0356, -0.0037,  0.0024,  0.0016, -0.0116, -0.0215, -0.0066,
         0.0026, -0.0100,  0.0241,  0.0103, -0.0086, -0.0452, -0.0036, -0.0385,
        -0.0356, -0.0668,  0.0095, -0.0090, -0.0682, -0.0069, -0.0117, -0.0254,
        -0.0318,  0.0459, -0.0011,  0.0575,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0368,  0.2986,  0.0278,  0.0443, -0.0109, -0.0200,  0.0432,  0.0147,
         0.0025,  0.0049,  0.0072,  0.0344,  0.0463, -0.0104, -0.0561, -0.0206,
         0.0562,  0.1232,  0.0524, -0.0893,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1666, -0.0442, -0.0079,  0.0138,  0.0482, -0.0090,  0.0031,  0.0076,
         0.0146,  0.0173, -0.0098,  0.0047,  0.0106,  0.0229, -0.0263, -0.0049,
         0.0101, -0.0402, -0.0421, -0.0030, -0.0241, -0.0165, -0.0005,  0.0202,
         0.0022, -0.0112,  0.0073,  0.0142,  0.0045, -0.0006, -0.0567, -0.0081,
        -0.0513, -0.1465,  0.0180,  0.0257,  0.0414, -0.0443,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0099, -0.0626, -0.0426, -0.1115, -0.0077, -0.0077,  0.0168,  0.0090,
        -0.0038, -0.0279, -0.0553, -0.0840,  0.0056,  0.0125,  0.0069, -0.0455,
        -0.0084, -0.0393,  0.0124,  0.0141,  0.0079,  0.0247,  0.0053, -0.0118,
        -0.0158,  0.0299, -0.0133, -0.0083, -0.0821,  0.0092, -0.0147, -0.0080,
         0.0230,  0.0136, -0.0231, -0.0185,  0.0035,  0.0292, -0.0066, -0.0403,
         0.0277], device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.0408e-03,  4.1366e-01, -9.1325e-02,  1.3379e-01, -4.9311e-05,
         7.8404e-02, -4.2185e-02,  1.5254e-02,  2.4812e-02,  2.3656e-03,
         2.0567e-02,  1.2291e-03,  1.2318e-02, -1.6398e-02, -2.2658e-02,
         2.7615e-02,  1.0062e-02,  2.2918e-02, -1.4435e-02,  3.0515e-03,
         2.6755e-02,  1.5101e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0362, -0.0103, -0.0094, -0.0042, -0.0056, -0.0208, -0.0043,  0.0273,
        -0.0489, -0.0113, -0.0073, -0.0770, -0.0096, -0.0229,  0.0242, -0.0647,
        -0.0239, -0.0349,  0.0220, -0.0487, -0.1031,  0.0399,  0.0181, -0.0279,
        -0.0256,  0.0025,  0.0093, -0.0392, -0.0880, -0.0166, -0.0167, -0.0296,
        -0.0474, -0.0214,  0.0012,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0668, -0.0174,  0.0788,  0.0215,  0.0679,  0.0232, -0.0144, -0.0360,
         0.0335, -0.0026,  0.0639,  0.0351, -0.0033,  0.0039,  0.0307,  0.0439,
        -0.0670,  0.0090,  0.0631, -0.0117,  0.0499,  0.0716, -0.0593,  0.0237,
        -0.0034,  0.0003,  0.0592, -0.0279,  0.0110,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0037, -0.1692, -0.0433, -0.0311, -0.0058, -0.0450, -0.0104,  0.0223,
        -0.0270, -0.0285, -0.0404, -0.0332, -0.0025, -0.0266, -0.0103, -0.0494,
        -0.0527, -0.0062, -0.0435,  0.0013,  0.0543, -0.0715, -0.1001, -0.0238,
        -0.0980,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0669, -0.0987,  0.0991,  0.0838,  0.1602, -0.0006,  0.0282,  0.0280,
         0.0194,  0.0262, -0.0395,  0.0397,  0.0096, -0.0143,  0.0501,  0.0714,
         0.0109,  0.0174, -0.0216, -0.0055,  0.0372, -0.0644,  0.0073,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0998, -0.0698,  0.0024,  0.0022, -0.0156,  0.0195, -0.0076, -0.0176,
        -0.0083, -0.0093,  0.0103, -0.0121, -0.0413, -0.1246,  0.0063, -0.0046,
         0.0052, -0.0150, -0.0173, -0.0083, -0.0395, -0.0039,  0.0070, -0.0393,
         0.0166, -0.0076, -0.0503, -0.0122, -0.0403, -0.0212, -0.0707, -0.0206,
        -0.0131, -0.0247, -0.0067, -0.0266,  0.0280, -0.0615, -0.0132,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0272, -0.1475,  0.0374, -0.1444, -0.1813, -0.0560, -0.1616, -0.0931,
        -0.1151,  0.0087, -0.0004, -0.0274,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0695, -0.0672, -0.1512, -0.0155, -0.0218,  0.0291, -0.0437, -0.0164,
        -0.0242, -0.0243, -0.0088, -0.0016, -0.0043,  0.0153, -0.0266, -0.0267,
        -0.0075, -0.0342, -0.0095, -0.0366, -0.0212, -0.0206, -0.0184, -0.0087,
         0.0012,  0.0070,  0.0225,  0.0004, -0.0087,  0.0266, -0.0098,  0.0549,
        -0.0179,  0.0017, -0.0456, -0.0442, -0.0092, -0.0043, -0.0058, -0.0023,
        -0.0052,  0.0015,  0.0050, -0.0036,  0.0198,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0243, -0.2501, -0.0308, -0.0551, -0.0369, -0.0573, -0.1246, -0.0082,
        -0.0776, -0.0850,  0.1597, -0.0904,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0201,  0.4163,  0.1463,  0.1737,  0.0136, -0.0706,  0.1210, -0.0385,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.0791e-02, -1.1488e-01, -1.4044e-03,  5.0152e-02, -2.8321e-02,
        -6.5313e-03,  5.1387e-03, -1.1394e-02, -2.3394e-02, -5.2732e-02,
        -2.7691e-02, -6.1417e-03,  9.2858e-04, -4.5215e-03, -3.7505e-03,
         3.7085e-03,  1.1739e-02, -2.4423e-02,  6.5883e-03, -4.0385e-04,
        -7.1576e-03, -3.9814e-03,  6.3876e-03,  4.4428e-03,  1.1396e-03,
        -1.5641e-02, -1.6889e-02, -5.0005e-03, -2.1773e-02,  1.0714e-02,
        -1.3697e-02,  2.8911e-04, -1.0684e-03, -2.7043e-02,  5.8311e-03,
         2.5971e-03, -2.3465e-02, -7.2118e-03, -6.4618e-03, -1.5463e-03,
        -1.4006e-02,  7.1320e-03, -8.0580e-03,  9.0485e-03, -2.6385e-03,
         1.3857e-01, -3.6218e-02,  4.1815e-03, -3.1856e-02, -3.6659e-02,
        -8.2739e-03, -7.3837e-03, -1.3888e-02,  3.8819e-03, -2.1662e-03,
         1.7031e-03, -9.8887e-03,  6.6225e-05, -2.8325e-03,  1.5570e-02,
         9.0131e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1077, -0.0688, -0.0037, -0.0342, -0.1660, -0.0222, -0.0203, -0.0068,
         0.0269, -0.0031, -0.0213, -0.0171, -0.0152, -0.0145, -0.0888, -0.0002,
        -0.0191, -0.0311, -0.0285,  0.0133, -0.0456, -0.0134, -0.0105, -0.0034,
        -0.0205, -0.0311, -0.0178,  0.0074, -0.0219, -0.0407, -0.0351, -0.0410,
        -0.0028,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1705, -0.0211, -0.0462, -0.0110, -0.0027, -0.0085, -0.0389,  0.0090,
         0.0222, -0.0114,  0.0544,  0.0126, -0.0095, -0.0035, -0.0031,  0.0143,
         0.0098, -0.0725, -0.2494,  0.0031,  0.0563, -0.0311,  0.1157, -0.0234,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1043, -0.0654, -0.0366,  0.0051, -0.0538, -0.0739, -0.0379, -0.0364,
         0.0188, -0.0836, -0.0250, -0.0398, -0.1243, -0.0311,  0.0110, -0.0108,
        -0.0128, -0.0318, -0.0053,  0.0025, -0.0233,  0.0369,  0.0299,  0.0997,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1203, -0.0676, -0.0636, -0.0274,  0.0128,  0.0205, -0.0163, -0.0289,
         0.0125, -0.0241, -0.0050, -0.0059,  0.0178,  0.0074, -0.0296, -0.0434,
        -0.0240,  0.0371,  0.0213, -0.0458, -0.0397, -0.0239, -0.0168, -0.0889,
         0.0171,  0.0003, -0.0503, -0.0125,  0.0094,  0.0040, -0.0460, -0.0140,
        -0.0121,  0.0336,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0695, -0.0076, -0.0063, -0.0015, -0.0251, -0.0222, -0.0162,  0.0360,
        -0.0485, -0.0194,  0.0083, -0.0703,  0.0025, -0.2816, -0.0291, -0.1086,
        -0.0606, -0.0225,  0.0120, -0.0203,  0.0042, -0.0390, -0.0394,  0.0233,
        -0.0259,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0101,  0.0313,  0.0865,  0.0578,  0.0894, -0.0313,  0.0614,  0.0230,
         0.0392,  0.3435,  0.0200,  0.0256, -0.0341,  0.0256, -0.0703, -0.0509,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0159, -0.2079, -0.0350, -0.0808, -0.0427, -0.0286, -0.1138, -0.0283,
        -0.0228, -0.0392, -0.0202, -0.0346, -0.0511, -0.0633, -0.0271,  0.0119,
        -0.0087, -0.0194, -0.0729, -0.0268, -0.0034, -0.0208,  0.0246,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1276,  0.2246, -0.0320,  0.1396,  0.1909,  0.0798, -0.0032,  0.0579,
         0.0399, -0.0006,  0.0099,  0.0495,  0.0378,  0.0068,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.1128, -0.0323, -0.0118, -0.1344, -0.0232,  0.0258,  0.0038,  0.0569,
        -0.0037,  0.0023, -0.0536, -0.0338, -0.0416,  0.0033, -0.0169, -0.0090,
         0.0122, -0.1205, -0.0392, -0.0216, -0.0414, -0.0205,  0.0568,  0.1224,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1928, -0.0586, -0.1284,  0.0100, -0.1146, -0.0032, -0.1725, -0.0312,
        -0.0563, -0.0895, -0.1428,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1278, -0.0334, -0.0223,  0.0229, -0.0557, -0.0640, -0.1059,  0.0182,
        -0.0684, -0.0038, -0.0383,  0.0295, -0.0021,  0.0090, -0.0218,  0.0248,
        -0.0424, -0.1576,  0.0437,  0.0200,  0.0885,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2235, -0.1681,  0.0151, -0.0453, -0.0301, -0.0627,  0.0011, -0.1326,
        -0.0314, -0.0481, -0.0260, -0.0444, -0.1065, -0.0650,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1010, -0.0013, -0.0097,  0.0439, -0.0889,  0.0119, -0.0449,  0.0020,
         0.0186, -0.0955,  0.0265, -0.1086, -0.1079, -0.0121, -0.0350,  0.0556,
         0.0570,  0.0086,  0.0232,  0.1129,  0.0348,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1789, -0.1499,  0.0292, -0.0666, -0.0840, -0.0908, -0.0479, -0.0991,
        -0.0385, -0.0719, -0.0531, -0.0077, -0.0826,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0620, -0.0229, -0.0071, -0.0021, -0.0069, -0.0150, -0.0137, -0.0514,
        -0.0186, -0.0217, -0.0044,  0.0033, -0.0084,  0.0189, -0.0059, -0.0086,
        -0.0047,  0.0144, -0.0004,  0.0058, -0.0158, -0.0071, -0.0336, -0.0089,
        -0.1654,  0.0208, -0.0475, -0.0295, -0.0260, -0.0420, -0.0088,  0.0189,
         0.0110,  0.0125,  0.0288, -0.0178, -0.0178, -0.0053, -0.0084, -0.0058,
        -0.0053,  0.0173,  0.0028,  0.0614,  0.0852], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0564, -0.1072, -0.0267,  0.0010,  0.0014,  0.0132, -0.0151,  0.0017,
        -0.0061, -0.0237, -0.0361, -0.0117, -0.0153,  0.0180, -0.0447, -0.0076,
        -0.1379,  0.0152,  0.0027,  0.0121, -0.0100, -0.0371, -0.0242, -0.0774,
         0.0058,  0.0087,  0.0192, -0.0249, -0.0870, -0.0548,  0.0028, -0.0023,
        -0.0387, -0.0534,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1313,  0.2589,  0.0407,  0.0044,  0.0195,  0.0025, -0.0071,  0.0368,
         0.0113,  0.0138,  0.0225,  0.0328,  0.0104,  0.0017, -0.0173, -0.0432,
        -0.0040, -0.0276,  0.0054,  0.0456,  0.0558, -0.0065,  0.0171,  0.0377,
        -0.0443, -0.1017,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0262, -0.2256, -0.0282, -0.0938,  0.0453, -0.0585, -0.0160, -0.0761,
        -0.0582, -0.0286, -0.0212, -0.0169, -0.0131, -0.0411, -0.0299, -0.0244,
        -0.1337, -0.0118, -0.0232,  0.0014, -0.0070, -0.0197,  0.0004,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0778, -0.0671,  0.0283, -0.0088, -0.0136, -0.0098, -0.0256, -0.0113,
        -0.0489,  0.0023, -0.0133, -0.0142, -0.0126, -0.0147, -0.1972, -0.0083,
         0.0147, -0.0620, -0.0334, -0.0472, -0.0162, -0.0094,  0.0003,  0.0225,
        -0.0045,  0.0059, -0.0172, -0.0491,  0.0023,  0.0093, -0.0024, -0.0148,
         0.0090, -0.0037,  0.0079, -0.0019, -0.0020, -0.0569,  0.0099, -0.0436,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0665, -0.1058, -0.0425, -0.0329, -0.0446, -0.0382,  0.0296, -0.0385,
        -0.0082, -0.0182, -0.0150, -0.0034, -0.0112, -0.0348, -0.0313, -0.0310,
        -0.0155, -0.0225,  0.0274, -0.0186,  0.0267, -0.0265, -0.0062,  0.0091,
         0.0077,  0.0032, -0.0047, -0.0394, -0.0128,  0.0601, -0.0573, -0.0114,
        -0.0172, -0.0821,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.1324, -0.0448, -0.0037, -0.0145, -0.0188, -0.0202, -0.0284, -0.0326,
         0.0007,  0.0233, -0.0121, -0.0036, -0.0037,  0.0016, -0.0376, -0.0258,
         0.0423, -0.0067, -0.0173,  0.0195, -0.0141, -0.0016, -0.0367, -0.0080,
        -0.0302, -0.0955,  0.0305, -0.0125,  0.0130, -0.0169,  0.0111,  0.0113,
         0.0024,  0.0021, -0.0430,  0.0153, -0.0062, -0.0133, -0.0016, -0.0156,
        -0.0236, -0.0050, -0.0235, -0.0042, -0.0045,  0.0079,  0.0118,  0.0008,
        -0.0046, -0.0023, -0.0029, -0.0387], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0824, -0.0022, -0.0692, -0.0714,  0.0769,  0.0284, -0.0308, -0.0250,
        -0.0430,  0.0345,  0.0113, -0.0103, -0.0031,  0.0637,  0.0279,  0.0296,
         0.0845, -0.0479,  0.0359, -0.0262,  0.0176,  0.0070,  0.0541,  0.0038,
        -0.0106, -0.0239, -0.0044, -0.0063,  0.0205, -0.0235,  0.0242,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2326,  0.0477, -0.0050, -0.0038,  0.0194, -0.0043,  0.0467,  0.0180,
         0.0481, -0.0286,  0.0078,  0.0501, -0.0015,  0.0043, -0.0330,  0.0092,
        -0.0621,  0.0224,  0.0498,  0.0693,  0.0170,  0.0460,  0.0523, -0.0757,
         0.0453,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0283,  0.0436,  0.0386, -0.0123,  0.0899,  0.1728,  0.3330, -0.0094,
         0.0683,  0.0830, -0.0832, -0.0376,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.5105e-02,  2.1338e-01,  1.1729e-03,  3.7855e-02,  3.2537e-02,
         5.1536e-02,  2.1916e-02,  1.2162e-02, -3.8859e-03,  3.0972e-02,
         2.6170e-02, -5.3149e-03,  1.5706e-02,  7.4339e-02,  3.7997e-02,
         6.0540e-02,  2.1974e-02,  2.3769e-02, -3.2443e-02,  4.9913e-05,
         1.6618e-02,  1.6747e-02,  7.2861e-04, -2.1740e-02, -3.2536e-02,
         5.0852e-02,  4.0368e-02, -2.5572e-02,  4.6010e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0230,  0.2732, -0.0315, -0.0370, -0.0192,  0.0621,  0.2565, -0.0867,
        -0.0036,  0.0358,  0.0163,  0.1551,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1634, -0.2839, -0.0086, -0.0547,  0.0746, -0.0865, -0.1211,  0.0051,
        -0.0267, -0.0421,  0.0028, -0.1305,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2228,  0.0164,  0.0087,  0.0520,  0.0633, -0.0337,  0.0255,  0.0440,
         0.1108, -0.0109,  0.0075,  0.0707,  0.0276, -0.0319, -0.0234,  0.0861,
         0.0433,  0.0042, -0.0660, -0.0513,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0089, -0.2099, -0.0042, -0.0278, -0.0758,  0.0164,  0.0081,  0.0029,
        -0.0196,  0.0109,  0.0257,  0.0097,  0.0068,  0.0197, -0.0126,  0.0471,
        -0.0025,  0.0018,  0.0163, -0.0377, -0.0048,  0.0292,  0.0208,  0.0474,
         0.0122,  0.0207,  0.0370, -0.0197, -0.0242, -0.0206, -0.0129, -0.0096,
        -0.0013, -0.0111, -0.0450,  0.0723, -0.0468,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1426, -0.3989,  0.0331,  0.0914, -0.0363, -0.0317,  0.0032,  0.0881,
        -0.1747,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-9.9418e-02, -1.9160e-01,  5.7416e-02, -7.2772e-02,  3.8865e-03,
        -1.2525e-02, -2.5002e-02, -6.9608e-02,  2.6543e-02, -4.1577e-02,
        -1.6558e-02,  4.1095e-02, -2.2253e-02, -6.6197e-02,  5.3202e-03,
         3.3711e-03,  1.1166e-04,  1.5740e-02, -4.2008e-02, -5.4824e-02,
        -3.2214e-03,  7.3567e-03, -5.6845e-03,  2.5815e-03, -1.2925e-03,
         7.7704e-02, -3.4331e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0180,  0.1259,  0.0364,  0.0197,  0.0705,  0.0287,  0.0010, -0.0069,
        -0.0026,  0.0220,  0.0136,  0.0396,  0.0125, -0.0237,  0.0128,  0.0031,
         0.0135, -0.0412,  0.0193,  0.0099,  0.0155,  0.0602,  0.0402, -0.0292,
        -0.0007,  0.0226,  0.0077,  0.0215,  0.2206,  0.0031,  0.0148, -0.0009,
         0.0090,  0.0329,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 1.0909e-01,  2.0071e-02,  2.6949e-02,  2.8654e-03, -1.3896e-02,
         1.2102e-02,  1.5705e-03, -8.0782e-04, -2.8252e-02,  5.4492e-03,
        -3.3754e-02, -8.1759e-02, -1.9185e-02, -1.5903e-02, -1.7320e-02,
        -2.9515e-02, -3.5544e-02, -8.8915e-03, -1.6548e-02,  3.0921e-03,
         1.1191e-02,  1.0017e-02, -3.0128e-02,  2.4542e-03, -3.4243e-02,
        -2.2727e-02, -1.4821e-02,  3.2615e-04, -1.6433e-03, -6.4618e-03,
        -2.1961e-02, -4.0258e-02, -6.8710e-02, -1.6534e-02,  8.7433e-03,
        -1.6660e-02, -3.5014e-02, -9.9980e-03,  1.6278e-03, -1.1650e-02,
         1.2312e-05, -5.2210e-02,  1.2677e-02,  2.0249e-02,  6.7115e-02,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1085, -0.0611,  0.0064,  0.0098, -0.0567, -0.0726, -0.2671, -0.0043,
        -0.0566, -0.0440, -0.0190,  0.0262, -0.0527, -0.0595,  0.0057,  0.1498,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.3402, -0.0060,  0.0345,  0.0020, -0.0294, -0.0020, -0.0847, -0.0066,
        -0.0423, -0.0339,  0.0025, -0.0545, -0.0155,  0.0124, -0.0562,  0.0501,
        -0.0331, -0.0133,  0.0171, -0.0115, -0.0072, -0.0190, -0.0242, -0.0149,
         0.0250, -0.0228, -0.0392,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0729, -0.0151, -0.0153,  0.0032, -0.0181, -0.0100, -0.0087, -0.0134,
        -0.0368, -0.0260, -0.0139, -0.0123, -0.0179, -0.0042, -0.0103, -0.0030,
        -0.0037,  0.0016,  0.0429, -0.0293,  0.0092, -0.0234, -0.0246, -0.1262,
         0.0058, -0.0213, -0.0110, -0.0079,  0.0114, -0.0111, -0.0359, -0.0018,
        -0.0169,  0.0231,  0.0002,  0.0167, -0.0259, -0.0334,  0.0417,  0.0317,
        -0.0150, -0.0448, -0.0085, -0.0325,  0.0398, -0.0217], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1409,  0.0112,  0.0308,  0.0276, -0.0219,  0.0606,  0.0092,  0.0338,
         0.0169,  0.0616,  0.0231, -0.0104,  0.0169,  0.0080, -0.0058,  0.0012,
         0.0288,  0.0105,  0.0725,  0.0761,  0.0473,  0.0363,  0.0062,  0.0223,
         0.0366,  0.0695,  0.0181,  0.0371,  0.0157, -0.0431,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2827, -0.0025, -0.0081, -0.0121, -0.0112, -0.0543, -0.1190,  0.0016,
         0.0339, -0.1516,  0.3229,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0160, -0.2796,  0.0085, -0.0945, -0.0461, -0.0122, -0.0459, -0.1187,
         0.0035, -0.0854,  0.0198, -0.0203, -0.0045,  0.0069,  0.0139, -0.0395,
        -0.0342, -0.0109, -0.0604, -0.0792,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0937, -0.1809, -0.0798, -0.0565,  0.1306, -0.0660, -0.1573,  0.0247,
         0.0096, -0.1184, -0.0825,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0470, -0.1754, -0.0822, -0.1524, -0.0879,  0.0260, -0.0176, -0.0115,
        -0.0781, -0.0439, -0.0211, -0.0414, -0.0007, -0.0301, -0.0085, -0.0718,
        -0.0390, -0.0174, -0.0088,  0.0087, -0.0304,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0017,  0.1389,  0.0375, -0.0235,  0.0160,  0.3613,  0.0198, -0.0440,
         0.0603,  0.0056,  0.0669,  0.0577, -0.0042,  0.0170,  0.0218,  0.0565,
        -0.0558, -0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0362, -0.0883, -0.1287,  0.0217, -0.0085,  0.0221,  0.0158,  0.0091,
        -0.0120, -0.0046,  0.0062,  0.0060, -0.0159, -0.0286, -0.0345, -0.0215,
        -0.0576, -0.0680,  0.0007,  0.0003, -0.0185, -0.0357, -0.0795, -0.0133,
        -0.0403,  0.0082, -0.0208, -0.0108, -0.0136, -0.0862, -0.0045,  0.0048,
        -0.0184, -0.0054, -0.0124, -0.0079, -0.0022,  0.0107,  0.0205,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0586,  0.3260,  0.0731,  0.0221,  0.0086,  0.0099,  0.0076, -0.0466,
        -0.0138,  0.0110,  0.0029,  0.0255,  0.0204,  0.0093, -0.0036, -0.0036,
         0.0157, -0.0031,  0.0133, -0.0051,  0.0038, -0.0020, -0.0035, -0.0027,
        -0.0186,  0.0322,  0.0096,  0.0115,  0.0478,  0.0489,  0.0058,  0.0117,
         0.0609,  0.0611,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0816, -0.1945,  0.0605, -0.0949,  0.0089, -0.1686, -0.1420, -0.0330,
        -0.2161,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0640,  0.0834,  0.0566, -0.0142, -0.0868, -0.0375,  0.0085, -0.0153,
        -0.0162,  0.0082,  0.0131,  0.0053, -0.0040,  0.0115, -0.0303,  0.0392,
         0.0434,  0.2019, -0.0129,  0.0027,  0.0030, -0.0014, -0.0092, -0.0071,
        -0.0026, -0.0446,  0.0130, -0.0026, -0.0390,  0.0051,  0.0041,  0.0310,
         0.0127,  0.0054, -0.0047, -0.0028,  0.0063,  0.0061,  0.0098, -0.0345,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.6880e-02, -1.2900e-01, -6.9335e-02,  1.0076e-02, -1.4431e-02,
         1.8205e-02,  1.1997e-02, -8.2806e-03, -2.7455e-03, -2.6740e-02,
        -9.2473e-03,  6.4995e-03,  3.7511e-04, -3.2614e-03, -1.4787e-02,
         9.1365e-03, -1.0723e-02, -1.9116e-03, -1.6012e-02,  1.8105e-03,
         6.6798e-03, -1.0329e-02,  3.5116e-02, -3.0908e-02,  1.1760e-02,
        -1.8009e-02, -7.6016e-03,  4.1233e-02, -4.0113e-02, -1.6258e-02,
        -6.0921e-03, -2.8018e-02, -3.1565e-05,  5.6847e-03,  3.0799e-03,
        -2.2779e-02, -1.0576e-02, -2.4681e-02,  2.2003e-02,  1.7525e-02,
         4.2012e-02, -3.0001e-03,  3.1863e-03, -9.9443e-03,  1.1970e-02,
         1.4784e-02, -1.6827e-03,  1.0272e-02,  1.3785e-02, -1.7214e-02,
         5.7545e-03, -7.6462e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0032, -0.1355, -0.0647,  0.0093, -0.0448, -0.0307, -0.0394, -0.0316,
        -0.0260, -0.0070,  0.0579, -0.0587, -0.0340, -0.0182, -0.0286, -0.0661,
        -0.0481, -0.0021, -0.0777, -0.0610, -0.0597,  0.0958,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0208, -0.3627, -0.0429, -0.0361, -0.0080, -0.0556, -0.0616, -0.0079,
         0.0174, -0.0557, -0.0228, -0.0354, -0.0073, -0.0589, -0.0289, -0.0570,
        -0.0187,  0.0014, -0.0663,  0.0143, -0.0203,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.8910e-01, -3.6731e-02, -2.1037e-03, -1.4624e-02,  4.4816e-05,
        -3.4145e-03, -1.7651e-03,  4.6625e-03, -1.3539e-02, -8.5501e-03,
        -9.2168e-03, -1.6721e-03, -4.4602e-03,  9.2575e-03, -4.9612e-03,
        -5.1820e-04, -2.6964e-03, -5.0984e-04, -2.3627e-02, -1.0481e-02,
         1.7235e-03, -2.5595e-02,  1.0234e-02, -9.1254e-03, -9.1310e-03,
         7.6045e-03, -1.7350e-02, -1.6005e-02, -5.4312e-03, -9.0379e-03,
        -2.0325e-02, -8.4822e-03, -1.0025e-02, -4.1348e-03, -2.1321e-02,
        -5.6234e-03,  1.1765e-03,  2.4192e-03, -7.5086e-03, -7.3397e-03,
        -1.9315e-01, -2.5508e-02,  5.5962e-03, -2.8863e-02, -1.5400e-02,
        -2.8302e-03, -1.3459e-02, -1.9594e-02,  2.1476e-03,  4.8813e-03,
        -2.2510e-02, -1.8552e-02,  2.6540e-02, -3.8170e-02, -4.1272e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0868, -0.0365, -0.0054, -0.0144, -0.0252, -0.0227, -0.0154, -0.0217,
         0.0182,  0.0037,  0.0328, -0.0066, -0.0245, -0.0123, -0.0139, -0.0136,
         0.0031, -0.0140, -0.0046, -0.0346, -0.0326, -0.0315, -0.0192, -0.0083,
         0.0338,  0.0046, -0.0195, -0.0752, -0.0009, -0.0127,  0.0007, -0.0245,
        -0.0287, -0.0692,  0.0123, -0.0252, -0.0175, -0.0219, -0.0259, -0.0085,
        -0.0258,  0.0174, -0.0186,  0.0457, -0.0099,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.6319e-01,  1.2139e-01,  5.0778e-02,  5.7706e-02,  3.6829e-03,
         1.7208e-04,  2.8397e-02,  1.1630e-02,  3.3493e-02, -1.5773e-02,
        -2.1062e-02,  3.8010e-02,  4.5830e-02,  3.8045e-02,  2.8413e-02,
        -1.3214e-02, -1.0699e-04,  6.6952e-02, -2.4514e-02,  2.7023e-02,
         8.5364e-02,  3.1198e-02,  6.3763e-03,  8.7682e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0374, -0.0518, -0.0052,  0.0014, -0.0610,  0.0151, -0.0150, -0.0268,
         0.0001,  0.0049, -0.0069, -0.0128, -0.0140, -0.0414, -0.0583, -0.0294,
        -0.0718, -0.0298, -0.0153, -0.0431, -0.0530,  0.0153, -0.0339, -0.0834,
        -0.0048, -0.0368, -0.0458, -0.0348, -0.0094, -0.0916,  0.0052, -0.0020,
        -0.0425,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0473, -0.2763, -0.0359, -0.0588, -0.0112, -0.0268,  0.0113,  0.0004,
        -0.1875, -0.0135,  0.0058,  0.0097,  0.1552, -0.0360,  0.0224, -0.0593,
         0.0082,  0.0165, -0.0179,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.1667e-01, -6.5257e-02, -3.4936e-02, -4.2053e-02, -6.7901e-02,
         5.6103e-02, -6.5538e-02, -1.8350e-02, -1.3177e-01,  7.5583e-03,
         1.4920e-02,  2.4957e-02, -2.1525e-02, -3.5296e-03, -5.5173e-03,
        -5.7908e-05, -1.1110e-02, -4.4636e-02, -6.4681e-03,  1.0203e-02,
        -3.7232e-02, -4.7561e-02, -1.2222e-03,  4.1619e-03,  6.0768e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1892, -0.1199, -0.0706, -0.0187, -0.0552, -0.0508, -0.2399,  0.0041,
        -0.0078, -0.0283,  0.0124, -0.0113, -0.0340, -0.0202, -0.0533,  0.0499,
         0.0345,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0593,  0.0343, -0.0260, -0.0093, -0.0324, -0.0038,  0.0165, -0.0071,
        -0.0105, -0.0451, -0.0335, -0.0542, -0.0164, -0.0024, -0.0478, -0.0234,
        -0.0329, -0.0886,  0.0128, -0.0156,  0.0009, -0.0565, -0.1378, -0.0014,
         0.0355, -0.0416, -0.0248, -0.0072, -0.0170, -0.0141, -0.0365, -0.0547,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0694,  0.0974,  0.0285,  0.0095,  0.0311,  0.0186,  0.0142,  0.0838,
        -0.0005,  0.0146, -0.0075,  0.0105, -0.0233,  0.0376,  0.0215, -0.0056,
         0.0397,  0.0336,  0.0404,  0.0521,  0.0291,  0.0063,  0.0589,  0.1002,
         0.0036,  0.0193,  0.1166, -0.0268,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0409, -0.2536, -0.0100,  0.0187, -0.0247, -0.0511, -0.1314, -0.0179,
        -0.0100, -0.0237, -0.0217, -0.0578, -0.0143, -0.0411, -0.0127, -0.0101,
        -0.0130, -0.0484, -0.0005,  0.0078, -0.0076, -0.0124,  0.0084, -0.0086,
        -0.0209,  0.0093,  0.0189, -0.0233,  0.0811,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0466,  0.0070, -0.0115, -0.0202, -0.0985, -0.0481,  0.0031, -0.0150,
         0.0084,  0.0323, -0.0244, -0.0466, -0.0417, -0.0732,  0.0066, -0.0387,
        -0.0033, -0.1118, -0.0080, -0.0543, -0.0373, -0.0341, -0.0105, -0.0159,
        -0.0212, -0.0021, -0.0147, -0.0551, -0.0183,  0.0201, -0.0032, -0.0603,
        -0.0076,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1654, -0.1300, -0.0116, -0.0253, -0.0368, -0.0605,  0.0105, -0.0565,
        -0.0151, -0.0362, -0.0372, -0.0365, -0.0187, -0.0172, -0.0400,  0.0032,
        -0.0254, -0.0301, -0.1408,  0.0059, -0.0362, -0.0278,  0.0123, -0.0207,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0218, -0.1433, -0.1135, -0.0255, -0.0196,  0.0107, -0.0500, -0.0789,
        -0.0127, -0.0535, -0.0229, -0.0128, -0.0278, -0.0075,  0.0176, -0.0045,
         0.0024, -0.0764, -0.0248, -0.2128, -0.0217,  0.0160,  0.0232,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0479, -0.1420, -0.0214,  0.0055, -0.0096, -0.0777,  0.0022,  0.0080,
        -0.0100, -0.0525, -0.0188, -0.0259, -0.0031,  0.0155, -0.0162,  0.0038,
        -0.0508, -0.0221, -0.0083, -0.0556,  0.0117, -0.0427,  0.0197, -0.0188,
        -0.0219, -0.0168, -0.0061,  0.0041, -0.0239, -0.0400, -0.0325, -0.0118,
        -0.0222, -0.0131, -0.0233, -0.0149, -0.0029,  0.0330, -0.0439,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.5138e-02, -9.6302e-02, -3.0159e-02, -2.4813e-02, -8.9749e-02,
        -8.6659e-03,  3.0550e-06, -1.0273e-02, -1.9246e-02,  3.4590e-03,
        -3.0008e-03, -1.2187e-02, -2.3676e-03, -1.1319e-02,  3.9459e-02,
        -3.4305e-02,  1.3538e-02,  1.0707e-03,  1.4415e-03, -8.0798e-03,
        -3.2109e-02, -2.2038e-02,  4.2490e-03, -3.1834e-02, -5.1301e-02,
        -2.8554e-02, -1.8566e-02, -4.4263e-03, -1.3968e-02, -2.5928e-02,
         1.1614e-02,  2.4902e-03, -3.0855e-02, -9.9107e-03, -1.5011e-02,
         2.7262e-04, -2.6367e-02, -8.8197e-02, -1.0201e-02, -1.6664e-02,
        -1.5448e-02, -5.7355e-02, -4.8065e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0234, -0.1905,  0.0178,  0.0147,  0.0096, -0.0257, -0.0178, -0.0006,
        -0.0297,  0.0255,  0.0173, -0.0089, -0.0164, -0.0027,  0.0051, -0.0111,
        -0.0199, -0.0220,  0.0147, -0.0134,  0.0091, -0.0157, -0.0103, -0.0094,
        -0.0029, -0.0104, -0.0038, -0.0029,  0.0150, -0.0225,  0.0015, -0.0023,
        -0.0454, -0.0688, -0.0003, -0.0107, -0.0246, -0.0145, -0.0267, -0.0104,
        -0.0149, -0.0300, -0.0039,  0.0112,  0.0015, -0.0238, -0.0030, -0.0062,
         0.0020, -0.0317, -0.0235, -0.0113, -0.0160,  0.0039,  0.0079, -0.0043,
         0.0113], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0542,  0.4168,  0.1087,  0.0264,  0.0793,  0.0294,  0.0273,  0.0894,
         0.0361, -0.0095,  0.0394, -0.0415,  0.0156,  0.0265,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1526, -0.0850, -0.0616,  0.0036, -0.0261, -0.0320, -0.0318,  0.0063,
        -0.0225,  0.0002, -0.0172, -0.0612, -0.0022,  0.0317, -0.0512, -0.0494,
        -0.0236, -0.0472,  0.0187, -0.0476, -0.0748,  0.0030, -0.0204, -0.0004,
         0.0083, -0.0110, -0.0092,  0.0287,  0.0380,  0.0346,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([0.1750, 0.2880, 0.0075, 0.2168, 0.0485, 0.0023, 0.1312, 0.1306, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0032, -0.1646, -0.0080, -0.0196, -0.0379, -0.1048, -0.0196,  0.0038,
         0.0405,  0.0052,  0.0185, -0.0353, -0.0295, -0.0514, -0.0170, -0.0016,
        -0.0055, -0.0022, -0.0005, -0.0270,  0.0274, -0.0320, -0.0181,  0.0032,
        -0.0202, -0.0155, -0.0367, -0.0114, -0.0460, -0.0913,  0.0357,  0.0410,
         0.0259,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0000e-01, -1.1902e-01, -5.5287e-02, -3.8731e-03,  1.4743e-02,
        -1.5064e-02,  3.4918e-03, -1.1135e-02,  7.5972e-03, -1.8416e-02,
        -1.1472e-03,  6.1764e-05, -9.6562e-03, -3.6404e-03, -2.3860e-03,
         2.3305e-03, -2.7498e-03,  1.5748e-02, -5.2841e-04,  5.1399e-04,
        -8.6300e-04,  1.2825e-02,  3.8346e-03, -5.0034e-03,  1.0137e-03,
        -2.9377e-03, -1.3501e-02, -7.7007e-02, -2.6452e-03,  1.3659e-03,
         1.2202e-02, -1.6918e-02, -1.4147e-02, -8.5137e-02,  1.6401e-02,
        -1.2900e-02, -1.2735e-02,  1.2935e-03, -2.0313e-03,  1.0639e-02,
         8.5974e-03, -5.8593e-03, -7.5151e-03, -8.9609e-03,  4.2649e-03,
         9.0785e-03,  1.1103e-02,  5.7660e-03,  1.3584e-02,  7.7998e-03,
        -1.2792e-02, -2.3008e-02, -6.8933e-03, -1.1583e-02,  3.3259e-03,
         1.5712e-02,  1.6254e-03, -1.3255e-02, -1.5438e-02, -1.5287e-02,
        -9.6852e-03, -9.3649e-03, -3.2312e-03, -1.0044e-02,  2.6844e-03,
        -2.2459e-03,  1.7347e-03,  1.5879e-02, -1.1459e-02, -5.2522e-03,
        -3.4175e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1266, -0.0353, -0.0084, -0.0103, -0.0364, -0.0115, -0.0723, -0.0115,
        -0.0175, -0.0046, -0.0140, -0.0102, -0.0071, -0.0379, -0.0368, -0.0021,
        -0.0054,  0.0066, -0.0418, -0.0861,  0.0196,  0.0027,  0.0189, -0.0716,
        -0.0051, -0.0319, -0.0255, -0.0053,  0.0290, -0.0066, -0.0161, -0.0172,
        -0.0052, -0.0445,  0.0024, -0.0103, -0.0208, -0.0064, -0.0125, -0.0124,
        -0.0158,  0.0379,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0148, -0.1763,  0.0122, -0.0255, -0.0046,  0.0119,  0.0018, -0.0071,
        -0.0408, -0.0601, -0.0357, -0.0179, -0.0548, -0.0116,  0.0211,  0.0150,
        -0.0012, -0.0414, -0.0575, -0.0140, -0.0177, -0.0055, -0.0082, -0.0058,
        -0.0319, -0.0747, -0.0053, -0.0148, -0.0354,  0.0151, -0.0212, -0.0093,
        -0.0037, -0.0569,  0.0228,  0.0406, -0.0055,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0040, -0.3367,  0.0678, -0.0715, -0.0347, -0.0767, -0.0310, -0.0643,
        -0.0250, -0.0174, -0.0541,  0.0071, -0.0411, -0.0915,  0.0769,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2770,  0.0683,  0.0107,  0.0011,  0.0200,  0.0455,  0.0561,  0.0271,
         0.0796,  0.0079, -0.0027,  0.0176,  0.0451,  0.0133,  0.0159, -0.0276,
         0.0858,  0.1225,  0.0472,  0.0289,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1481,  0.0236, -0.0083,  0.0023, -0.1230, -0.0170, -0.0085, -0.0027,
        -0.0180, -0.0103,  0.0403, -0.0349, -0.0116, -0.0038, -0.0336, -0.0232,
        -0.0023, -0.0179,  0.0148, -0.0155,  0.0172, -0.0127, -0.0052, -0.0063,
         0.0111, -0.0142, -0.0761, -0.0041, -0.0254,  0.0050, -0.0297, -0.0173,
        -0.0022, -0.0131, -0.0002, -0.0670, -0.0020, -0.0099, -0.0023, -0.0034,
         0.0440,  0.0226,  0.0493,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0088,  0.0284,  0.0147, -0.0305, -0.0314, -0.0403, -0.0088, -0.0306,
        -0.0023, -0.0195, -0.0459, -0.0378, -0.0419, -0.0291, -0.0941, -0.0138,
        -0.0118, -0.0169, -0.0109,  0.0067, -0.0302, -0.0221, -0.0566, -0.0090,
        -0.0278, -0.0219,  0.0333,  0.0092, -0.0855, -0.0283, -0.0163, -0.0384,
        -0.0083, -0.0071, -0.0146, -0.0129, -0.0143,  0.0401,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0172, -0.1498, -0.0746, -0.1428, -0.0179, -0.0359,  0.0555,  0.0117,
        -0.0199, -0.1462,  0.0172, -0.0020, -0.0228,  0.0162,  0.0340,  0.0163,
        -0.0162, -0.0403,  0.0757,  0.0055,  0.0678,  0.0146,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0483, -0.0550, -0.0008,  0.0334,  0.0176, -0.0462, -0.0049,  0.0255,
        -0.0130, -0.0167, -0.0233,  0.0124, -0.0087, -0.0004, -0.0202, -0.0179,
        -0.0966, -0.0195, -0.0320, -0.0423, -0.0119, -0.0233, -0.0308, -0.0410,
        -0.0094,  0.0118, -0.0322, -0.0025,  0.0070, -0.0391, -0.0549,  0.0059,
         0.0027, -0.0018, -0.0306, -0.0902,  0.0074, -0.0044, -0.0105, -0.0480,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0911, -0.0213, -0.0104, -0.0237, -0.0426, -0.0265,  0.0098, -0.0191,
        -0.0446, -0.0394, -0.0210, -0.0029,  0.0045, -0.0695, -0.0287, -0.0568,
        -0.0163, -0.0163, -0.0494, -0.0170, -0.0088, -0.0109, -0.0154, -0.0325,
        -0.0070, -0.1466,  0.0061, -0.0509, -0.0560, -0.0201,  0.0269,  0.0081,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0038, -0.2800,  0.0175,  0.0011, -0.0496, -0.0105, -0.0298, -0.0289,
        -0.0708, -0.0214, -0.0162, -0.0112,  0.0065, -0.0153,  0.0047, -0.0333,
        -0.0097,  0.0012,  0.0091, -0.0141,  0.0098, -0.0308, -0.0271, -0.0414,
        -0.0061,  0.0162, -0.0040, -0.0152,  0.0062, -0.0066,  0.0097,  0.0022,
         0.0138,  0.0173, -0.1589,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0950, -0.0551, -0.0800, -0.0199, -0.0464, -0.0608, -0.1214, -0.0867,
        -0.0010, -0.1290, -0.0451, -0.0223,  0.0123, -0.0353, -0.0273, -0.0116,
        -0.0400, -0.0387, -0.0155, -0.0383, -0.0183,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1712, -0.3240,  0.0093, -0.0464, -0.0548, -0.0578, -0.0237, -0.0568,
        -0.1207,  0.0301,  0.0137,  0.0208,  0.0089, -0.0127, -0.0009, -0.0053,
        -0.0139,  0.0091,  0.0197,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0214, -0.1169, -0.0263, -0.0445, -0.0191, -0.0140,  0.0124, -0.1200,
         0.0212, -0.0427, -0.0368,  0.0234, -0.0468, -0.1530, -0.0095, -0.0319,
        -0.0480, -0.0556,  0.0524, -0.1042,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0312, -0.2131, -0.0178, -0.0146,  0.0085, -0.0161, -0.0390, -0.0489,
        -0.0156, -0.0509, -0.0087, -0.0364,  0.0088, -0.0491, -0.0197, -0.0282,
        -0.1009, -0.0352, -0.0225, -0.0488,  0.0114,  0.0793,  0.0953,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0818, -0.1207, -0.0528, -0.0383,  0.0412, -0.0110, -0.0206, -0.0106,
        -0.0198, -0.1487, -0.0054, -0.0176, -0.0036, -0.0923,  0.0762,  0.0323,
        -0.0092, -0.0092, -0.0795, -0.0181, -0.0119,  0.0003, -0.0251, -0.0687,
         0.0050,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.5555e-02, -3.2891e-01,  6.1449e-03,  5.7977e-02,  3.8650e-03,
        -3.1881e-02,  4.4139e-02,  2.6224e-03,  3.0904e-02, -6.5190e-02,
        -3.3561e-02, -6.7264e-04, -3.8946e-04, -1.6175e-02,  3.0664e-02,
         1.6976e-02, -3.0061e-03, -7.0773e-03,  1.1814e-03, -1.9110e-03,
        -5.4669e-02, -2.5479e-02, -3.0019e-02, -5.1147e-02, -3.6990e-02,
        -1.1852e-02,  9.7813e-03,  3.2054e-03,  7.2946e-03, -3.2303e-04,
         1.8617e-02,  3.3437e-04, -4.0984e-03,  3.6302e-03,  3.7586e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1018,  0.0041,  0.0224, -0.0699, -0.0325,  0.0168, -0.0232,  0.0116,
        -0.1273, -0.0285, -0.0145,  0.0276,  0.1035, -0.0271, -0.0241, -0.0860,
        -0.0192, -0.0683, -0.0747, -0.0187, -0.0399, -0.0583,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1398, -0.0533,  0.0288,  0.0002,  0.0021, -0.0646, -0.0423,  0.0148,
        -0.0101, -0.1237, -0.1256,  0.0589,  0.0422,  0.0020, -0.0988, -0.0679,
        -0.0076,  0.0135,  0.0528, -0.0290, -0.0221,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0469, -0.3497,  0.0155,  0.0127, -0.0228,  0.0395, -0.0526, -0.1306,
        -0.0280, -0.0159, -0.0075, -0.0086,  0.0040, -0.0907, -0.0015, -0.1253,
         0.0156, -0.0285, -0.0041,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0679, -0.0040, -0.0149, -0.0099, -0.0133, -0.0437,  0.0032, -0.0176,
        -0.0002, -0.0231,  0.0047, -0.0004, -0.0087, -0.0206, -0.0053,  0.0029,
        -0.0154, -0.0377, -0.0099, -0.0220,  0.0087, -0.0211,  0.0337, -0.0044,
         0.0311,  0.0201, -0.0238, -0.0939, -0.0331, -0.0906, -0.0222,  0.0019,
        -0.0259, -0.0559, -0.0015, -0.0002,  0.0355,  0.0013,  0.0259, -0.0054,
        -0.0184, -0.0086,  0.0049,  0.0103, -0.0109, -0.0160,  0.0693],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0036, -0.1510, -0.0325, -0.0045,  0.0092, -0.0090,  0.0073,  0.0439,
        -0.0075,  0.0035,  0.0276,  0.0208,  0.0140,  0.0058,  0.0204, -0.0257,
        -0.1287, -0.0071, -0.0459,  0.0034, -0.0136, -0.0285, -0.0212, -0.0986,
        -0.0177,  0.0385,  0.0230, -0.0058, -0.0078,  0.0029, -0.0436,  0.0196,
        -0.0081,  0.0994,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1104, -0.3191,  0.0027,  0.0009, -0.0440, -0.0239,  0.0027, -0.0011,
        -0.0099, -0.0136,  0.0253,  0.0060, -0.0160, -0.0286,  0.0031,  0.0311,
        -0.0460,  0.0261, -0.0358, -0.0448, -0.0079,  0.0199,  0.0042,  0.0153,
        -0.0010, -0.0026, -0.0030, -0.0055,  0.0022, -0.0179,  0.0005, -0.0380,
         0.0430, -0.0476,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0397, -0.2532,  0.0536, -0.0145, -0.0416, -0.0504, -0.0206,  0.0086,
        -0.0106, -0.0799, -0.0877,  0.0196, -0.0664, -0.0827, -0.0282, -0.0329,
         0.0138,  0.0961,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0520, -0.1186, -0.0875, -0.0150, -0.0189,  0.0044, -0.0187, -0.0165,
        -0.0625, -0.0510,  0.0030, -0.0287, -0.0218, -0.0395, -0.0156,  0.0262,
         0.0340,  0.0993,  0.0040, -0.0839,  0.0236,  0.0276, -0.0109,  0.0144,
        -0.0180, -0.0892, -0.0151,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0746, -0.0883, -0.0803, -0.0346, -0.0222,  0.0130, -0.0388, -0.0710,
        -0.0823, -0.0102, -0.0305, -0.0275, -0.0566, -0.0561,  0.0071, -0.0217,
        -0.0400, -0.0231, -0.0038, -0.0778, -0.0129, -0.0169, -0.0278,  0.0211,
        -0.0462, -0.0154,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0250,  0.2908,  0.0279,  0.0486,  0.0678,  0.0229,  0.0006,  0.0830,
         0.0156,  0.0304,  0.0604,  0.0033, -0.0060, -0.0225,  0.0510,  0.0247,
         0.0230, -0.0085, -0.0362,  0.0970, -0.0549,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0385, -0.2137, -0.1251, -0.1552, -0.0099,  0.0253, -0.0578, -0.0100,
        -0.0161, -0.0121, -0.0430, -0.0480,  0.0334,  0.1217, -0.0900,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0970,  0.0166,  0.0709,  0.0287, -0.0114,  0.1590, -0.0014,  0.0288,
         0.0441, -0.0051,  0.0408, -0.0310,  0.2433,  0.0599,  0.1272, -0.0093,
         0.0169,  0.0087,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2562, -0.1200, -0.0615, -0.1934, -0.0260,  0.0048, -0.0070, -0.0098,
         0.0009, -0.0464, -0.0442, -0.0219,  0.0485, -0.0344, -0.0354,  0.0159,
         0.0005, -0.0460,  0.0273,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0008, -0.0605, -0.0212,  0.0026, -0.0210, -0.0166, -0.0035, -0.0095,
        -0.0205, -0.0027, -0.0117, -0.0062, -0.0188, -0.0159,  0.0025, -0.0286,
        -0.0121, -0.0123,  0.0104,  0.0728, -0.0183,  0.0047, -0.0543, -0.0120,
        -0.0316, -0.0280, -0.0471, -0.0044, -0.0082, -0.0099,  0.0098, -0.0299,
        -0.0639, -0.0179, -0.0311, -0.0252, -0.0762, -0.0126, -0.0224, -0.0073,
         0.0047,  0.0336, -0.0415, -0.0554], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0913, -0.0653, -0.0389, -0.0308, -0.0308, -0.0386, -0.0415,  0.0042,
        -0.0857, -0.0201, -0.0162, -0.0239,  0.0074, -0.0053,  0.0285, -0.0307,
        -0.0426, -0.1163, -0.0005, -0.0109, -0.0242, -0.0133, -0.0045, -0.0024,
        -0.0104, -0.0161, -0.0004, -0.0583,  0.0169,  0.0014, -0.0483, -0.0250,
         0.0094, -0.0009,  0.0389,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0081, -0.3921, -0.0064, -0.0343,  0.0203,  0.0260, -0.0664, -0.0363,
         0.0341, -0.0053, -0.0136, -0.0258, -0.0264, -0.0052, -0.0005, -0.1078,
        -0.0419,  0.0160,  0.0165, -0.0046, -0.0211, -0.0249, -0.0255, -0.0410,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0072, -0.1428, -0.0290, -0.0157, -0.0251, -0.0080, -0.0122, -0.0323,
        -0.0726,  0.0018,  0.0022, -0.0012,  0.0349, -0.0223, -0.0485,  0.0353,
         0.0466,  0.0088, -0.0913,  0.0072, -0.0195, -0.0332, -0.0124, -0.0099,
         0.0039, -0.0394, -0.0320, -0.0210,  0.0005,  0.0268, -0.0314, -0.0181,
        -0.0795, -0.0276,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0216, -0.0663, -0.0409, -0.0248,  0.0454, -0.0071, -0.0873, -0.0407,
        -0.0094, -0.0562, -0.0562, -0.0014, -0.0437, -0.0357, -0.0549, -0.1538,
        -0.0034, -0.0181, -0.0026,  0.0453, -0.0057, -0.0027, -0.0118, -0.0245,
        -0.0015,  0.0015, -0.0233,  0.0037,  0.0338, -0.0766,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-8.5569e-02, -2.2197e-01, -7.4554e-02, -1.3562e-01, -1.6439e-02,
        -2.7050e-02, -4.4352e-02, -2.1059e-02, -3.9505e-02, -1.6290e-01,
        -7.2155e-02,  2.1772e-04, -9.8605e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0302, -0.0541,  0.0688, -0.0313, -0.0430, -0.0650,  0.0895, -0.0395,
        -0.1391, -0.0817, -0.0197, -0.0950,  0.0270, -0.0563, -0.1074,  0.0221,
        -0.0226, -0.0076,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1330, -0.2347, -0.1167, -0.0973, -0.1256,  0.0954, -0.0203, -0.1771,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.2555e-02, -4.9986e-02, -1.2536e-02,  5.6220e-03,  3.3722e-03,
        -2.1809e-03, -4.9421e-03,  7.7533e-03,  1.1349e-02,  2.8586e-03,
         5.4277e-06, -1.3239e-02, -1.9586e-02, -5.6289e-04, -1.0413e-02,
        -4.7340e-03,  5.0394e-03,  1.2102e-02, -9.3706e-03, -9.2364e-03,
        -5.4156e-03, -4.8581e-04, -1.2303e-02, -2.9062e-02, -1.8212e-02,
        -2.8523e-03, -8.4875e-02, -4.3203e-02, -1.2278e-01, -2.7156e-02,
         1.6484e-02, -4.5054e-02,  7.7882e-03, -6.0333e-02,  5.4225e-03,
        -3.4782e-02, -7.1727e-02, -1.3290e-02,  1.5169e-02, -1.0109e-02,
        -9.0916e-03, -1.4518e-02, -4.5782e-03,  5.1325e-03, -1.7089e-02,
         2.2388e-02,  3.3254e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0813, -0.0831, -0.2140, -0.2733, -0.0022,  0.0175, -0.0380,  0.0781,
        -0.0563,  0.0966, -0.0595,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0188, -0.1636,  0.0280,  0.0016, -0.0489, -0.0523, -0.0601, -0.2088,
        -0.0161, -0.0833, -0.0589, -0.0474, -0.0680, -0.0100, -0.0012, -0.0389,
        -0.0232, -0.0709,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0649, -0.1390, -0.2216, -0.0561, -0.0295, -0.0265, -0.1626, -0.0434,
        -0.0452, -0.0165,  0.0516,  0.0147, -0.0321,  0.0866, -0.0097,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0391,  0.2610, -0.0056,  0.0031, -0.0976,  0.0527,  0.1062, -0.0854,
         0.0954,  0.0515,  0.1359,  0.0111, -0.0007, -0.0289, -0.0259,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0172, -0.0304, -0.0390,  0.0064,  0.0009, -0.0502, -0.0716, -0.0586,
        -0.0860, -0.0548, -0.0383, -0.0985, -0.0867, -0.0182, -0.0113, -0.0134,
        -0.0461,  0.0113,  0.0031,  0.0280, -0.0405, -0.0630, -0.0209,  0.0472,
        -0.0282, -0.0300,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0929, -0.1189, -0.0713, -0.0484, -0.0265, -0.0244, -0.0376, -0.0319,
        -0.3465, -0.0017, -0.0401,  0.0438,  0.0232,  0.0319, -0.0609,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0395, -0.1880, -0.0840, -0.0156, -0.0648, -0.0064, -0.0357,  0.0288,
        -0.0236, -0.0975, -0.0051,  0.0530, -0.0302, -0.0096, -0.0039, -0.0221,
        -0.0144,  0.0532, -0.0364,  0.0337, -0.0375,  0.0101,  0.0502,  0.0129,
        -0.0267,  0.0170,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0600,  0.0060,  0.0292,  0.0200, -0.0329, -0.0269, -0.0368, -0.0059,
        -0.0245, -0.0499, -0.1169, -0.0278, -0.0178,  0.0569, -0.0758, -0.0251,
         0.0672, -0.0558, -0.0190, -0.0450,  0.0076, -0.0347,  0.0043,  0.0029,
         0.0095, -0.1415,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0175, -0.2405,  0.0125, -0.0271, -0.0326, -0.0112, -0.0149, -0.0339,
         0.0111, -0.0396, -0.0188, -0.0235, -0.0569, -0.0139, -0.0375, -0.0333,
         0.0049, -0.0456, -0.0097,  0.0097,  0.0325, -0.0278,  0.0188, -0.0218,
        -0.0572, -0.0344, -0.0230,  0.0052, -0.0077,  0.0122,  0.0027, -0.0187,
        -0.0070, -0.0087, -0.0123,  0.0151,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0006, -0.1279, -0.0612, -0.0188, -0.0525, -0.0285, -0.0188, -0.0074,
         0.0254,  0.0160, -0.0127,  0.0052,  0.0280,  0.0438, -0.0152, -0.0035,
         0.0191, -0.0067,  0.0249, -0.0049, -0.0164, -0.0516, -0.0175, -0.0170,
        -0.0314, -0.1066,  0.0047, -0.0098, -0.0374, -0.0197, -0.0126, -0.0285,
        -0.0859,  0.0042, -0.0069,  0.0289,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0830, -0.3301, -0.0987, -0.0238,  0.0200,  0.0121, -0.0439,  0.0103,
        -0.0048,  0.0032,  0.0053, -0.0295, -0.0736, -0.0351, -0.0116, -0.0288,
        -0.0963,  0.0044,  0.0428,  0.0425,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0099, -0.1061, -0.0032, -0.0120,  0.0075, -0.0126, -0.0632, -0.0293,
         0.0281, -0.0148, -0.0040, -0.0029, -0.0125, -0.0111, -0.0963, -0.0157,
        -0.0005, -0.0159,  0.0002, -0.0293,  0.0094,  0.0017,  0.0074, -0.0107,
        -0.0395,  0.0048,  0.0005, -0.0316, -0.0217, -0.0250, -0.0271, -0.0354,
         0.0019, -0.0614,  0.0712, -0.0624, -0.0501,  0.0632,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7276e-02, -4.4117e-02, -4.9511e-03, -7.4734e-02, -3.2443e-03,
        -1.5429e-02,  1.3940e-02,  1.4217e-02, -2.6058e-02, -5.4870e-02,
        -5.7172e-02, -9.8674e-02, -1.4193e-03, -4.6854e-03, -8.4015e-04,
        -4.4742e-02,  3.9546e-03, -6.5896e-02, -1.9038e-02,  9.9083e-05,
        -1.3120e-02,  5.7790e-03,  5.9128e-03, -1.4960e-02,  3.8107e-02,
        -8.0320e-03,  3.1944e-03, -2.8789e-02, -8.1632e-03, -9.4782e-04,
        -1.2074e-03, -1.0192e-02,  2.9989e-03,  1.1721e-02,  3.7744e-03,
        -5.1001e-03, -1.0865e-03,  7.9289e-02,  1.6303e-03,  1.3286e-01,
        -5.7782e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0447, -0.3832,  0.0176, -0.0007,  0.0321, -0.0446, -0.1130, -0.0250,
        -0.0216, -0.0153,  0.0022, -0.0233, -0.0421,  0.0165, -0.0297, -0.0548,
        -0.0124, -0.0369,  0.0052, -0.0238, -0.0045, -0.0506,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1254, -0.0164, -0.0114, -0.0125, -0.0327, -0.0278, -0.0226, -0.0028,
        -0.0296, -0.0221,  0.0376, -0.0743, -0.0027, -0.0420,  0.0224, -0.0401,
        -0.0110,  0.0027, -0.0025, -0.0488, -0.1306,  0.0124, -0.0044, -0.0348,
        -0.0184, -0.0054,  0.0029, -0.0354, -0.0863, -0.0047,  0.0021, -0.0047,
         0.0011, -0.0586,  0.0105,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.5104e-02,  1.9915e-03, -1.1897e-01, -8.6502e-02, -1.0423e-01,
        -2.3364e-03,  3.5333e-02,  2.4773e-02,  3.3258e-02, -4.2955e-02,
        -3.9298e-02,  2.7960e-02, -1.9287e-02, -2.3908e-02,  1.7998e-02,
        -4.0406e-02,  7.1349e-04, -6.1932e-03, -6.0590e-02,  1.3359e-02,
        -5.5963e-02, -3.3159e-05, -3.9452e-02, -4.8088e-02,  6.1762e-03,
         2.9564e-02, -2.2303e-02, -2.9899e-02,  3.3501e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2950, -0.1451, -0.0324,  0.0071, -0.0309, -0.0342,  0.0029, -0.0019,
        -0.0202, -0.0047, -0.0352, -0.0443, -0.0047, -0.0183, -0.0152, -0.0204,
        -0.0153,  0.0037, -0.0201, -0.0007, -0.0173, -0.0661, -0.0177, -0.0125,
         0.1341,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0126, -0.0264, -0.0716, -0.0508, -0.1218, -0.0782, -0.0604, -0.0936,
        -0.0416,  0.0102,  0.0066,  0.0652,  0.0311,  0.0178, -0.0781, -0.0638,
        -0.0320, -0.0186, -0.0273, -0.0303, -0.0476,  0.0038, -0.0106,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1005,  0.0030,  0.0018, -0.0268, -0.0015,  0.0349, -0.0311, -0.0234,
        -0.0041,  0.0049,  0.0015,  0.0033, -0.0274, -0.0680,  0.0207, -0.0454,
        -0.0079, -0.0109, -0.0209, -0.0414,  0.0308, -0.0540, -0.0018,  0.0323,
        -0.0032, -0.0401, -0.0978, -0.0062, -0.0378, -0.0282, -0.0131, -0.0270,
        -0.0341, -0.0111, -0.0085, -0.0043, -0.0130, -0.0201,  0.0575,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0589,  0.0529, -0.0352, -0.1019, -0.2251, -0.0392, -0.0700, -0.1379,
        -0.0750,  0.0660, -0.0603, -0.0774,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0036, -0.0616, -0.1038,  0.0020, -0.0184, -0.0459, -0.0637, -0.0092,
        -0.0123, -0.0113, -0.0110, -0.0014, -0.0014,  0.0075, -0.0274, -0.0027,
         0.0021, -0.0361,  0.0029, -0.0196, -0.0020, -0.0191, -0.0079,  0.0041,
         0.0063,  0.0025,  0.0049,  0.0308, -0.0637,  0.0529,  0.0135, -0.0008,
        -0.0277,  0.0077, -0.0310, -0.0322, -0.0029, -0.0256,  0.0079, -0.0111,
         0.0082,  0.0158,  0.0050,  0.0489, -0.1237,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0613,  0.1658, -0.0396,  0.0530,  0.0680,  0.0600,  0.1334,  0.0816,
         0.1229,  0.1308, -0.0492,  0.0344,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1669, -0.3435, -0.1453, -0.0280, -0.0634,  0.0776,  0.0627,  0.1126,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.2636e-02, -1.1177e-01, -1.2606e-03,  2.9802e-03, -1.1813e-03,
         2.9585e-03,  2.5396e-02,  2.7448e-03, -6.0418e-03, -1.0898e-01,
        -8.4245e-05,  1.3416e-02, -1.2259e-02, -2.2245e-03, -1.1215e-02,
         1.3109e-02,  2.7065e-02, -1.7169e-02, -8.2608e-04, -2.7487e-03,
         2.3288e-03,  1.1607e-02, -5.6810e-03, -9.1938e-03, -8.3220e-03,
        -7.2830e-03, -4.3503e-03, -2.0913e-02, -3.3842e-02,  9.8411e-04,
        -3.8137e-03, -1.1410e-02, -6.3394e-03, -1.6918e-02, -7.5559e-03,
        -4.0314e-03,  1.4071e-03, -6.3926e-03, -2.7042e-03,  4.0966e-03,
         5.8416e-04, -2.3781e-03,  3.6206e-03, -1.3706e-02, -1.6204e-02,
        -7.8429e-03, -1.1338e-01,  5.3026e-03, -3.7399e-02, -5.7499e-02,
        -8.4222e-03,  5.3899e-03, -1.5747e-03,  6.6866e-03, -2.3646e-02,
        -1.1407e-03, -2.7088e-02,  6.9457e-03, -7.0438e-03,  4.1847e-04,
         3.8492e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0765, -0.0584,  0.0059, -0.0265, -0.0923,  0.0210, -0.0344, -0.0158,
        -0.0115, -0.0270, -0.0654, -0.0354, -0.0053, -0.0117, -0.0792, -0.0036,
         0.0268, -0.0042, -0.0219, -0.0192, -0.0610, -0.0248, -0.0419, -0.0112,
        -0.0239, -0.0237, -0.0153, -0.0148, -0.0202, -0.0237, -0.0178, -0.0657,
        -0.0141,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.0338e-01, -7.2195e-02, -1.0000e-01, -3.2906e-02, -1.9298e-02,
        -1.8856e-02, -5.4564e-02,  2.3843e-02, -5.0411e-03,  1.1390e-02,
        -2.0043e-03, -2.4567e-02, -4.8075e-02, -1.0791e-03, -1.9958e-02,
         9.1782e-03,  4.4777e-06, -4.4293e-02, -2.1245e-01, -2.6572e-02,
         4.1554e-02, -3.2747e-02, -7.7677e-02, -1.8354e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0242, -0.0825, -0.0519, -0.0293, -0.0463, -0.0791, -0.0243, -0.0500,
        -0.0115, -0.0859, -0.0570, -0.0322, -0.1552, -0.0235, -0.0067, -0.0140,
         0.0023, -0.0121,  0.0517,  0.0275, -0.0088,  0.0107,  0.0704, -0.0427,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1013,  0.0019, -0.0713, -0.0042,  0.0102,  0.0200, -0.0158, -0.0309,
         0.0201, -0.0113, -0.0262, -0.0028,  0.0208,  0.0121, -0.0321, -0.0656,
        -0.0381, -0.0234, -0.0175, -0.0175, -0.0224, -0.0114, -0.0195, -0.1148,
         0.0214,  0.0168, -0.0545, -0.0147,  0.0163, -0.0116, -0.0558, -0.0222,
        -0.0074,  0.0682,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0619, -0.0705,  0.0346,  0.0155, -0.0659, -0.0224, -0.0113,  0.0041,
        -0.0177, -0.0145,  0.0465, -0.0658,  0.0500, -0.0961, -0.0370, -0.0081,
        -0.1274, -0.0296,  0.0133, -0.0493, -0.0498, -0.0803,  0.0049, -0.0159,
        -0.0076,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0090,  0.0578, -0.0598, -0.0180, -0.0786,  0.0008, -0.1139, -0.0106,
        -0.0585, -0.3832, -0.1147, -0.0155, -0.0063,  0.0137, -0.0128, -0.0469,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0954,  0.1374,  0.0072,  0.0596,  0.0286,  0.0393,  0.2195, -0.0179,
        -0.0400,  0.0008, -0.0293, -0.0882,  0.0277,  0.0387,  0.0031, -0.0021,
         0.0113,  0.0405,  0.0331,  0.0068,  0.0121, -0.0197,  0.0416,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0408,  0.2267,  0.0319,  0.0320,  0.3378,  0.1078, -0.0554,  0.0014,
        -0.0127, -0.0088, -0.0346,  0.0171,  0.0499,  0.0433,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.1629, -0.0590, -0.0310, -0.1262, -0.0058,  0.0122,  0.0014,  0.0326,
        -0.0158, -0.0092, -0.0544, -0.0179, -0.0297,  0.0101, -0.0068, -0.0254,
         0.0161, -0.1307, -0.0386, -0.0210, -0.0375, -0.0485,  0.0921,  0.0151,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1114, -0.0702, -0.2239, -0.0525, -0.0554, -0.0035, -0.2576, -0.1408,
        -0.0317, -0.0377, -0.0152,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0223, -0.0890, -0.0124, -0.0284, -0.0048, -0.0757, -0.1209,  0.0105,
        -0.1454, -0.0123, -0.0423, -0.0208, -0.0312,  0.0394, -0.0410,  0.0030,
        -0.0227, -0.0818, -0.0673, -0.0497, -0.0791,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1683, -0.0469, -0.0575,  0.0276, -0.0589, -0.0638, -0.0164, -0.1429,
        -0.0288, -0.0791,  0.0215, -0.0498, -0.0369, -0.2016,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0337,  0.1407, -0.0286, -0.0302,  0.0125,  0.0090, -0.0183, -0.0695,
        -0.0510,  0.0333,  0.0520,  0.0928,  0.1765,  0.0125,  0.0038, -0.0341,
        -0.0681,  0.0289, -0.0137,  0.0018,  0.0889,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2286, -0.3741, -0.0459,  0.0232, -0.0251, -0.0487, -0.0096, -0.0484,
        -0.0591,  0.0134, -0.0510, -0.0220, -0.0509,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0105, -0.0413,  0.0014,  0.0041,  0.0064, -0.0208, -0.0181, -0.0610,
        -0.0114, -0.0336, -0.0118,  0.0042,  0.0166,  0.0123, -0.0099, -0.0128,
        -0.0033,  0.0152,  0.0021, -0.0077, -0.0187, -0.0028, -0.0184, -0.0180,
        -0.1344, -0.0270, -0.0870, -0.0867, -0.0643,  0.0225, -0.0103,  0.0050,
         0.0170,  0.0024,  0.0346,  0.0215, -0.0164, -0.0003, -0.0041, -0.0090,
        -0.0162,  0.0095, -0.0166,  0.0463,  0.0065], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0910, -0.0293, -0.0566, -0.0273, -0.0162,  0.0133, -0.0041, -0.0117,
         0.0052, -0.0086, -0.0390,  0.0054,  0.0074,  0.0330, -0.0103, -0.0041,
        -0.1484, -0.0059,  0.0036,  0.0177, -0.0083, -0.0778, -0.0269, -0.0878,
         0.0133, -0.0461,  0.0009, -0.0211, -0.0581, -0.0244,  0.0364,  0.0247,
         0.0117,  0.0245,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0640, -0.0210, -0.0489, -0.0519, -0.0490, -0.0091, -0.0248, -0.0451,
         0.0506, -0.0259, -0.0075, -0.0156, -0.0157,  0.0288, -0.0114,  0.0413,
         0.0199,  0.0058,  0.0009, -0.0793, -0.2031, -0.0496, -0.0367,  0.0399,
         0.0519,  0.0023,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0819, -0.2753, -0.0138, -0.0123,  0.0126, -0.0671, -0.0139, -0.0794,
        -0.0485, -0.0167, -0.0305, -0.0092,  0.0048, -0.0155, -0.0171, -0.0348,
        -0.1091,  0.0360, -0.0187,  0.0175, -0.0252,  0.0282,  0.0319,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1494, -0.0546, -0.0037, -0.0161,  0.0007, -0.0394, -0.0062, -0.0030,
        -0.0263, -0.0029, -0.0002, -0.0483, -0.0095, -0.0129, -0.1118,  0.0538,
         0.0148, -0.0281, -0.0441, -0.0241,  0.0016, -0.0334, -0.0059, -0.0101,
        -0.0116, -0.0048, -0.0111, -0.0303,  0.0219,  0.0368,  0.0518, -0.0144,
         0.0218,  0.0079,  0.0077, -0.0025,  0.0002,  0.0221,  0.0342,  0.0200,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0904, -0.0795, -0.0106, -0.0344, -0.0554, -0.0096,  0.0140, -0.0406,
        -0.0124, -0.0567, -0.0181, -0.0277, -0.0474, -0.0282, -0.0234, -0.0428,
        -0.0240,  0.0171,  0.0284, -0.0104, -0.0151, -0.0333, -0.0056,  0.0124,
        -0.0058, -0.0362,  0.0054, -0.0449, -0.0056,  0.0140, -0.0914, -0.0006,
         0.0099, -0.0485,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0260,  0.0059,  0.0019, -0.0052, -0.0566, -0.0251, -0.0163,  0.0048,
        -0.0068,  0.0193, -0.0142, -0.0093,  0.0297, -0.0088, -0.1120, -0.0118,
        -0.0090,  0.0117, -0.0376,  0.0032, -0.0132, -0.0168, -0.0094,  0.0089,
        -0.0328, -0.0888, -0.0213, -0.0359, -0.0231, -0.0341,  0.0034, -0.0068,
         0.0151, -0.0230, -0.0207, -0.0029, -0.0091, -0.0205, -0.0180, -0.0108,
        -0.0125,  0.0076, -0.0111, -0.0089, -0.0094,  0.0026,  0.0101,  0.0038,
        -0.0194, -0.0136, -0.0030,  0.0680], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1079, -0.0103,  0.0689,  0.0208, -0.0847, -0.0177, -0.0012,  0.0393,
        -0.0061,  0.0121, -0.0013,  0.0116, -0.0109, -0.0567, -0.0279, -0.0396,
         0.0124,  0.0257, -0.0513,  0.0509, -0.0071, -0.0256, -0.0193, -0.1591,
         0.0161,  0.0202,  0.0095,  0.0111,  0.0085,  0.0042, -0.0621,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1141, -0.0638,  0.0451,  0.0085, -0.0177,  0.0274,  0.0130,  0.0141,
        -0.0405,  0.0006, -0.0070, -0.0473, -0.0119,  0.0553, -0.0141, -0.0468,
         0.1391, -0.0172, -0.0585, -0.0820, -0.0373, -0.0136, -0.0451,  0.0214,
         0.0585,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0896, -0.0798, -0.0339,  0.0423, -0.0752, -0.0855, -0.3236, -0.0047,
        -0.0386, -0.0250,  0.0757,  0.1261,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0436, -0.1492, -0.0239, -0.0339, -0.0484, -0.0471,  0.0290, -0.0088,
         0.0095, -0.0075, -0.0240,  0.0310, -0.0176, -0.1417,  0.0051, -0.0687,
        -0.0079,  0.0149,  0.0150, -0.0089, -0.0264, -0.0201,  0.0397,  0.0314,
         0.0273, -0.0627, -0.0121,  0.0133, -0.0313,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0404, -0.2392,  0.1037,  0.0389,  0.0719, -0.0700, -0.1308,  0.1218,
         0.1343,  0.0004, -0.0060, -0.0426,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0785, -0.3316,  0.1241, -0.0322,  0.0615, -0.1101, -0.0947, -0.0156,
        -0.0399, -0.0291, -0.0679, -0.0148,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1127,  0.0611, -0.0799, -0.0094, -0.1291, -0.0136,  0.0091, -0.0219,
        -0.0783, -0.0127, -0.0129, -0.1205, -0.0217,  0.0922, -0.0747, -0.0447,
         0.0417, -0.0157,  0.0281, -0.0201,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0836, -0.2835, -0.0100, -0.0042, -0.0790,  0.0069,  0.0041, -0.0065,
         0.0237,  0.0282,  0.0070, -0.0105,  0.0019, -0.0237, -0.0200,  0.0093,
        -0.0226,  0.0122,  0.0109, -0.0332, -0.0056,  0.0056,  0.0074,  0.0316,
         0.0105, -0.0060,  0.0100,  0.0012, -0.0261, -0.0087, -0.0057, -0.0128,
        -0.0115, -0.0385, -0.0244,  0.0631, -0.0503,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1050,  0.5009,  0.0402,  0.0714, -0.0402, -0.0312,  0.0558, -0.1217,
        -0.0336,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0974, -0.0294, -0.0583, -0.0034, -0.0093,  0.0102, -0.0576, -0.1003,
         0.0060, -0.0323, -0.0083, -0.0056, -0.0615, -0.1592,  0.0209,  0.0418,
        -0.0123,  0.0528, -0.0395, -0.0704, -0.0430,  0.0060, -0.0134, -0.0031,
         0.0230, -0.0162, -0.0186,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0605,  0.2396,  0.0063, -0.0044, -0.0067,  0.0331,  0.0070,  0.0071,
         0.0027,  0.0020, -0.0236,  0.0411,  0.0155, -0.0206,  0.0139, -0.0438,
         0.0429, -0.0213, -0.0420, -0.0114, -0.0460,  0.0239, -0.0029, -0.0246,
         0.0144, -0.0277,  0.0081,  0.0144,  0.1194, -0.0005, -0.0042, -0.0463,
         0.0068, -0.0153,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0361, -0.0060,  0.0009, -0.0003, -0.0203, -0.0100,  0.0063,  0.0203,
        -0.0125, -0.0025, -0.0069, -0.0985, -0.0185, -0.0225, -0.0002, -0.0211,
        -0.0781,  0.0161,  0.0039,  0.0072,  0.0189, -0.0130, -0.0350,  0.0053,
         0.0042, -0.0204, -0.0059, -0.0291,  0.0308, -0.0200, -0.0279, -0.0577,
        -0.0697,  0.0070,  0.0031, -0.0303, -0.0162,  0.0214,  0.0144, -0.0094,
        -0.0153,  0.0253,  0.0217, -0.0414,  0.0685,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0034, -0.0870, -0.1043, -0.0718, -0.0691, -0.0506, -0.1429, -0.0049,
        -0.0543, -0.0650, -0.0380, -0.0433, -0.0631,  0.1193, -0.0756,  0.0073,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2567,  0.0489, -0.0229, -0.0370, -0.0056, -0.0212, -0.0180,  0.0294,
        -0.0512, -0.0528, -0.0251, -0.0543, -0.0191, -0.0162, -0.0350,  0.0272,
        -0.0454, -0.0341,  0.0121,  0.0081, -0.0259,  0.0433, -0.0139,  0.0089,
         0.0064,  0.0487, -0.0325,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0676, -0.0748, -0.0150, -0.0080, -0.0152, -0.0126, -0.0240, -0.0107,
        -0.0170, -0.0085,  0.0044, -0.0088, -0.0051, -0.0106, -0.0038,  0.0111,
        -0.0101,  0.0144, -0.0093, -0.0290,  0.0135, -0.0329, -0.0265, -0.0766,
         0.0110,  0.0133, -0.0162, -0.0465,  0.0131, -0.0652, -0.0173, -0.0373,
        -0.0160,  0.0133, -0.0378, -0.0057,  0.0057, -0.0541, -0.0003,  0.0284,
        -0.0167, -0.0367, -0.0044,  0.0029, -0.0161,  0.0323], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0451,  0.0217, -0.0130, -0.0283, -0.0089, -0.0827, -0.0089, -0.0356,
        -0.0129, -0.0249, -0.0200, -0.0002, -0.0021,  0.0050, -0.0051,  0.0207,
        -0.0136, -0.0435, -0.0947, -0.1972, -0.0406, -0.0221, -0.0110, -0.0165,
        -0.0194, -0.0196, -0.0489, -0.0792,  0.0139,  0.0447,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2226,  0.0307,  0.0868,  0.0552, -0.0223,  0.1207,  0.1765,  0.0790,
        -0.1937, -0.0107,  0.0017,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1248, -0.1789,  0.0132,  0.0461, -0.0249, -0.0405, -0.0491, -0.1176,
        -0.0246, -0.0168,  0.0029,  0.0095, -0.1005, -0.0341,  0.0447, -0.0635,
        -0.0138, -0.0122, -0.0367, -0.0455,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2107,  0.1393,  0.1155,  0.0717,  0.0699,  0.1052,  0.1167, -0.0512,
        -0.0074,  0.0582,  0.0542,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0387, -0.1877, -0.0584, -0.1427, -0.1199,  0.0398, -0.0268, -0.0242,
        -0.0689, -0.0318,  0.0058, -0.0074, -0.0278, -0.0068, -0.0186, -0.0422,
         0.0334,  0.0119, -0.0320, -0.0246, -0.0505,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.2291e-01,  1.3252e-01,  3.2281e-02, -6.4923e-02,  2.6161e-02,
         6.8314e-02, -1.1864e-01,  4.3177e-02,  3.3172e-02,  1.0800e-01,
         1.5801e-02,  2.7805e-03,  3.3449e-02,  2.5823e-02,  7.8160e-03,
         5.7597e-02, -1.8995e-05, -1.0661e-01,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0372, -0.0486, -0.0668,  0.0324, -0.0030, -0.0013,  0.0595,  0.0056,
        -0.0071, -0.0027,  0.0247,  0.0031, -0.0154, -0.0190, -0.0459,  0.0202,
        -0.0062, -0.0466, -0.0031, -0.0372, -0.0323, -0.0416, -0.0746, -0.0086,
        -0.0269, -0.0020, -0.0203, -0.0159, -0.0145, -0.0812, -0.0199, -0.0128,
        -0.0186,  0.0060, -0.0229, -0.0127,  0.0005, -0.0740, -0.0292,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0361, -0.3407, -0.0215, -0.0292, -0.0204,  0.0118, -0.0437,  0.0330,
        -0.0217, -0.0246, -0.0215, -0.0102,  0.0206,  0.0057,  0.0124, -0.0182,
        -0.0508, -0.0341, -0.0055, -0.0030, -0.0159, -0.0177, -0.0039,  0.0033,
        -0.0171, -0.0150, -0.0305,  0.0057, -0.0413, -0.0013, -0.0055,  0.0038,
        -0.0060,  0.0683,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0255, -0.3486,  0.1634, -0.0938,  0.0697, -0.0610,  0.0032, -0.2145,
        -0.0203,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1176,  0.1167, -0.0370, -0.0038,  0.0113,  0.0386,  0.0032,  0.0046,
        -0.0192,  0.0037,  0.0344,  0.0106,  0.0030, -0.0012, -0.0060,  0.0829,
         0.0109,  0.1801,  0.0009, -0.0092, -0.0164,  0.0092,  0.0279, -0.0107,
         0.0218,  0.0055, -0.0056,  0.0137, -0.0062, -0.0019, -0.0081,  0.0319,
         0.0184,  0.0108,  0.0094,  0.0098, -0.0052, -0.0094, -0.0194, -0.0634,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0044, -0.0540, -0.1741,  0.0058, -0.0036,  0.0058, -0.0097,  0.0029,
        -0.0136, -0.0707, -0.0066,  0.0049,  0.0012,  0.0017, -0.0131,  0.0080,
         0.0139, -0.0166, -0.0397,  0.0035, -0.0144, -0.0161,  0.0235, -0.0125,
         0.0234, -0.0150, -0.0323, -0.0040, -0.0228, -0.0627, -0.0202, -0.0142,
        -0.0030,  0.0013,  0.0031, -0.0351, -0.0204, -0.0434, -0.0023,  0.0130,
        -0.0053,  0.0184, -0.0015, -0.0085,  0.0033, -0.0018,  0.0320, -0.0062,
         0.0326, -0.0117,  0.0173, -0.0251,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0198, -0.1727, -0.0477, -0.0302, -0.0078, -0.1177, -0.0390, -0.0005,
        -0.0210, -0.0048, -0.0312, -0.0681, -0.0282, -0.0216, -0.0028, -0.0787,
        -0.0350, -0.0103, -0.0756, -0.0527, -0.0987, -0.0358,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0248,  0.3593,  0.0898,  0.0399,  0.0329, -0.0359,  0.0389,  0.0286,
        -0.0252,  0.0292,  0.0047,  0.0314,  0.0018,  0.0186, -0.0301,  0.0691,
         0.0121, -0.0266, -0.0285,  0.0142,  0.0586,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0293, -0.0382,  0.0097, -0.0185,  0.0017,  0.0107, -0.0069,  0.0100,
        -0.0254, -0.0154, -0.0062, -0.0057, -0.0028,  0.0040, -0.0172,  0.0071,
        -0.0017,  0.0023, -0.0185,  0.0031, -0.0105,  0.0020, -0.0104,  0.0023,
         0.0114,  0.0002, -0.0264, -0.0280, -0.0023,  0.0005, -0.0460, -0.0267,
        -0.0133, -0.0126, -0.0172, -0.0040, -0.0166,  0.0714, -0.0309, -0.0302,
        -0.0494, -0.0289, -0.0103, -0.0464, -0.0391, -0.0073, -0.0039, -0.0270,
         0.0311,  0.0278, -0.0312,  0.0011, -0.0222, -0.0572, -0.0197],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0290, -0.0083,  0.0176, -0.0008,  0.0160,  0.0003, -0.0334,  0.0145,
        -0.0029,  0.0248,  0.0475, -0.0092, -0.0021, -0.0051, -0.0118, -0.0143,
        -0.0199, -0.0262, -0.0274,  0.0199,  0.0027, -0.0141, -0.0299, -0.0122,
         0.0281, -0.0180, -0.0451, -0.0357, -0.0394, -0.0201, -0.0718,  0.0038,
        -0.0264, -0.0847, -0.0115, -0.0339,  0.0048, -0.0459, -0.0182, -0.0319,
        -0.0198, -0.0024,  0.0223, -0.0300, -0.0164,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1259,  0.0890,  0.0627,  0.0698,  0.0211,  0.0171,  0.0288,  0.0104,
         0.0588, -0.0238, -0.0083,  0.0449,  0.0472,  0.0074,  0.0361, -0.0010,
        -0.0062,  0.1103,  0.0055,  0.0250,  0.1126,  0.0362,  0.0262,  0.0259,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1162, -0.0829,  0.0030,  0.0163, -0.0265, -0.0171,  0.0123, -0.0129,
         0.0194,  0.0214, -0.0058,  0.0071,  0.0007, -0.0666, -0.0447, -0.0414,
        -0.0556, -0.0225, -0.0314, -0.0551,  0.0026,  0.0022, -0.0269, -0.0719,
         0.0539, -0.0246, -0.0606, -0.0071, -0.0065, -0.0594,  0.0104,  0.0121,
        -0.0029,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1042, -0.2998, -0.0149, -0.0194, -0.0310, -0.0100,  0.0515,  0.0443,
        -0.1740,  0.0037,  0.0065,  0.0316,  0.0394,  0.0038,  0.0321,  0.0183,
         0.0179,  0.0773,  0.0205,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.7752e-02, -2.4798e-02,  1.2261e-02, -4.9373e-02, -1.0149e-01,
        -5.1035e-02, -4.7929e-02, -2.2191e-02, -9.9648e-02,  1.4833e-02,
        -1.4305e-02, -1.4272e-02,  1.1523e-04, -5.7218e-03,  1.3410e-02,
         3.0675e-02,  5.0673e-02, -3.7383e-02, -8.9314e-03,  1.4081e-02,
        -1.6359e-02, -8.4847e-02, -2.7574e-02, -4.0235e-02,  1.2011e-01,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1096,  0.1557,  0.0834,  0.0246,  0.0285,  0.0684,  0.2219,  0.0402,
         0.0456,  0.0267,  0.0068,  0.0080,  0.0273,  0.0171,  0.0377, -0.0848,
        -0.0138,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0818,  0.0253, -0.0406, -0.0194,  0.0025, -0.0399,  0.0416, -0.0267,
         0.0007, -0.0138, -0.0091, -0.0235, -0.0018, -0.0021, -0.0260, -0.0175,
        -0.0443, -0.0815, -0.0080, -0.0367, -0.0093, -0.0881, -0.0688, -0.0235,
        -0.0414, -0.0565,  0.0100,  0.0345, -0.0385,  0.0062, -0.0586,  0.0216,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0288, -0.1023, -0.0579, -0.0383, -0.0203, -0.0208, -0.0540, -0.0743,
        -0.0239, -0.0152,  0.0169,  0.0056, -0.0338, -0.0215, -0.0095,  0.0349,
        -0.0309, -0.0201, -0.0269, -0.0008,  0.0008, -0.0217, -0.0245, -0.1008,
         0.0121, -0.0508, -0.1511,  0.0014,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0529, -0.2083, -0.0240, -0.0185,  0.0009, -0.0361, -0.1455,  0.0034,
        -0.0275, -0.0508,  0.0059, -0.0352,  0.0102, -0.0764, -0.0027, -0.0424,
        -0.0139, -0.0377, -0.0140, -0.0211,  0.0027,  0.0017, -0.0179, -0.0059,
        -0.0234,  0.0231, -0.0028,  0.0598, -0.0354,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0344,  0.0325, -0.0063, -0.0162, -0.1244, -0.0090, -0.0286, -0.0155,
        -0.0038, -0.0372,  0.0269, -0.0213, -0.0256, -0.0414, -0.0269, -0.0405,
        -0.0316, -0.0482, -0.0153, -0.0152, -0.0395, -0.0170, -0.0404, -0.0062,
        -0.0375, -0.0298, -0.0293, -0.0042, -0.0373,  0.0195, -0.0134, -0.0415,
        -0.0835,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0209, -0.1853, -0.0278,  0.0025,  0.0016, -0.0497,  0.0276, -0.0391,
        -0.0288, -0.0258, -0.0512, -0.0250, -0.0030, -0.0247, -0.0268, -0.0054,
        -0.0392, -0.0378, -0.1673, -0.0798,  0.0343, -0.0179, -0.0088, -0.0697,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0318, -0.0858, -0.0455, -0.0019, -0.0180, -0.0587, -0.0639, -0.0428,
         0.0086, -0.0400, -0.0308, -0.0455, -0.0416, -0.0335, -0.0200,  0.0100,
        -0.0046, -0.0379, -0.0129, -0.2394, -0.0157,  0.0664,  0.0446,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0444, -0.0608, -0.0055,  0.0005, -0.0011, -0.0567, -0.0011, -0.0502,
        -0.0083, -0.0041, -0.0144, -0.0138, -0.0136,  0.0195,  0.0123, -0.0106,
        -0.0682, -0.0216, -0.0232,  0.0107,  0.0255, -0.0730,  0.0249, -0.0119,
        -0.0473, -0.0187,  0.0044, -0.0219, -0.0282, -0.0632, -0.0351, -0.0425,
        -0.0224, -0.0204, -0.0295, -0.0408, -0.0080, -0.0384, -0.0035,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0372, -0.0595, -0.0158,  0.0002, -0.0230,  0.0034, -0.0051,  0.0209,
        -0.0396,  0.0025,  0.0061, -0.0787, -0.0084, -0.0118,  0.0027, -0.1122,
        -0.0106, -0.0087, -0.0039, -0.0276, -0.0556, -0.0075, -0.0063, -0.0895,
        -0.0330, -0.0276, -0.0020, -0.0023, -0.0081, -0.0244, -0.0044, -0.0132,
        -0.0461, -0.0255, -0.0097, -0.0181, -0.0115, -0.0052, -0.0158, -0.0063,
        -0.0043,  0.0736, -0.0320,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.3093e-03, -1.8356e-01, -1.3888e-02,  1.9149e-03,  8.6763e-03,
        -5.0486e-03, -1.6345e-02,  4.5470e-05, -5.9235e-02,  1.0523e-02,
         1.0286e-02,  5.0745e-03, -1.1245e-02,  7.1228e-03, -3.9579e-03,
        -1.9071e-02, -1.5236e-02, -6.4039e-02,  1.2691e-02, -1.7985e-02,
         1.4389e-02, -1.3880e-03, -2.1087e-02, -1.3268e-02, -1.0424e-02,
        -5.0791e-03,  2.7296e-02,  1.4080e-02, -3.2203e-02, -1.2198e-02,
        -1.2487e-02, -1.7630e-03, -3.5703e-02,  1.6647e-03, -7.8519e-03,
         1.2342e-02, -1.5894e-02,  8.7601e-03, -3.9174e-02, -4.5876e-03,
        -1.8124e-02, -4.9521e-02, -2.8505e-03, -1.6713e-04, -6.5348e-03,
        -1.5847e-02, -7.6198e-03,  3.9502e-03, -1.0350e-03, -1.2173e-02,
        -2.5086e-02,  2.2228e-03, -1.4645e-02, -1.9834e-03,  9.3497e-03,
        -2.5444e-02, -4.1550e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0286, -0.4353, -0.0203,  0.0092, -0.0510,  0.0343, -0.0708, -0.1170,
        -0.0161,  0.0378, -0.0445,  0.0274, -0.0391, -0.0687,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0372, -0.1301, -0.0255, -0.0023,  0.0058, -0.0229, -0.0064, -0.0184,
        -0.0160,  0.0002, -0.0179, -0.0080,  0.0480,  0.0285, -0.0585, -0.0433,
        -0.0188, -0.0349,  0.0240, -0.0177, -0.2002, -0.0116,  0.0084, -0.0076,
         0.0162, -0.0380,  0.0107,  0.0187,  0.1131, -0.0111,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1876,  0.1421, -0.0113,  0.0475, -0.0271,  0.0104,  0.1228, -0.4510,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0671, -0.1095,  0.0066, -0.0205, -0.0270, -0.0915,  0.0135, -0.0069,
         0.0171, -0.0097, -0.0195, -0.0573, -0.0485, -0.0674, -0.0116, -0.0086,
         0.0027, -0.0083,  0.0013, -0.0085,  0.0296, -0.0012, -0.0069, -0.1030,
        -0.0338, -0.0260, -0.0310,  0.0165, -0.0430, -0.0485,  0.0160,  0.0346,
         0.0068,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0022, -0.1278, -0.0511,  0.0380, -0.0050, -0.0315, -0.0012, -0.0071,
        -0.0260, -0.0151,  0.0020,  0.0080, -0.0149, -0.0065, -0.0039, -0.0155,
        -0.0018,  0.0012, -0.0101, -0.0033, -0.0014,  0.0064, -0.0040, -0.0064,
         0.0051, -0.0262, -0.0083, -0.0581, -0.0072,  0.0001, -0.0055, -0.0004,
        -0.0221, -0.0437,  0.0071, -0.0226, -0.0176,  0.0046, -0.0085,  0.0101,
         0.0045, -0.0069, -0.0154, -0.0056,  0.0331,  0.0154, -0.0088, -0.0007,
        -0.0080, -0.0009, -0.0061, -0.0084, -0.0053, -0.0010,  0.0054,  0.0002,
        -0.0025, -0.0146,  0.0021, -0.0111, -0.0053, -0.0084, -0.0419, -0.0286,
        -0.0051, -0.0010,  0.0085,  0.0434, -0.0024, -0.0394, -0.0223],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0491, -0.0029, -0.0120,  0.0054, -0.0332, -0.0312, -0.0918, -0.0106,
        -0.0159, -0.0043,  0.0035, -0.0197,  0.0239,  0.0159, -0.0118, -0.0035,
        -0.0011,  0.0117, -0.0360, -0.0671,  0.0265,  0.0110,  0.0734, -0.0746,
        -0.0030, -0.0264, -0.0320,  0.0179,  0.0144, -0.0203,  0.0251, -0.0050,
        -0.0204, -0.0186, -0.0314, -0.0138, -0.0404,  0.0006, -0.0051,  0.0116,
        -0.0157, -0.0623,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.0638e-02, -1.2689e-01, -2.2029e-02,  1.0087e-02, -1.2880e-03,
         5.0530e-03,  4.8959e-03,  1.1404e-02,  2.5935e-02, -4.5479e-02,
        -1.7710e-02, -2.6136e-02, -7.5852e-02, -2.0476e-02,  8.4277e-05,
         1.1011e-02,  5.4416e-03, -3.7545e-02,  5.4865e-03, -1.6335e-02,
        -2.8428e-02, -2.0930e-02, -1.0332e-02, -1.4642e-02, -1.7688e-03,
        -4.3903e-02, -9.5373e-03, -1.8830e-02, -9.3638e-02, -1.4753e-02,
        -2.0732e-02,  5.2646e-03,  2.8173e-02, -3.8539e-02, -1.8785e-02,
         2.8193e-02, -7.3779e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2343, -0.3550,  0.0661, -0.0288,  0.0129,  0.0370, -0.0544, -0.0653,
        -0.0142,  0.0119,  0.0141, -0.0148, -0.0402, -0.0309,  0.0201,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0485,  0.1056, -0.0549,  0.0154,  0.0592,  0.0254,  0.0353,  0.0329,
         0.1374,  0.0243, -0.0192, -0.1184,  0.0713, -0.0186,  0.0104, -0.0458,
         0.0059,  0.0482, -0.0384, -0.0848,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0513,  0.0268, -0.0066,  0.0354, -0.1766, -0.0076, -0.0068, -0.0163,
         0.0022, -0.0199, -0.0129, -0.0394, -0.0287,  0.0028, -0.0278, -0.0221,
         0.0085, -0.0192,  0.0089, -0.0259,  0.0088,  0.0151,  0.0062, -0.0029,
         0.0105, -0.0234, -0.0628, -0.0020,  0.0057, -0.0159, -0.0441, -0.0189,
        -0.0012, -0.0160, -0.0157, -0.0475, -0.0067, -0.0067, -0.0172, -0.0170,
         0.0185,  0.0760, -0.0156,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0115, -0.0311,  0.0487, -0.0274, -0.0098, -0.0406, -0.0191, -0.1534,
        -0.0319,  0.0044, -0.0011,  0.0415, -0.0172, -0.0147, -0.0625, -0.0004,
        -0.0061,  0.0280,  0.0048, -0.0308, -0.0191, -0.0141,  0.0016,  0.0053,
        -0.0006,  0.0225,  0.0436,  0.0373, -0.0775,  0.0083, -0.0021, -0.0313,
        -0.0079, -0.0230, -0.0326,  0.0315, -0.0418,  0.0147,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0652, -0.1479, -0.0563, -0.0949, -0.0171, -0.1262, -0.0227,  0.0454,
        -0.0054, -0.0720,  0.0053,  0.0007, -0.0511, -0.0153,  0.0616, -0.0326,
         0.0188, -0.0612,  0.0312, -0.0341,  0.0146,  0.0207,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0665, -0.0950, -0.0017, -0.0252,  0.0216, -0.0114,  0.0041,  0.0163,
         0.0394, -0.0449, -0.0122,  0.0006, -0.0079, -0.0095, -0.0287, -0.0168,
        -0.1817, -0.0368, -0.0402, -0.0179,  0.0148,  0.0245, -0.0195, -0.0265,
         0.0049, -0.0131,  0.0242, -0.0123, -0.0089, -0.0390, -0.0149, -0.0037,
         0.0080, -0.0123, -0.0287, -0.0044,  0.0167, -0.0170, -0.0078, -0.0207,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0645, -0.0096, -0.0301, -0.1269, -0.0097, -0.0247,  0.0030,  0.0005,
        -0.0608, -0.0230, -0.0015, -0.0057, -0.0145, -0.0776, -0.0490, -0.0089,
        -0.0421,  0.0465, -0.0761,  0.0052, -0.0125, -0.0177, -0.0328, -0.0296,
        -0.0296, -0.0513,  0.0180, -0.0427, -0.0385,  0.0026, -0.0161,  0.0286,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0781, -0.2544, -0.0007,  0.0294, -0.0062, -0.0098, -0.0246, -0.0454,
        -0.0758, -0.0217,  0.0102, -0.0298, -0.0131, -0.0231, -0.0208, -0.0467,
        -0.0224,  0.0003,  0.0113,  0.0233,  0.0014, -0.0071, -0.0103, -0.0556,
         0.0032,  0.0210, -0.0102, -0.0013,  0.0216, -0.0158,  0.0096, -0.0011,
        -0.0253,  0.0445, -0.0248,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([0.0370, 0.0319, 0.0605, 0.0523, 0.0276, 0.1474, 0.1057, 0.0966, 0.0548,
        0.0762, 0.0118, 0.0306, 0.0226, 0.0512, 0.0089, 0.0112, 0.0388, 0.0166,
        0.0252, 0.0611, 0.0319, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0598, -0.3384,  0.0487, -0.0474, -0.0571, -0.0392, -0.0216, -0.0470,
        -0.0749, -0.0145,  0.0050, -0.0136,  0.0185,  0.0436, -0.0302,  0.0287,
        -0.0384,  0.0270, -0.0463,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0162, -0.2862, -0.0681, -0.0060, -0.0273, -0.0198, -0.0126, -0.0265,
        -0.0356, -0.0805, -0.0580, -0.0408, -0.0064, -0.0977, -0.0143,  0.0040,
        -0.0179, -0.0155,  0.0574, -0.1092,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1573, -0.2311, -0.0056, -0.0811, -0.0301, -0.0309, -0.0169, -0.0272,
        -0.0096, -0.0161, -0.0043, -0.0255, -0.0050, -0.0339, -0.0331, -0.0217,
        -0.0568, -0.0044,  0.0014, -0.0671,  0.0199,  0.0969,  0.0240,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0158, -0.2023, -0.0353,  0.0113, -0.0044, -0.0264,  0.0218, -0.0153,
        -0.0239, -0.1544,  0.0029, -0.0154, -0.0057, -0.1668,  0.0244,  0.0331,
         0.0139, -0.0270, -0.0455, -0.0348, -0.0431,  0.0167,  0.0149,  0.0067,
        -0.0381,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0527,  0.2370,  0.0116,  0.0117,  0.0039,  0.0060, -0.0253,  0.0394,
         0.0021, -0.0357,  0.0576,  0.0126,  0.0061, -0.0011, -0.0042, -0.0539,
         0.0069, -0.0132, -0.0264,  0.0090,  0.0161,  0.0451,  0.0342,  0.0212,
         0.0255,  0.0188, -0.0141, -0.0025, -0.0303, -0.0236, -0.0202, -0.0282,
         0.0198,  0.0634, -0.0205,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1099,  0.0022,  0.0256, -0.0920,  0.0055,  0.0151, -0.0371, -0.0231,
        -0.1563, -0.0256,  0.0168, -0.0034, -0.0303, -0.0438, -0.0172, -0.0909,
        -0.0291, -0.0613, -0.0660, -0.0198,  0.0586, -0.0704,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1751,  0.0722,  0.0243,  0.0175,  0.0918,  0.0424,  0.0291,  0.0150,
         0.0239,  0.0915,  0.0151,  0.0491,  0.0486,  0.0529,  0.0320,  0.0111,
        -0.0259,  0.0891, -0.0069,  0.0470,  0.0395,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0922, -0.2784,  0.0111,  0.0034,  0.0171, -0.0475, -0.0492, -0.1232,
        -0.0245,  0.0278,  0.0440,  0.0464, -0.0157, -0.0790,  0.0513, -0.0458,
         0.0043,  0.0210,  0.0184,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0291, -0.0304, -0.0574, -0.0013, -0.0180, -0.0647, -0.0155,  0.0081,
         0.0037, -0.0018, -0.0097, -0.0043,  0.0107, -0.0341, -0.0116,  0.0084,
        -0.0092, -0.0379,  0.0038, -0.0214,  0.0141, -0.0341, -0.0138, -0.0045,
        -0.0230,  0.0100,  0.0085, -0.0665, -0.0523, -0.0708, -0.0016, -0.0168,
        -0.0381, -0.0786, -0.0061, -0.0047,  0.0018,  0.0184,  0.0456, -0.0009,
         0.0025, -0.0068,  0.0056,  0.0021, -0.0177, -0.0235,  0.0501],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1309, -0.0533, -0.0549, -0.0101, -0.0004, -0.0199,  0.0034,  0.0102,
         0.0024, -0.0122, -0.0090, -0.0044,  0.0109, -0.0192,  0.0390, -0.0098,
        -0.0327,  0.0004,  0.0271, -0.0073, -0.0012, -0.0157, -0.0612, -0.1402,
        -0.0054,  0.0077, -0.1483,  0.0104,  0.0053,  0.0170,  0.0075,  0.0119,
         0.0429,  0.0679,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0958,  0.3303,  0.0172, -0.0458, -0.0473,  0.0198,  0.0067,  0.0059,
         0.0169,  0.0077, -0.0204, -0.0071,  0.0171,  0.0669, -0.0056, -0.0300,
        -0.0079,  0.0316,  0.0335,  0.0322,  0.0081,  0.0308,  0.0034,  0.0204,
         0.0048, -0.0010,  0.0092,  0.0140,  0.0033,  0.0192, -0.0027, -0.0042,
         0.0217,  0.0112,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.1482, -0.2872, -0.0044, -0.0419, -0.0159, -0.0566, -0.0014,  0.0229,
        -0.0579, -0.0491, -0.0491, -0.0418, -0.0374, -0.0709, -0.0529,  0.0075,
        -0.0296,  0.0253,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1037, -0.0798, -0.0595, -0.0246, -0.0361,  0.0089, -0.0190, -0.0006,
        -0.0084, -0.1585,  0.0278, -0.0192, -0.0124, -0.0441, -0.0471, -0.0332,
        -0.0264, -0.0033,  0.0223, -0.1138,  0.0145,  0.0112, -0.0101,  0.0003,
        -0.0064, -0.0531, -0.0557,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0559,  0.0022, -0.0202, -0.0133, -0.0382, -0.0646, -0.0072,  0.0165,
        -0.0669, -0.0121, -0.0028, -0.0636, -0.0692, -0.0640, -0.0077,  0.0049,
        -0.0303, -0.0500,  0.0065, -0.2190, -0.0444, -0.0108,  0.0440,  0.0171,
        -0.0013,  0.0673,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0454, -0.2517, -0.0319, -0.0453, -0.0816,  0.0053, -0.0097,  0.0143,
        -0.0487, -0.0401, -0.0429, -0.0486, -0.0366, -0.0237, -0.0788, -0.0478,
        -0.0332, -0.0240,  0.0235, -0.0443,  0.0224,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1184,  0.1971,  0.0623,  0.0923,  0.0079,  0.0880,  0.0769,  0.0525,
         0.0088, -0.0135,  0.0588,  0.0264, -0.0536,  0.1307, -0.0128,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.2314e-01,  2.4189e-02, -5.7246e-02, -5.9829e-02, -1.2574e-02,
         9.7332e-02, -2.7579e-02, -2.3212e-02,  9.6568e-03, -5.4051e-03,
        -1.9302e-02,  8.7781e-02,  1.7148e-01,  1.0354e-01,  1.4682e-01,
        -1.0436e-04,  6.4477e-03,  2.4360e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0445, -0.1333, -0.0592, -0.1418, -0.0195, -0.0380, -0.0663, -0.0227,
        -0.0367, -0.1013, -0.0651, -0.0246,  0.0278, -0.0623, -0.0319,  0.0038,
        -0.0093,  0.0290,  0.0831,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0461, -0.0684,  0.0097,  0.0594, -0.0098, -0.0085, -0.0093,  0.0056,
        -0.0101,  0.0041,  0.0038, -0.0100, -0.0126,  0.0065, -0.0005, -0.0055,
        -0.0285, -0.0122,  0.0425, -0.0167, -0.1240, -0.0040, -0.0428, -0.0142,
        -0.0583, -0.0293, -0.0342, -0.0075, -0.0002, -0.0174,  0.0010, -0.0052,
        -0.0467, -0.0120, -0.0308, -0.0196, -0.0462,  0.0182, -0.0145, -0.0260,
         0.0137,  0.0377,  0.0032, -0.0236], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.8051e-02,  1.4333e-02,  3.2580e-02,  1.3519e-05, -2.8421e-02,
        -5.3497e-03, -4.0506e-02, -8.9576e-03,  1.4548e-03,  2.0033e-02,
        -1.7967e-02,  5.1450e-02,  1.9941e-03,  5.7123e-03,  2.8450e-03,
         1.1827e-02, -2.2696e-03, -1.0201e-01, -1.1471e-02, -1.5320e-02,
        -2.4846e-02,  8.3002e-03, -1.4882e-02,  1.5754e-02,  3.4058e-02,
        -2.1943e-03,  2.5442e-02, -1.0966e-01, -3.3926e-04, -4.1114e-02,
        -6.0140e-02,  9.1440e-03,  9.8449e-03, -2.2043e-01,  1.2816e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0524, -0.3035, -0.0238, -0.0130, -0.0394, -0.0312, -0.0820, -0.0369,
        -0.0019, -0.0052,  0.0113, -0.0098, -0.1092,  0.0087,  0.0335, -0.0575,
        -0.0449, -0.0070,  0.0197, -0.0080, -0.0276, -0.0184, -0.0237, -0.0315,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0930, -0.2041, -0.0288,  0.0088,  0.0066, -0.0224, -0.0138,  0.0136,
        -0.0433, -0.0090, -0.0168,  0.0073,  0.0246, -0.0367, -0.0611, -0.0102,
        -0.0069, -0.0056, -0.0835, -0.0047, -0.0101, -0.0332, -0.0188, -0.0324,
        -0.0064, -0.0219, -0.0049, -0.0087, -0.0017,  0.0024, -0.0378, -0.0149,
        -0.0919,  0.0142,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0754, -0.0456,  0.0096, -0.0102,  0.0261,  0.0259, -0.1293, -0.0386,
        -0.0104,  0.0028,  0.0213,  0.0156, -0.0501, -0.0201, -0.0362, -0.2441,
        -0.0087,  0.0163, -0.0314,  0.0033,  0.0143, -0.0158,  0.0066, -0.0373,
         0.0009,  0.0051, -0.0286, -0.0111,  0.0119,  0.0473,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0236, -0.1421, -0.0885, -0.2576, -0.0263, -0.0262,  0.0069, -0.0250,
        -0.0557, -0.1790, -0.0511,  0.0433,  0.0748,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1894,  0.0009, -0.0527, -0.0807, -0.0061, -0.0417, -0.0029, -0.0249,
        -0.1767, -0.0518, -0.0140, -0.0569,  0.0254, -0.0198, -0.0467, -0.0093,
        -0.0551,  0.1448,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1926,  0.2955,  0.0772,  0.0389,  0.1364,  0.1034,  0.1339,  0.0220,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0685, -0.0926, -0.0172, -0.0003, -0.0129, -0.0256, -0.0112, -0.0076,
         0.0229,  0.0063, -0.0086,  0.0010, -0.0010, -0.0008, -0.0017, -0.0066,
         0.0024, -0.0014,  0.0043,  0.0061,  0.0010,  0.0063, -0.0197, -0.0412,
        -0.0087, -0.0432, -0.0292, -0.0273, -0.1045, -0.0091, -0.0261, -0.0373,
        -0.0017, -0.0861, -0.0157, -0.0138, -0.0189, -0.0166,  0.0227, -0.0012,
        -0.0008,  0.0029,  0.0004,  0.0164,  0.0583,  0.0573,  0.0346],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0918,  0.0768,  0.1336, -0.0401, -0.0852, -0.0020,  0.1736, -0.0461,
         0.1009, -0.0781, -0.1717,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1604,  0.1815,  0.0415,  0.0343,  0.0294,  0.0321,  0.0256,  0.1936,
         0.0779,  0.0381,  0.0332, -0.0045,  0.0642, -0.0161, -0.0182,  0.0292,
         0.0154,  0.0048,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0328,  0.0794,  0.2718,  0.1435, -0.0523,  0.0092,  0.0322, -0.0217,
         0.0262, -0.0393, -0.0173, -0.0405,  0.0539, -0.1111,  0.0687,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0698, -0.2839, -0.1174, -0.0054,  0.0277, -0.0179, -0.1519,  0.0144,
        -0.0381, -0.0068, -0.0799, -0.0166,  0.0870,  0.0113,  0.0719,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0490,  0.0877, -0.0169, -0.0220, -0.0197, -0.0517, -0.0169,  0.0180,
         0.0620,  0.0520,  0.0071,  0.0736,  0.0922, -0.0001, -0.0048, -0.0745,
         0.0346, -0.0523,  0.0068, -0.0153,  0.0374,  0.0874, -0.0016,  0.0112,
         0.0037,  0.1014,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0555, -0.0538, -0.0773, -0.1286, -0.0667, -0.0295,  0.0706, -0.0997,
        -0.3008, -0.0257, -0.0403,  0.0179, -0.0171, -0.0157,  0.0008,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.4858e-02,  1.7608e-01,  1.1815e-02, -1.0811e-02,  8.8134e-02,
         4.6176e-02,  3.8163e-02, -1.0389e-02,  1.1088e-02,  2.2196e-01,
        -2.7462e-05,  5.0115e-03,  2.4203e-02,  1.9220e-02,  3.3318e-02,
         9.5951e-03,  1.6763e-02, -3.8774e-03,  2.3397e-02, -8.7512e-02,
         2.3943e-03, -2.3900e-02,  1.0459e-02, -8.3241e-03, -3.3127e-02,
        -3.9397e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0018, -0.0225,  0.0121, -0.0471, -0.0351, -0.0386, -0.0122, -0.0398,
        -0.0093, -0.0391, -0.1884, -0.0231, -0.0262, -0.0129, -0.0765, -0.0669,
         0.0219, -0.0610,  0.0070, -0.0349,  0.0616, -0.0042,  0.0238, -0.0054,
         0.0839, -0.0448,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 3.4289e-02, -2.6652e-01,  5.8022e-02,  7.1128e-05, -4.7545e-03,
        -1.0624e-02, -1.5454e-02, -3.1373e-02, -2.3848e-02, -2.9498e-02,
        -1.8659e-02, -2.4089e-02, -7.5332e-02, -2.6362e-02, -5.0711e-03,
         2.4357e-02, -2.7224e-03, -1.2729e-02, -2.9275e-03, -7.9358e-03,
        -4.4514e-03, -2.9397e-02,  1.7670e-02, -2.1404e-02, -5.0665e-02,
        -1.3118e-02, -3.5609e-02,  6.7067e-03, -3.6056e-03, -3.7836e-03,
         2.2153e-02, -3.1113e-02,  9.8818e-03,  8.3850e-03, -2.0751e-02,
        -4.6667e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1062, -0.0803, -0.0333, -0.0250, -0.0394, -0.0604, -0.0350, -0.0892,
         0.0056,  0.0050, -0.0191, -0.0157,  0.0109, -0.0030, -0.0101, -0.0253,
         0.0023, -0.0092,  0.0143,  0.0114, -0.0100, -0.0539, -0.0322, -0.0170,
        -0.0244, -0.0892, -0.0144, -0.0120, -0.0463, -0.0443, -0.0105, -0.0142,
        -0.0023, -0.0015, -0.0189, -0.0086,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0046,  0.4095, -0.0044, -0.0097,  0.0347,  0.0619,  0.0362,  0.0625,
         0.0049, -0.0219,  0.0616, -0.0119,  0.0536,  0.0169,  0.0301, -0.0054,
         0.0532,  0.0097,  0.0842,  0.0233,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1005, -0.0410, -0.0175, -0.0179, -0.0074, -0.0249, -0.0597, -0.0464,
        -0.0047, -0.0167, -0.0248, -0.0018,  0.0107, -0.0228, -0.0610, -0.0191,
         0.0298, -0.0411, -0.0076, -0.0236, -0.0034, -0.0097, -0.0006, -0.0450,
        -0.0250, -0.0180,  0.0036,  0.0003, -0.0285, -0.0089, -0.0241,  0.0456,
         0.0324, -0.0578,  0.0297,  0.0435,  0.0082,  0.0365,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1690e-01, -5.4941e-02, -8.1052e-03, -6.7908e-02, -5.4379e-03,
         3.2944e-05,  1.8000e-02,  3.7192e-03, -2.4505e-02, -1.3313e-02,
        -3.0413e-02, -3.2630e-02, -8.5941e-03,  2.3822e-02,  3.1461e-02,
        -5.6684e-02,  2.9167e-02, -3.4267e-02, -1.5290e-02, -7.8974e-03,
        -4.1798e-03,  2.2059e-02, -9.6407e-03,  2.0892e-02, -5.3468e-02,
         2.8550e-03, -2.4342e-02, -2.9818e-02, -3.8376e-02,  2.3892e-03,
        -1.0686e-02,  3.6445e-02,  9.9470e-03, -8.4960e-03, -7.4979e-03,
         1.6999e-02,  1.3423e-02,  2.1572e-02, -3.9776e-03,  5.8140e-02,
        -2.1705e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0802, -0.3221,  0.1141, -0.0762, -0.0543, -0.0203, -0.0624, -0.0214,
        -0.0148, -0.0098, -0.0127, -0.0200, -0.0265,  0.0096, -0.0170, -0.0252,
        -0.0035, -0.0323, -0.0106, -0.0140,  0.0357,  0.0174,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0420, -0.0160, -0.0009,  0.0144,  0.0292, -0.0416, -0.0026,  0.0220,
        -0.0461, -0.0066, -0.0424, -0.0564, -0.0034, -0.0229, -0.0105, -0.0336,
         0.0022, -0.0182, -0.0203, -0.0350, -0.1887,  0.0254, -0.0281, -0.0310,
        -0.0097, -0.0148, -0.0056, -0.0349, -0.0081,  0.0137,  0.0050, -0.0291,
        -0.0376,  0.0585, -0.0432,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0382, -0.0190,  0.0616,  0.0246,  0.0624,  0.0097,  0.0062,  0.0231,
         0.0173,  0.0224,  0.0531, -0.0005,  0.0806,  0.0133,  0.0047,  0.0430,
         0.0070,  0.0176,  0.0552, -0.0051,  0.0167,  0.0149,  0.0601,  0.0469,
         0.0292, -0.0056,  0.0469,  0.1037, -0.1114,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.5191e-02,  2.0943e-01,  6.0997e-02,  6.2026e-02,  2.6811e-02,
         7.0752e-02, -1.6786e-04,  1.2527e-02,  1.6799e-02, -6.2834e-03,
         3.1320e-02,  2.3657e-02, -3.7573e-02, -4.6199e-03,  9.0861e-03,
         2.1228e-02,  4.5688e-02,  1.8488e-03,  5.4793e-02,  1.4610e-02,
         4.2363e-02,  6.1384e-02,  8.9598e-02, -5.0089e-02, -1.1154e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1317, -0.0452, -0.0074, -0.0103, -0.1321, -0.0363, -0.0429, -0.0911,
        -0.0531, -0.0005, -0.0304,  0.0340, -0.0185,  0.0320, -0.0402, -0.0724,
        -0.0579, -0.0112,  0.0064, -0.0260, -0.0035, -0.0508,  0.0663,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0362, -0.0340,  0.0129,  0.0073, -0.0257,  0.0170,  0.0252, -0.0263,
        -0.0104,  0.0037,  0.0008,  0.0022, -0.0522, -0.1109,  0.0367, -0.0255,
        -0.0158, -0.0301, -0.0175,  0.0008, -0.0497, -0.0124, -0.0182, -0.0211,
        -0.0103, -0.0087, -0.0817, -0.0349, -0.0411, -0.0479, -0.0248, -0.0084,
        -0.0239,  0.0022, -0.0051, -0.0035,  0.0068,  0.0516,  0.0567,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0102,  0.0599, -0.0364, -0.0882, -0.2899, -0.0460,  0.0226, -0.0780,
        -0.1711,  0.0337, -0.0670, -0.0970,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.1028, -0.0893, -0.0908,  0.0035, -0.0195, -0.0324, -0.0305, -0.0183,
         0.0030, -0.0190, -0.0218, -0.0003,  0.0007, -0.0061, -0.0265, -0.0080,
        -0.0011, -0.0689, -0.0058, -0.0196, -0.0171, -0.0167,  0.0054, -0.0149,
         0.0112,  0.0005, -0.0151,  0.0115, -0.0257,  0.0074, -0.0133,  0.0043,
        -0.0702, -0.0029, -0.0380, -0.0285, -0.0267, -0.0044,  0.0093,  0.0013,
         0.0236,  0.0147, -0.0177,  0.0167, -0.0349,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0811, -0.4234, -0.0149, -0.0812, -0.0435, -0.0093,  0.0523, -0.0904,
        -0.1435,  0.0202, -0.0301, -0.0100,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1962,  0.4642,  0.0860,  0.1053, -0.0246, -0.0440, -0.0245,  0.0553,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-9.9435e-03, -1.7025e-01, -1.6519e-02,  1.0971e-02,  1.5587e-02,
         2.0155e-02,  1.5871e-02,  3.0529e-03, -2.3432e-02, -6.8944e-02,
        -3.4499e-02,  1.1535e-02, -2.5784e-03,  1.2853e-04, -4.7597e-03,
         1.5322e-02,  1.1909e-02, -3.9060e-02,  5.2320e-04, -5.5131e-03,
        -1.0314e-02, -3.6635e-04, -1.2558e-03,  1.3885e-03,  1.1103e-03,
        -2.2378e-02, -4.4849e-03, -1.0659e-02, -2.1322e-02,  8.5207e-03,
        -1.7369e-02,  9.4779e-03, -7.1452e-03, -1.2260e-02,  6.7729e-03,
        -1.3634e-02,  2.6132e-03, -3.4093e-03, -1.0566e-02, -4.5890e-03,
        -5.6670e-04, -4.3806e-03, -6.4064e-03, -1.1544e-02,  6.8592e-03,
         7.7200e-02, -4.2139e-02, -1.8141e-02, -2.1076e-02, -4.1670e-02,
        -1.5318e-02, -6.9054e-03, -1.9431e-02,  4.2950e-03, -4.8635e-03,
         9.3815e-03, -1.3506e-02,  3.1383e-03,  1.3740e-04,  3.3047e-02,
         9.8067e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.8383e-02, -3.3882e-02,  1.5835e-02, -1.5283e-02, -1.1751e-01,
        -2.9205e-02, -4.9638e-02, -1.5824e-05, -1.6678e-02,  1.4690e-02,
        -2.7917e-02, -2.7263e-02, -7.0510e-03, -2.0557e-02, -8.8557e-02,
        -9.5126e-03, -1.2644e-02,  4.0808e-02, -6.4446e-02, -5.8135e-03,
        -7.9265e-02, -3.3515e-02, -1.0137e-01, -4.0261e-03, -3.4324e-02,
        -2.6981e-02, -4.4172e-03, -4.1790e-03, -9.2775e-03, -1.7540e-02,
        -8.3083e-03,  3.0028e-02, -1.1077e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0007, -0.0996, -0.0962, -0.0509, -0.0724, -0.0276, -0.0578, -0.0120,
         0.0015,  0.0219,  0.0244, -0.0128, -0.0359, -0.0258, -0.0337,  0.0152,
        -0.0123, -0.0733, -0.2256,  0.0073, -0.0084, -0.0294,  0.0198,  0.0352,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0542, -0.0538, -0.0956, -0.0498, -0.0166, -0.1186, -0.0248, -0.0297,
         0.0150, -0.0864, -0.0563, -0.0284, -0.1129,  0.0004, -0.0315, -0.0134,
        -0.0095,  0.0044,  0.0325,  0.0294, -0.0209,  0.0025,  0.0003, -0.1134,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1029, -0.0932, -0.0734, -0.0548,  0.0037, -0.0064, -0.0229, -0.0170,
        -0.0011, -0.0079, -0.0264,  0.0088,  0.0058,  0.0052, -0.0291, -0.0674,
        -0.0563, -0.0254, -0.0188, -0.0322, -0.0308, -0.0134, -0.0422, -0.0754,
        -0.0149, -0.0178, -0.0286, -0.0120, -0.0150,  0.0080, -0.0122, -0.0286,
         0.0319,  0.0108,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0573,  0.0033, -0.0316,  0.0062, -0.0943, -0.0231, -0.0078, -0.0059,
         0.0386, -0.0108,  0.0104, -0.0101, -0.0396, -0.1972, -0.0394, -0.0483,
        -0.0904, -0.0313, -0.0445, -0.0241, -0.0304, -0.0623, -0.0041, -0.0267,
        -0.0622,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0948,  0.0232, -0.0750, -0.0318, -0.0500, -0.0221, -0.0874, -0.0595,
        -0.0738, -0.1818,  0.0457, -0.0513, -0.0353, -0.0677, -0.0596,  0.0409,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0626, -0.1324, -0.0314, -0.0645, -0.0323, -0.0598, -0.0990, -0.0225,
        -0.0061, -0.0315, -0.0268,  0.0717, -0.0661, -0.0393, -0.0463,  0.0066,
        -0.0162, -0.0215, -0.0820, -0.0448,  0.0199,  0.0107,  0.0059,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0003,  0.1559,  0.0481,  0.1243,  0.2094,  0.1612, -0.0144, -0.0038,
         0.0087, -0.0782, -0.0567,  0.0507, -0.0803, -0.0079,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.1415, -0.0279, -0.0377, -0.1289, -0.0142,  0.0067, -0.0113, -0.0116,
        -0.0147, -0.0166, -0.0595, -0.0680, -0.0279, -0.0099, -0.0070, -0.0103,
        -0.0236, -0.1447, -0.0349, -0.0416, -0.0578, -0.0355, -0.0357, -0.0326,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0377,  0.0111, -0.2232, -0.0574, -0.0981, -0.1147, -0.1815, -0.0760,
         0.0167, -0.1141, -0.0693,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0197, -0.0093, -0.0024, -0.0792, -0.0024, -0.0415, -0.0944, -0.0537,
        -0.0965, -0.0008, -0.0186,  0.0315, -0.0335, -0.0105, -0.0615, -0.0140,
        -0.0562, -0.0238, -0.0985, -0.1802, -0.0717,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.4068,  0.0797, -0.0491,  0.0676,  0.0101,  0.0186, -0.0046,  0.1328,
         0.0516,  0.1048, -0.0242,  0.0173, -0.0145,  0.0184,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1038, -0.0546, -0.0052, -0.0197,  0.0148, -0.0147,  0.0034, -0.0416,
         0.0050, -0.0683,  0.0179, -0.0014,  0.3231, -0.0109,  0.0197, -0.0033,
        -0.0177,  0.0255,  0.0197, -0.1390, -0.0905,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1151,  0.2368,  0.0125, -0.0035, -0.1479,  0.1525, -0.0111,  0.0532,
         0.0358,  0.0871,  0.0613,  0.0546,  0.0287,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0014, -0.0402,  0.0231,  0.0080, -0.0127, -0.0045, -0.0217, -0.0315,
        -0.0078, -0.0065, -0.0100,  0.0309,  0.0058, -0.0006,  0.0085, -0.0026,
         0.0145,  0.0167, -0.0071, -0.0050, -0.0171,  0.0008, -0.0400, -0.0202,
        -0.1956,  0.0074, -0.0752, -0.0344, -0.0284, -0.0244, -0.0210, -0.0241,
         0.0279, -0.0123,  0.0625, -0.0121, -0.0403, -0.0074,  0.0014, -0.0034,
        -0.0215,  0.0132,  0.0169, -0.0152,  0.0180], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0545, -0.0342, -0.0471, -0.0177, -0.0206,  0.0047, -0.0281, -0.0233,
        -0.0040, -0.0198, -0.0545,  0.0193, -0.0110, -0.0071, -0.0214,  0.0086,
        -0.0508, -0.0160, -0.0110, -0.0118, -0.0432, -0.1071, -0.0142, -0.0673,
         0.0349,  0.0165, -0.0005,  0.0016, -0.0758, -0.0016, -0.0004,  0.0310,
        -0.0800,  0.0604,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0773, -0.1326, -0.1093, -0.0148, -0.0045, -0.0266,  0.0281, -0.0176,
        -0.0794, -0.0249, -0.0179, -0.0247, -0.0162,  0.0067, -0.0091,  0.0224,
         0.0225, -0.0097,  0.0075, -0.0966, -0.1501,  0.0158, -0.0150,  0.0090,
        -0.0294,  0.0321,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0485, -0.2497,  0.0119, -0.0428,  0.0403, -0.0560,  0.0658, -0.0503,
        -0.0012, -0.0075, -0.0108,  0.0004,  0.0120, -0.0057, -0.0306, -0.0359,
        -0.1152,  0.0635, -0.0518,  0.0282,  0.0181,  0.0393, -0.0143,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0748, -0.0653, -0.0206, -0.0390, -0.0016, -0.0174,  0.0034, -0.0187,
        -0.0314, -0.0077,  0.0024,  0.0271,  0.0048,  0.0006, -0.1824,  0.0385,
         0.0198, -0.0282, -0.0299, -0.0491, -0.0113,  0.0071,  0.0124, -0.0295,
         0.0034, -0.0214, -0.0391, -0.0614,  0.0094,  0.0059, -0.0057, -0.0008,
         0.0077,  0.0233,  0.0146,  0.0122, -0.0191,  0.0037, -0.0144, -0.0349,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0192, -0.0433, -0.0281, -0.0253, -0.0898, -0.0094, -0.0172, -0.0755,
        -0.0282, -0.0208, -0.0288, -0.0233, -0.0262, -0.0065, -0.0259, -0.0326,
        -0.0310, -0.0043,  0.0433, -0.0218, -0.0082, -0.0387, -0.0173,  0.0166,
        -0.0586,  0.0033,  0.0071,  0.0228,  0.0175, -0.0080, -0.1002, -0.0084,
        -0.0483, -0.0448,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0479, -0.0687, -0.0090,  0.0032,  0.0127, -0.0047, -0.0067, -0.0147,
         0.0034,  0.0175, -0.0158, -0.0051,  0.0074, -0.0352, -0.1401, -0.0005,
         0.0084,  0.0070, -0.0410,  0.0071,  0.0143,  0.0061, -0.0023,  0.0081,
        -0.0474, -0.0891,  0.0023, -0.0358,  0.0233,  0.0068, -0.0225, -0.0029,
         0.0059,  0.0713, -0.0224,  0.0015, -0.0251, -0.0162, -0.0099, -0.0214,
        -0.0185, -0.0056, -0.0159, -0.0054, -0.0051,  0.0102,  0.0002,  0.0046,
        -0.0126,  0.0005, -0.0096,  0.0213], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0245,  0.0018,  0.0213, -0.0187, -0.1006, -0.0343, -0.0065,  0.0202,
        -0.0091, -0.0403,  0.0017, -0.0230, -0.0344, -0.0989, -0.0528, -0.0645,
        -0.0923,  0.0081, -0.0310, -0.0205, -0.0384, -0.0420, -0.0331, -0.0645,
        -0.0048,  0.0360,  0.0170,  0.0222, -0.0204,  0.0045,  0.0125,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.6441e-01, -3.7933e-02, -4.6847e-03, -1.1427e-02, -1.3427e-02,
        -9.9074e-05, -2.3895e-03,  1.1551e-02, -3.5059e-02, -1.1269e-02,
        -1.7509e-02, -4.3808e-02, -5.7870e-02, -3.3295e-02, -3.2416e-02,
        -4.4738e-02,  8.0231e-02, -2.7552e-02, -6.1429e-02, -6.9038e-02,
        -5.1242e-02,  2.7402e-02, -9.3628e-02, -3.5177e-02, -3.2422e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0113, -0.0152,  0.0034,  0.0506, -0.0019, -0.0605, -0.5674, -0.0078,
        -0.0612, -0.0815,  0.1346, -0.0045,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1368, -0.2481, -0.0014, -0.0189, -0.0384, -0.0078,  0.0073, -0.0109,
         0.0130, -0.0116,  0.0234,  0.0179, -0.0398, -0.0835, -0.0143, -0.0253,
        -0.0043,  0.0260,  0.0046,  0.0010, -0.0058,  0.0003, -0.0263,  0.0191,
         0.0035, -0.0617, -0.0060,  0.0533, -0.0898,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0290,  0.2393, -0.1508, -0.0067, -0.0370,  0.0612,  0.1939, -0.0788,
        -0.1020, -0.0015, -0.0361,  0.0636,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0631, -0.2610,  0.0204, -0.0475,  0.1155, -0.1355, -0.1967,  0.0204,
        -0.0641, -0.0346,  0.0194, -0.0219,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0398, -0.0855,  0.0645,  0.0359,  0.0667, -0.0634,  0.0127,  0.0508,
         0.0770,  0.0173,  0.0280,  0.0873,  0.0442, -0.0197,  0.0194,  0.0721,
         0.0467,  0.0068,  0.0621, -0.1002,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0888, -0.2933, -0.0174, -0.0026, -0.1078, -0.0130,  0.0201, -0.0012,
        -0.0174, -0.0143,  0.0025, -0.0081,  0.0026, -0.0047, -0.0196,  0.0037,
        -0.0169,  0.0064,  0.0039, -0.0350, -0.0149,  0.0108,  0.0135,  0.0489,
         0.0040, -0.0045,  0.0074, -0.0014, -0.0046,  0.0111, -0.0117, -0.0029,
        -0.0256, -0.0199, -0.0082,  0.0438, -0.0875,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1519,  0.3842,  0.1513, -0.0454,  0.0232,  0.0311, -0.0377, -0.0320,
         0.1432,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0423, -0.1133, -0.0556, -0.1996, -0.0482, -0.0380, -0.0829, -0.0803,
        -0.0117,  0.0031,  0.0057,  0.0107, -0.0368, -0.0714, -0.0005,  0.0188,
         0.0075,  0.0534, -0.0193, -0.0532,  0.0163,  0.0022,  0.0188, -0.0047,
         0.0012,  0.0014, -0.0031,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.3818e-01, -2.3938e-01, -1.4389e-03, -1.5747e-02, -1.9758e-02,
        -1.0631e-02, -2.1321e-04,  3.6965e-03,  5.8104e-04, -5.5993e-03,
         7.4692e-03, -4.4399e-02, -1.9164e-02, -7.1373e-03, -2.9437e-03,
         7.5703e-03, -2.6521e-02,  2.2230e-02,  6.4522e-04, -1.6310e-02,
         1.1223e-01, -3.4679e-02, -2.7567e-02, -4.5085e-03, -6.2856e-03,
         9.5046e-03,  2.2548e-03, -1.5698e-02, -1.3593e-01, -1.0534e-02,
        -5.6839e-03, -5.9321e-03,  3.0492e-02, -9.0795e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.1675, -0.0087,  0.0036,  0.0094, -0.0198, -0.0120,  0.0103,  0.0059,
        -0.0077,  0.0057, -0.0132, -0.1054, -0.0238, -0.0088, -0.0095, -0.0161,
        -0.0488,  0.0203, -0.0131, -0.0023,  0.0192,  0.0055, -0.0352,  0.0038,
        -0.0041, -0.0079, -0.0129,  0.0026,  0.0083, -0.0010,  0.0129, -0.0230,
        -0.1186, -0.0117, -0.0051, -0.0215, -0.0721,  0.0049, -0.0031,  0.0103,
        -0.0200, -0.0068,  0.0067,  0.0352,  0.0355,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.6393e-02,  2.8324e-02,  2.6147e-05, -5.6062e-03, -5.6906e-02,
        -8.2656e-02, -2.7185e-01, -2.2068e-02, -2.1305e-02, -1.0276e-01,
        -5.1176e-02, -1.2400e-01, -6.9600e-02, -1.0026e-01, -2.0305e-02,
        -6.7649e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.6851e-02,  4.5992e-02, -5.1868e-03,  1.1600e-02,  1.1694e-02,
        -4.1786e-02, -2.1784e-02, -9.7992e-03, -6.8630e-02, -7.5916e-02,
        -5.5210e-02, -8.0041e-02, -6.5530e-02, -1.5888e-02, -3.9475e-02,
         1.2032e-02, -2.2333e-02, -2.7707e-02,  3.0193e-03,  2.4177e-02,
         6.8919e-05,  4.1010e-02, -2.2115e-02, -2.2987e-02,  2.7386e-02,
         7.1183e-02, -9.0598e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.7089e-01, -1.5203e-02, -4.4180e-02,  1.1037e-03, -1.4720e-02,
        -9.0425e-03, -1.4801e-02, -9.1329e-03,  1.2502e-03, -4.5066e-03,
        -1.0256e-02, -1.7255e-02, -1.8204e-02, -5.3780e-03, -1.9049e-03,
        -2.4849e-03, -1.0314e-02, -2.9800e-03,  2.6278e-03, -3.2839e-02,
        -3.2098e-03, -1.7542e-02, -2.4349e-02, -1.1924e-01,  3.0426e-03,
        -1.6638e-03, -9.8865e-03, -2.7620e-02,  3.5961e-03, -2.3878e-02,
         1.3003e-02, -1.9976e-02, -2.7969e-02, -2.6535e-02, -7.4961e-03,
        -2.4011e-02, -1.1960e-02, -3.7100e-02,  1.3710e-04,  2.1023e-02,
        -1.3687e-02, -3.9749e-02, -1.1638e-02,  4.7988e-03,  2.0447e-02,
         9.7377e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0860,  0.0344, -0.0178, -0.0366,  0.0214, -0.0393, -0.0018, -0.0172,
        -0.0447, -0.0176,  0.0099,  0.0062,  0.0130,  0.0065, -0.0351,  0.0058,
        -0.0174, -0.0162, -0.0711, -0.1192, -0.0244,  0.0143, -0.0004,  0.0067,
        -0.0051, -0.0473, -0.0672,  0.0104,  0.1126,  0.0947,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([0.1497, 0.0803, 0.1117, 0.0440, 0.0605, 0.1914, 0.0571, 0.1194, 0.1072,
        0.0572, 0.0215, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0861, -0.2784, -0.0376, -0.0725, -0.0280, -0.0415, -0.0287, -0.1680,
        -0.0015, -0.0328, -0.0031,  0.0293, -0.0274, -0.0271, -0.0232, -0.0139,
        -0.0140, -0.0180,  0.0220, -0.0471,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1110, -0.2411, -0.0833, -0.0771,  0.0883, -0.0665, -0.1518,  0.0021,
        -0.0623,  0.0442, -0.0723,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.4529e-02, -2.8570e-01, -2.7430e-02, -8.5042e-02, -9.5676e-02,
         4.2973e-02, -1.2328e-02, -3.1133e-02, -8.5885e-02,  1.8013e-02,
         1.4563e-02, -1.4358e-02, -3.7618e-02,  7.2117e-05, -3.9600e-02,
        -4.1228e-02,  2.2886e-02, -1.2519e-02, -4.7485e-02,  1.4873e-02,
        -2.6086e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0462, -0.0982, -0.0115, -0.0023,  0.0431, -0.3956,  0.0476,  0.0004,
        -0.0246, -0.0455, -0.0030, -0.0356, -0.0274,  0.0284, -0.0370, -0.0395,
         0.0794,  0.0346,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0749, -0.0763, -0.1155, -0.0111, -0.0327, -0.0241, -0.0006,  0.0126,
        -0.0081, -0.0232, -0.0037,  0.0083, -0.0053, -0.0168, -0.0213,  0.0167,
        -0.0431, -0.0616, -0.0284,  0.0235, -0.0410, -0.0313, -0.0928, -0.0063,
        -0.0270, -0.0156, -0.0276, -0.0136, -0.0069, -0.0567,  0.0044, -0.0038,
        -0.0040,  0.0067, -0.0115, -0.0054,  0.0161,  0.0116, -0.0101,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0622,  0.2179,  0.0154,  0.0128,  0.0132, -0.0042, -0.0177, -0.0614,
        -0.0240, -0.0065, -0.0409,  0.0310, -0.0110,  0.0200, -0.0227, -0.0117,
         0.0621, -0.0003,  0.0087, -0.0020,  0.0121,  0.0109,  0.0039, -0.0017,
         0.0173,  0.0200,  0.0406,  0.0118,  0.0512,  0.0837,  0.0080,  0.0163,
         0.0609,  0.0161,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-3.0249e-01,  2.6401e-01,  3.7232e-02,  3.6766e-02,  6.7171e-02,
         1.2422e-01,  1.4380e-04,  8.1685e-02,  8.6289e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0050, -0.1786, -0.0233, -0.0432, -0.0144, -0.0185, -0.0046,  0.0119,
        -0.0059, -0.0176, -0.0106, -0.0165,  0.0036, -0.0147, -0.0055, -0.0301,
        -0.0108, -0.1618, -0.0168, -0.0401, -0.0118,  0.0064,  0.0168, -0.0171,
        -0.0480, -0.0157,  0.0131,  0.0042,  0.0320,  0.0042, -0.0317, -0.0243,
         0.0041, -0.0241,  0.0141,  0.0063, -0.0058,  0.0149, -0.0602, -0.0118,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0478, -0.0251, -0.1003, -0.0217, -0.0206, -0.0169, -0.0199, -0.0195,
        -0.0230,  0.0209, -0.0066,  0.0033, -0.0069, -0.0006,  0.0086, -0.0038,
        -0.0165, -0.0247, -0.0225,  0.0032,  0.0035,  0.0028, -0.0028, -0.0152,
        -0.0017, -0.0200, -0.0555, -0.0039, -0.0358, -0.0801, -0.0239, -0.0241,
        -0.0017, -0.0183, -0.0168, -0.0315, -0.0373, -0.0599, -0.0144, -0.0051,
         0.0251,  0.0107,  0.0164,  0.0037,  0.0017, -0.0029, -0.0139, -0.0070,
         0.0039, -0.0080, -0.0043,  0.0357,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1561, -0.2015, -0.0613,  0.0174, -0.0206, -0.0233, -0.0198, -0.0003,
         0.0011,  0.0107,  0.0214, -0.0290, -0.0411,  0.0165, -0.0356, -0.0494,
        -0.0292,  0.0065, -0.0403, -0.0533,  0.0074, -0.1582,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0287, -0.3773, -0.0655, -0.0318, -0.0762,  0.0117, -0.0144, -0.0264,
        -0.0393, -0.0393,  0.0194, -0.0107, -0.0060, -0.0168, -0.0110, -0.0487,
        -0.0396,  0.0260,  0.0245,  0.0564, -0.0303,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0830, -0.0613, -0.0020, -0.0192,  0.0512, -0.0204, -0.0187,  0.0120,
        -0.0140,  0.0004, -0.0129, -0.0140, -0.0138, -0.0067,  0.0041, -0.0033,
        -0.0084, -0.0130, -0.0220,  0.0040,  0.0099, -0.0176, -0.0032, -0.0109,
        -0.0044, -0.0040, -0.0384, -0.0268,  0.0042, -0.0067, -0.0831, -0.0231,
        -0.0064, -0.0182, -0.0215, -0.0185, -0.0034,  0.0076,  0.0012, -0.0045,
        -0.0715, -0.0067, -0.0321,  0.0047, -0.0304, -0.0035, -0.0100, -0.0283,
        -0.0232, -0.0055, -0.0160, -0.0052, -0.0262,  0.0091, -0.0297],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0397, -0.0414,  0.0222, -0.0252, -0.0339, -0.0128, -0.0194, -0.0088,
        -0.0048,  0.0089,  0.0196,  0.0033,  0.0043, -0.0170, -0.0207, -0.0197,
        -0.0960, -0.0370,  0.0030, -0.0050, -0.0111, -0.0295, -0.0297, -0.0096,
         0.0329, -0.0071, -0.0194, -0.0472, -0.0178,  0.0012,  0.0367, -0.0079,
        -0.0061, -0.0712, -0.0045, -0.0225, -0.0083, -0.0177, -0.0128, -0.0285,
        -0.0118, -0.0021,  0.0124,  0.0865,  0.0228,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1892e-01, -1.6071e-01, -7.4963e-02, -8.3933e-02,  1.7172e-02,
        -3.1458e-02, -3.4210e-02, -4.9154e-02, -4.3291e-02, -8.3418e-03,
        -2.1088e-05, -1.3966e-02, -1.9754e-02, -1.9713e-02, -3.5382e-02,
        -1.9760e-02, -9.1470e-03, -5.6283e-02,  2.3991e-02, -1.2654e-02,
        -1.3944e-01,  3.4270e-03, -4.4749e-03,  1.9834e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0083, -0.0264,  0.0145,  0.0158, -0.0103,  0.0006, -0.0044, -0.0386,
        -0.0227,  0.0202, -0.0077, -0.0023, -0.0067, -0.0636, -0.0211, -0.0008,
        -0.0659, -0.0408, -0.0514, -0.0656, -0.0243, -0.0201, -0.0477, -0.0721,
         0.0520, -0.0419, -0.0762, -0.0389, -0.0224, -0.0578,  0.0065,  0.0209,
        -0.0315,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.9132e-02, -2.9609e-01,  4.1223e-03, -2.5313e-02, -1.3217e-02,
        -1.2384e-02,  2.1308e-02, -3.4784e-02, -1.3211e-01, -1.0566e-03,
        -2.2636e-02, -5.0479e-04,  2.0460e-01, -4.6036e-02,  6.2291e-05,
        -1.9316e-02, -3.8686e-02, -3.6394e-04,  2.8270e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0739, -0.0120,  0.0250, -0.0641, -0.0568, -0.0027, -0.0641, -0.0386,
        -0.1260,  0.0007, -0.0080, -0.0396, -0.0005, -0.0224, -0.0099,  0.0172,
         0.0112, -0.0392, -0.0080, -0.0057, -0.0503, -0.1300, -0.0725, -0.0403,
         0.0812,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0518, -0.1483, -0.0293,  0.0053, -0.0875, -0.0659, -0.2900,  0.0138,
        -0.0354, -0.0240,  0.0254,  0.0106, -0.0124, -0.0139, -0.0443,  0.0793,
        -0.0627,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0246,  0.0078, -0.0441,  0.0211, -0.0160, -0.0321, -0.0272, -0.0499,
        -0.0167, -0.0595, -0.0128, -0.0356, -0.0024, -0.0028, -0.0127, -0.0157,
        -0.0286, -0.1303,  0.0069, -0.0150, -0.0035, -0.0813, -0.1393, -0.0091,
        -0.0166, -0.0500, -0.0659, -0.0241,  0.0007,  0.0122, -0.0123,  0.0232,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0935, -0.1388, -0.0432, -0.0300, -0.0106, -0.0219, -0.0279, -0.0812,
        -0.0009, -0.0015, -0.0155, -0.0028, -0.0285, -0.0113, -0.0160,  0.0396,
        -0.0192, -0.0281, -0.0146,  0.0032,  0.0052,  0.0335, -0.0680, -0.0689,
        -0.0292, -0.0116, -0.1400, -0.0155,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0223, -0.1561,  0.0054, -0.0088,  0.0616, -0.0395, -0.1443, -0.0115,
        -0.0330, -0.0150,  0.0176, -0.0951,  0.0311,  0.0005,  0.0172, -0.0116,
        -0.0252, -0.1025,  0.0061, -0.0175, -0.0095, -0.0098,  0.0157, -0.0222,
        -0.0530,  0.0456,  0.0041,  0.0159, -0.0023,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0194, -0.0690, -0.0049, -0.0469, -0.1183, -0.0693, -0.0059,  0.0044,
         0.0030,  0.0379,  0.0183, -0.0319, -0.0175, -0.0838,  0.0108, -0.0221,
        -0.0517, -0.1005, -0.0392,  0.0055,  0.0619, -0.0094, -0.0118, -0.0146,
        -0.0323, -0.0082, -0.0012, -0.0273, -0.0361, -0.0101,  0.0006,  0.0082,
         0.0179,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0327, -0.2955, -0.0182, -0.0718, -0.0265, -0.0682,  0.0020, -0.0516,
        -0.0057, -0.0230, -0.0142, -0.0146, -0.0359, -0.0137, -0.0170, -0.0026,
        -0.0312, -0.0069, -0.0937, -0.0366, -0.0620, -0.0307, -0.0109,  0.0347,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.0855e-01, -1.4111e-01, -7.7644e-02, -6.1899e-03, -1.8126e-02,
        -2.8897e-02, -5.3033e-02, -6.2213e-02, -1.2075e-02, -3.8878e-02,
        -3.7399e-02, -4.3097e-02, -4.1747e-02, -1.9601e-02, -8.0397e-03,
        -1.0575e-02, -7.6571e-05, -2.1617e-02, -1.6803e-02, -1.5741e-01,
        -2.7724e-02, -2.4336e-02, -4.4848e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0066, -0.0958, -0.0007, -0.0085,  0.0067, -0.0309,  0.0166, -0.0214,
        -0.0021, -0.0309,  0.0071, -0.0108, -0.0240,  0.0146, -0.0026,  0.0154,
        -0.0384, -0.0350, -0.0263, -0.0393,  0.0106, -0.0247, -0.0012, -0.0201,
        -0.0502, -0.0317, -0.0101, -0.0152, -0.0186, -0.0704, -0.0532, -0.0522,
        -0.0301, -0.0322, -0.0247, -0.0305, -0.0380,  0.0524, -0.0003,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0349, -0.0703, -0.0455, -0.0225, -0.1010, -0.0114,  0.0099, -0.0068,
        -0.0512,  0.0140, -0.0071, -0.0053, -0.0031, -0.0032,  0.0168, -0.0546,
         0.0113,  0.0151, -0.0150, -0.0109, -0.0259,  0.0063,  0.0008, -0.0280,
        -0.0202, -0.0373, -0.0174, -0.0029,  0.0015, -0.0181, -0.0098, -0.0039,
        -0.0643, -0.0170, -0.0050,  0.0024, -0.0181, -0.0572, -0.0121, -0.0046,
        -0.0101, -0.0655, -0.0619,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0084, -0.1977,  0.0089,  0.0037, -0.0084,  0.0127,  0.0048, -0.0107,
        -0.0707,  0.0179,  0.0233, -0.0004, -0.0223,  0.0085, -0.0094,  0.0047,
        -0.0053, -0.0226,  0.0052, -0.0055,  0.0143, -0.0178, -0.0044,  0.0002,
        -0.0157,  0.0091,  0.0014,  0.0039,  0.0109, -0.0040, -0.0130,  0.0037,
        -0.0569,  0.0080,  0.0009,  0.0373,  0.0292,  0.0032, -0.0320,  0.0043,
        -0.0292, -0.0506, -0.0151,  0.0123,  0.0131, -0.0266, -0.0122, -0.0027,
         0.0062,  0.0037, -0.0181, -0.0099, -0.0139, -0.0031,  0.0087,  0.0062,
        -0.0468], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.9643e-02, -3.8323e-01, -4.1690e-02, -3.8127e-03, -2.5891e-02,
         7.2074e-02, -1.6731e-03, -5.7407e-02, -5.5001e-02,  4.6815e-02,
        -6.4014e-02,  3.1819e-04,  1.2034e-01, -5.8093e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0038, -0.0997, -0.0678,  0.0807,  0.0526, -0.0350, -0.0240, -0.0036,
         0.0134,  0.0085, -0.0173, -0.2233, -0.0143, -0.0108, -0.0137, -0.0185,
        -0.0124, -0.0420,  0.0430, -0.0548, -0.0164, -0.0288, -0.0107, -0.0243,
        -0.0114, -0.0046,  0.0099,  0.0215, -0.0004,  0.0327,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1439,  0.2227,  0.2583,  0.2207,  0.1186, -0.0043,  0.0017,  0.0298,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0486, -0.0845,  0.0136, -0.0203, -0.0185, -0.0864,  0.0152,  0.0244,
        -0.0023, -0.0133,  0.0034, -0.0397, -0.0418, -0.0578, -0.0169, -0.0163,
        -0.0153, -0.0206, -0.0256, -0.0182, -0.0089, -0.0141, -0.0211,  0.0076,
        -0.0419, -0.0205, -0.0179,  0.0513, -0.0413, -0.0833,  0.0150,  0.0593,
         0.0352,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0822, -0.0803, -0.0415, -0.0034, -0.0118, -0.0154, -0.0040, -0.0005,
        -0.0038, -0.0215, -0.0105, -0.0042, -0.0136, -0.0085,  0.0091,  0.0136,
         0.0075,  0.0104, -0.0125,  0.0049, -0.0104, -0.0003, -0.0054, -0.0128,
        -0.0053, -0.0120, -0.0239, -0.0669, -0.0046, -0.0082, -0.0124, -0.0119,
        -0.0120, -0.0682, -0.0207,  0.0211, -0.0048, -0.0056, -0.0005, -0.0099,
         0.0050, -0.0003, -0.0046, -0.0223, -0.0085,  0.0366, -0.0008, -0.0004,
        -0.0002,  0.0060, -0.0064, -0.0106, -0.0026, -0.0077, -0.0061,  0.0011,
         0.0042, -0.0018, -0.0154, -0.0248, -0.0162, -0.0060, -0.0571, -0.0158,
         0.0032, -0.0055, -0.0002,  0.0217, -0.0037,  0.0209,  0.0082],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.3995e-02, -2.3098e-02, -1.4104e-03, -1.5809e-02, -4.3704e-02,
        -1.3255e-02, -1.9263e-01, -2.4504e-02, -1.7635e-02, -1.6131e-02,
        -1.1547e-02, -1.2815e-02, -1.3207e-02,  7.9716e-03, -1.3810e-02,
        -1.6971e-02, -9.3638e-03,  1.8367e-02,  2.5036e-02, -6.3771e-02,
         1.3092e-02,  3.4541e-03,  6.2922e-03, -3.7791e-02,  1.5781e-02,
        -1.1143e-02, -1.9825e-02, -7.9066e-03,  7.3715e-03, -3.6150e-03,
         2.6961e-02,  4.3801e-03, -4.8836e-02, -3.2685e-02,  8.1957e-03,
        -1.8649e-02, -1.8059e-02,  1.8266e-02,  5.5892e-05, -8.5794e-03,
         5.9014e-02, -3.5014e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.4264e-02, -1.2633e-01,  6.4834e-03,  9.0722e-03, -5.0545e-03,
        -4.0020e-03,  2.3761e-03,  1.1217e-02,  2.1015e-02, -3.4451e-02,
        -3.0696e-02, -1.4203e-02, -6.5522e-02, -1.5151e-02,  4.4728e-02,
        -2.1825e-02, -1.3748e-03, -3.5719e-02, -9.2841e-02, -1.1521e-02,
        -3.0329e-03, -7.2459e-03, -7.4450e-03, -1.2352e-02,  1.4944e-03,
        -3.5872e-02, -3.7204e-02, -2.6203e-02, -6.5563e-02, -2.0314e-02,
        -7.0542e-03, -1.8180e-02,  3.4537e-06, -3.4505e-02, -2.5598e-02,
         3.1341e-02, -2.8746e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1666, -0.3503,  0.0935, -0.1012, -0.0120,  0.0326, -0.0205, -0.0463,
        -0.0093, -0.0291,  0.0151, -0.0247, -0.0341, -0.0153,  0.0493,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1067,  0.0298,  0.0156,  0.0225, -0.0024,  0.0275,  0.0336,  0.0480,
         0.1558,  0.0274,  0.0315,  0.0236,  0.1507,  0.0548, -0.0575, -0.0087,
         0.1098,  0.0173,  0.0459,  0.0312,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0204, -0.0008, -0.0212, -0.0158, -0.1351, -0.0428, -0.0197, -0.0174,
        -0.0106, -0.0271, -0.0026, -0.0198, -0.0311, -0.0089, -0.0120, -0.0160,
         0.0109, -0.0195,  0.0164, -0.0324,  0.0008, -0.0088,  0.0039,  0.0016,
        -0.0045, -0.0193, -0.0692, -0.0058, -0.0009, -0.0297, -0.0322, -0.0125,
        -0.0025, -0.0161, -0.0218, -0.0551, -0.0198, -0.0158, -0.0028, -0.0244,
        -0.0090,  0.0917, -0.0713,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.5379e-02, -4.7942e-03,  3.9897e-03, -2.1796e-02, -3.2193e-02,
        -2.9778e-02, -2.6059e-02, -1.1314e-01, -2.6195e-02, -1.7083e-02,
        -5.1142e-02,  7.0939e-03, -3.9334e-02, -2.9660e-02, -6.7479e-02,
        -5.5509e-03,  1.6434e-03, -8.1012e-03,  1.2754e-02, -2.0606e-02,
        -3.7266e-02, -2.3447e-02,  2.0434e-02,  2.0536e-02,  6.5706e-03,
         2.2801e-03,  1.8564e-02,  2.0339e-02, -8.4087e-02,  9.6375e-05,
        -1.4319e-02, -2.3053e-02, -2.0514e-02, -1.5641e-02, -3.7426e-02,
        -1.4615e-02,  3.8161e-02,  3.8882e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0773, -0.1872, -0.0838, -0.1104, -0.0063, -0.0773, -0.0227,  0.0237,
        -0.0045, -0.1439, -0.0100, -0.0247, -0.0201, -0.0107, -0.0190, -0.0225,
        -0.0409, -0.0120,  0.0156, -0.0143,  0.0411,  0.0321,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0405, -0.0265,  0.0363, -0.0097,  0.0287,  0.0121, -0.0252,  0.0192,
         0.0126, -0.0721, -0.0132, -0.0079, -0.0034, -0.0112, -0.0104, -0.0312,
        -0.0883, -0.0203, -0.0145, -0.0729,  0.0216,  0.0133, -0.0269, -0.0248,
        -0.0301, -0.0054, -0.0167, -0.0125,  0.0060, -0.0507,  0.0055, -0.0002,
         0.0202,  0.0213, -0.0145, -0.0726,  0.0239,  0.0096, -0.0233, -0.0447,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0649, -0.0713,  0.0382, -0.0605,  0.0183,  0.0210,  0.0259,  0.0059,
        -0.0490, -0.0145,  0.0037, -0.0002,  0.0210, -0.0521, -0.0411, -0.0257,
        -0.0163,  0.0040, -0.0515, -0.0050, -0.0198,  0.0050,  0.0236,  0.0194,
        -0.0248, -0.1045,  0.0250, -0.0696, -0.0547,  0.0171, -0.0024, -0.0440,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0375, -0.3110, -0.0192, -0.0069, -0.0542,  0.0118, -0.0073, -0.0698,
        -0.0659, -0.0317,  0.0045, -0.0381,  0.0014,  0.0173, -0.0262,  0.0130,
        -0.0338,  0.0240, -0.0035, -0.0032, -0.0152,  0.0070, -0.0116, -0.0229,
        -0.0029,  0.0388, -0.0161, -0.0132,  0.0126, -0.0194,  0.0083,  0.0110,
         0.0025, -0.0302,  0.0079,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0530, -0.1028, -0.0931, -0.0872,  0.0233, -0.0430, -0.1325, -0.0903,
        -0.0199, -0.0742, -0.0200, -0.0144, -0.0015, -0.0526, -0.0332, -0.0335,
        -0.0325, -0.0132, -0.0111, -0.0429, -0.0259,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0039, -0.2987,  0.0710, -0.0400, -0.0153, -0.0316,  0.0059, -0.0634,
        -0.1446,  0.0332,  0.0084, -0.0041, -0.0251,  0.0218, -0.0440,  0.0257,
        -0.0498,  0.1024, -0.0110,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1276, -0.0831, -0.0434,  0.0041, -0.0081, -0.0368,  0.0301,  0.0357,
        -0.0233, -0.0356, -0.1279, -0.0014, -0.0558, -0.1553,  0.0334, -0.0446,
         0.0129, -0.0587,  0.0293, -0.0527,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.6497e-05, -3.2982e-01, -1.8656e-02, -9.0635e-02,  3.4230e-02,
        -2.0052e-02, -7.4798e-02, -3.4919e-02, -1.6867e-02, -1.5740e-02,
        -2.0199e-02, -1.0313e-02, -2.3352e-02, -2.7836e-02, -1.2758e-02,
        -3.2507e-02, -1.0048e-01, -1.4717e-02, -1.8450e-02, -4.0501e-02,
        -6.4585e-03, -3.5033e-02, -2.1622e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0397, -0.2093, -0.0448, -0.0621,  0.0100, -0.0069, -0.0260, -0.0146,
        -0.0227, -0.1691, -0.0272, -0.0466, -0.0295, -0.0855,  0.0290, -0.0087,
        -0.0035,  0.0188, -0.0073, -0.0266, -0.0416, -0.0053, -0.0046,  0.0122,
         0.0485,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1031,  0.2574,  0.0372, -0.0132, -0.0117,  0.0017,  0.0058,  0.0256,
         0.0282,  0.0145,  0.0571, -0.0011,  0.0041,  0.0587, -0.0347,  0.0503,
         0.0170,  0.0108, -0.0016,  0.0040,  0.0395,  0.0224,  0.0313,  0.0309,
         0.0132,  0.0121, -0.0136, -0.0028,  0.0118,  0.0141, -0.0059,  0.0087,
         0.0063,  0.0069, -0.0425,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1517,  0.0518,  0.0085, -0.0387,  0.0058, -0.0027, -0.0680,  0.0180,
        -0.1218, -0.0235, -0.0466,  0.0235, -0.0174, -0.0329, -0.0505, -0.1118,
        -0.0159, -0.0800, -0.0516,  0.0013,  0.0254,  0.0529,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0165, -0.0992,  0.0801, -0.0323, -0.0947, -0.0754, -0.0261,  0.0128,
        -0.0340, -0.0673, -0.0323, -0.0699,  0.0121, -0.1165, -0.0447, -0.0447,
        -0.0414, -0.0482, -0.0332,  0.0089,  0.0095,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0157, -0.4057,  0.0318,  0.0024, -0.0277,  0.0631, -0.0389, -0.0883,
         0.0026,  0.0375, -0.0006,  0.0247,  0.0368, -0.0042,  0.0067, -0.0598,
         0.0732, -0.0298,  0.0504,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0177, -0.0670, -0.0262, -0.0205, -0.0270, -0.0955, -0.0173, -0.0171,
         0.0126,  0.0092, -0.0053,  0.0018, -0.0073, -0.0223, -0.0094,  0.0115,
         0.0009, -0.0298, -0.0006, -0.0137,  0.0043, -0.0075, -0.0315,  0.0047,
        -0.0298, -0.0041, -0.0124, -0.0586, -0.0257, -0.1520, -0.0104, -0.0217,
        -0.0313, -0.0691,  0.0006,  0.0036,  0.0227,  0.0077, -0.0075, -0.0307,
         0.0101, -0.0003, -0.0012,  0.0071,  0.0036, -0.0164, -0.0126],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1119, -0.1476, -0.0503, -0.0134,  0.0117, -0.0023, -0.0178,  0.0131,
        -0.0302,  0.0025, -0.0054, -0.0020,  0.0056, -0.0059, -0.0044,  0.0132,
        -0.0997,  0.0026, -0.0487, -0.0074,  0.0321,  0.0152, -0.0217, -0.0610,
         0.0674, -0.0191, -0.0543, -0.0142,  0.0258,  0.0321, -0.0014,  0.0131,
        -0.0059, -0.0411,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1891, -0.2335, -0.0013,  0.0154,  0.0205, -0.0332, -0.0099, -0.0179,
        -0.0073, -0.0003,  0.0298, -0.0038, -0.0521, -0.0404, -0.0056, -0.0124,
        -0.0014,  0.0224,  0.0020, -0.0646, -0.0140,  0.0446, -0.0029,  0.0182,
         0.0012,  0.0108, -0.0092, -0.0153, -0.0035, -0.0085,  0.0062, -0.0239,
         0.0428, -0.0361,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0521, -0.2902, -0.0273,  0.0071, -0.0288, -0.0734, -0.0409, -0.0221,
         0.0097, -0.0574, -0.0380, -0.0270, -0.0330, -0.2079, -0.0443, -0.0152,
         0.0012, -0.0242,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1235,  0.0059,  0.0406,  0.0355,  0.0092, -0.0118,  0.0014, -0.0172,
         0.0781,  0.0917, -0.0441,  0.0307, -0.0175,  0.0213,  0.0515,  0.0478,
         0.0154,  0.0512,  0.0392,  0.0913, -0.0265,  0.0706,  0.0254,  0.0075,
        -0.0093, -0.0157, -0.0200,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0876, -0.0283,  0.0598, -0.0187,  0.0056,  0.0026, -0.0183, -0.0316,
        -0.1238, -0.0134, -0.0155,  0.0203, -0.1601, -0.0714, -0.0150,  0.0208,
        -0.0283, -0.0336, -0.0028, -0.0959, -0.0179,  0.0046, -0.0092,  0.0066,
         0.0076,  0.1008,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0120, -0.2949, -0.0102, -0.0143, -0.1354,  0.0717,  0.0686, -0.0175,
        -0.0190, -0.0528, -0.0446, -0.0104,  0.0525,  0.0658,  0.0255, -0.0029,
        -0.0046, -0.0162,  0.0175, -0.0150,  0.0485,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1137, -0.2132, -0.1253, -0.2427, -0.0051, -0.0658, -0.0419, -0.0156,
        -0.0114,  0.0577,  0.0103, -0.0289,  0.0311,  0.0186, -0.0187,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0908,  0.0451, -0.0139, -0.0598,  0.0743, -0.0599, -0.0284, -0.0439,
         0.0034, -0.1040,  0.0070, -0.0951, -0.1404,  0.0226,  0.0634, -0.1008,
         0.0388, -0.0085,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0432, -0.2140, -0.0960, -0.1173,  0.0046, -0.0040, -0.0554, -0.0027,
        -0.0173, -0.0824, -0.0751, -0.0149,  0.0379, -0.0576, -0.0479, -0.0511,
        -0.0032, -0.0541,  0.0214,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0125, -0.0537, -0.0179,  0.0090, -0.0112, -0.0149,  0.0349,  0.0055,
        -0.0165, -0.0069, -0.0245,  0.0013, -0.0112,  0.0010,  0.0097, -0.0131,
         0.0141, -0.0023,  0.0337, -0.0119, -0.0903,  0.0175, -0.0680, -0.0359,
        -0.0299, -0.0116, -0.0449,  0.0314, -0.0080, -0.0452, -0.0053, -0.0388,
        -0.0423,  0.0146, -0.0046, -0.0223, -0.0485,  0.0125, -0.0106,  0.0048,
         0.0121,  0.0262, -0.0169,  0.0523], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0554, -0.1387, -0.0146, -0.0119, -0.0450, -0.0583, -0.0071,  0.0067,
        -0.0879,  0.0361, -0.0366, -0.0214, -0.0054, -0.0012,  0.0237, -0.0022,
         0.0376, -0.0977,  0.0144, -0.0196, -0.0302, -0.0014, -0.0136, -0.0013,
        -0.0230, -0.0240, -0.0054, -0.0350,  0.0129, -0.0330, -0.0391, -0.0278,
        -0.0032, -0.0238, -0.0049,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0047, -0.3082, -0.0383, -0.0588, -0.0368,  0.0089, -0.0758, -0.0516,
        -0.0766,  0.0072,  0.0126, -0.0290, -0.0517, -0.0058,  0.0066, -0.0295,
         0.0106,  0.0058,  0.0234,  0.0030, -0.0115,  0.0068, -0.0901,  0.0467,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0101, -0.1731, -0.0194, -0.0245, -0.0454, -0.0553, -0.0148,  0.0169,
        -0.0278,  0.0062, -0.0084,  0.0071,  0.0206, -0.0452, -0.0142,  0.0061,
         0.0521,  0.0109, -0.1053, -0.0156, -0.0020, -0.0251, -0.0047, -0.0314,
        -0.0419, -0.0279, -0.0234, -0.0140, -0.0235, -0.0255, -0.0383,  0.0025,
        -0.0277,  0.0330,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1197, -0.0331, -0.0225, -0.0499, -0.0228,  0.0115, -0.0513, -0.0032,
        -0.0081, -0.0261, -0.0274, -0.0137, -0.0755, -0.0335, -0.0699, -0.0609,
        -0.0047,  0.0134, -0.0137, -0.0840,  0.0102, -0.0096, -0.0212, -0.0339,
        -0.0206, -0.0036, -0.0219, -0.0499, -0.0042, -0.0800,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0841,  0.0315, -0.1044, -0.3640, -0.0647, -0.0444, -0.0343, -0.0630,
        -0.0542, -0.1062, -0.0051, -0.0364,  0.0077,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1554, -0.0052, -0.0706,  0.0841, -0.0132, -0.0640,  0.0609, -0.0676,
        -0.1404, -0.0241, -0.0145, -0.0371, -0.0056, -0.0841, -0.1181, -0.0048,
         0.0403, -0.0100,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0223,  0.2346,  0.1369, -0.0122,  0.2536,  0.1182,  0.0350,  0.1872,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1051, -0.0158, -0.0036,  0.0016,  0.0011, -0.0111,  0.0077, -0.0122,
        -0.0055, -0.0020,  0.0011, -0.0066, -0.0104,  0.0013, -0.0096, -0.0008,
         0.0074,  0.0017,  0.0100,  0.0059,  0.0334, -0.0043, -0.0354, -0.0248,
        -0.0118, -0.0346,  0.0034, -0.0297, -0.1171, -0.0664,  0.0004, -0.0519,
        -0.0179, -0.1195, -0.0152, -0.0473, -0.0317, -0.0159,  0.0070, -0.0034,
        -0.0030, -0.0103, -0.0062,  0.0119,  0.0148, -0.0267, -0.0358],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1564,  0.1263,  0.0162,  0.1071,  0.0569,  0.0675, -0.0141, -0.1766,
        -0.0027, -0.1457,  0.1305,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0200,  0.2014,  0.0065,  0.1291, -0.0083, -0.0128,  0.0478,  0.1193,
         0.0646,  0.0777, -0.0209,  0.0110,  0.0953, -0.0103, -0.0332, -0.0149,
         0.1216,  0.0054,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1358, -0.2081, -0.2147,  0.1226,  0.0046,  0.0025, -0.0754, -0.0456,
        -0.0726, -0.0243,  0.0119,  0.0419, -0.0014,  0.0139,  0.0248,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0644, -0.4066, -0.0893, -0.0143, -0.0062, -0.0521, -0.1019, -0.0260,
        -0.0251,  0.0548, -0.0341,  0.0027,  0.0472, -0.0335,  0.0417,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0501,  0.0187,  0.0178, -0.0271, -0.0299, -0.0712,  0.0245,  0.0514,
         0.0512,  0.0640,  0.0580,  0.0652,  0.0782,  0.0062,  0.0044, -0.0022,
         0.0468, -0.0875,  0.0251, -0.0695,  0.0330,  0.0314, -0.0199, -0.0246,
         0.0375,  0.0044,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0523, -0.1393,  0.0093, -0.0996, -0.0176, -0.1089,  0.0242, -0.1208,
        -0.3000, -0.0212, -0.0302, -0.0493,  0.0086,  0.0149, -0.0039,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0061, -0.2128, -0.0437, -0.0208, -0.0743, -0.0243, -0.0494,  0.0086,
         0.0024, -0.1115,  0.0055,  0.0131, -0.0162, -0.0057,  0.0086, -0.0291,
        -0.0196,  0.0427, -0.0405,  0.0330, -0.0264,  0.0061,  0.0118,  0.0399,
        -0.0406, -0.1073,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.2625e-02, -4.3882e-03, -1.4787e-02, -6.6010e-02, -6.1809e-02,
        -5.8303e-02,  3.3641e-05,  2.8160e-03, -1.5843e-02, -6.2374e-02,
        -1.9380e-01, -3.0261e-02, -8.1363e-03, -2.5102e-03, -1.3015e-01,
        -3.6306e-02, -4.5111e-02, -7.0657e-02, -5.9249e-03, -2.3326e-02,
        -1.4919e-02, -2.0745e-02,  2.6069e-05, -2.9541e-02, -2.3436e-04,
         5.9364e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-6.4375e-02, -2.2066e-01, -2.1782e-02, -2.9984e-02, -7.9795e-03,
         2.1861e-02, -5.1450e-02, -2.4883e-02,  3.3650e-03, -8.9037e-03,
         5.3624e-03, -2.1717e-02, -2.4866e-02,  3.5232e-04, -1.1382e-02,
        -6.4622e-03, -4.9865e-02,  3.4965e-02, -5.3263e-02, -8.7568e-03,
        -2.7626e-02, -4.0735e-02, -7.6020e-03, -6.1455e-02, -4.2945e-02,
        -2.5457e-02, -4.3648e-02,  3.7606e-03,  7.7382e-03,  1.9289e-02,
         1.4456e-02, -1.3068e-02, -1.6481e-05, -6.3521e-04,  7.0790e-03,
        -1.2256e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0590, -0.0888, -0.0472, -0.0451, -0.0510, -0.0269, -0.0269, -0.0938,
        -0.0015,  0.0158,  0.0036,  0.0063, -0.0357,  0.0394, -0.0156, -0.0054,
         0.0058, -0.0040,  0.0147, -0.0120, -0.0153, -0.0481, -0.0298, -0.0515,
        -0.0345, -0.0281,  0.0152, -0.0290, -0.0570, -0.0236, -0.0005, -0.0147,
        -0.0171,  0.0098, -0.0206,  0.0064,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1886,  0.1755,  0.0937,  0.0267,  0.0115,  0.0690,  0.0257,  0.0040,
         0.0211, -0.0617, -0.0435,  0.0261, -0.0004, -0.0237, -0.0202, -0.0377,
         0.0770,  0.0321,  0.0262, -0.0358,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1266, -0.0803, -0.0055, -0.0126,  0.0191, -0.0041, -0.0465, -0.0217,
        -0.0032, -0.0259,  0.0005, -0.0230, -0.0059,  0.0027, -0.0516, -0.0423,
         0.0024, -0.0215,  0.0018, -0.0091,  0.0002, -0.0113, -0.0002, -0.0191,
        -0.0444, -0.0142, -0.0063, -0.0114, -0.0136, -0.0669, -0.0257,  0.0031,
        -0.0272, -0.1048,  0.0337,  0.0123, -0.0408, -0.0583,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0791, -0.0933, -0.0369, -0.1145, -0.0099,  0.0020, -0.0082, -0.0105,
         0.0105, -0.0090, -0.0467, -0.0595, -0.0160,  0.0049,  0.0035, -0.0719,
         0.0235, -0.0410, -0.0132, -0.0035,  0.0181,  0.0209,  0.0067, -0.0162,
         0.0117, -0.0019,  0.0028, -0.0210, -0.0911,  0.0023,  0.0018,  0.0116,
         0.0289,  0.0095, -0.0215, -0.0005, -0.0083,  0.0386, -0.0026,  0.0087,
        -0.0177], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1615, -0.3745, -0.0319, -0.0290, -0.0212, -0.0198,  0.0125, -0.0043,
        -0.0180, -0.0105,  0.0079,  0.0097, -0.0410, -0.0224, -0.0062, -0.0329,
        -0.0208, -0.0422,  0.0066,  0.0369, -0.0125,  0.0778,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0539, -0.0366, -0.0107,  0.0048,  0.0004, -0.0517, -0.0076,  0.0221,
        -0.0058, -0.0140, -0.0217, -0.0628, -0.0272, -0.0358,  0.0064, -0.0650,
        -0.0063, -0.0723, -0.0114, -0.0938, -0.0712,  0.0079,  0.0136, -0.0212,
        -0.0299,  0.0030, -0.0249, -0.0360, -0.0573, -0.0398, -0.0205, -0.0255,
        -0.0211,  0.0135, -0.0043,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1617, -0.0478, -0.0683, -0.0300, -0.0436, -0.0172,  0.0036, -0.0191,
        -0.0374, -0.0141, -0.0321, -0.0061, -0.0231,  0.0173,  0.0074, -0.0617,
         0.0548, -0.0184, -0.0954, -0.0407, -0.0238, -0.0128, -0.0266, -0.0143,
        -0.0223,  0.0040, -0.0293, -0.0479,  0.0193,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0099,  0.2927,  0.0820,  0.0288,  0.0600,  0.0471,  0.0178,  0.0275,
         0.0284,  0.0205,  0.0242,  0.0422, -0.0307,  0.0020,  0.0185,  0.0188,
         0.0757,  0.0314,  0.0379,  0.0022,  0.0157,  0.0154, -0.0058,  0.0171,
        -0.0476,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0878,  0.0911, -0.0354,  0.0930, -0.0835, -0.0330, -0.0047, -0.0823,
        -0.0484,  0.0070,  0.0098,  0.0099,  0.0038,  0.0311, -0.0531, -0.0475,
        -0.0222, -0.0532, -0.0267, -0.0264, -0.0340,  0.0365,  0.0791,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1156, -0.0153,  0.0073, -0.0038, -0.0108,  0.0188, -0.0092, -0.0155,
        -0.0008, -0.0068,  0.0102,  0.0193, -0.0439, -0.1275, -0.0117, -0.0529,
        -0.0194, -0.0206, -0.0189, -0.0040, -0.0053, -0.0032,  0.0091, -0.0108,
        -0.0313, -0.0209, -0.1365, -0.0398, -0.0686, -0.0167, -0.0256, -0.0033,
        -0.0194, -0.0262, -0.0080, -0.0062,  0.0115,  0.0158,  0.0095,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1183, -0.1311, -0.0640, -0.1153, -0.1459, -0.0423, -0.0236, -0.1424,
        -0.0823, -0.0539,  0.0128,  0.0679,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.0832, -0.0850, -0.0450, -0.0195, -0.0063, -0.0089, -0.0357,  0.0085,
        -0.0160, -0.0293, -0.0138, -0.0159, -0.0270,  0.0184, -0.0288,  0.0026,
        -0.0065, -0.0250, -0.0079, -0.0305, -0.0089, -0.0137,  0.0024,  0.0060,
         0.0121, -0.0092, -0.0215,  0.0179,  0.0636,  0.0055, -0.0256, -0.0427,
        -0.0270, -0.0046, -0.0398, -0.0243, -0.0362, -0.0143, -0.0006, -0.0039,
         0.0099, -0.0229,  0.0078, -0.0497, -0.0161,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0834,  0.3796, -0.0424,  0.0070, -0.1150,  0.0241,  0.0905,  0.0632,
         0.0974, -0.0271, -0.0563, -0.0140,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.3250e-04, -3.8807e-01, -4.6496e-02, -1.9267e-01,  9.1399e-02,
         1.1505e-01,  1.1277e-01, -5.3418e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0080, -0.1600, -0.0113, -0.0469, -0.0259, -0.0198, -0.0009, -0.0031,
        -0.0063, -0.0683, -0.0249, -0.0105, -0.0034,  0.0025, -0.0090,  0.0214,
         0.0374, -0.0171,  0.0051,  0.0052, -0.0113,  0.0059,  0.0053, -0.0077,
        -0.0097, -0.0268, -0.0051, -0.0089, -0.0120,  0.0077, -0.0337, -0.0058,
        -0.0113, -0.0012, -0.0109, -0.0094, -0.0071,  0.0028,  0.0027, -0.0123,
        -0.0154, -0.0112, -0.0077, -0.0148,  0.0220,  0.0003, -0.0205, -0.0489,
        -0.0324, -0.0437, -0.0194,  0.0027, -0.0173, -0.0127, -0.0153, -0.0005,
        -0.0060,  0.0030, -0.0042,  0.0165,  0.0008], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0722, -0.0209,  0.0286, -0.0183, -0.1040, -0.0127, -0.0533, -0.0381,
        -0.0081,  0.0129,  0.0048, -0.0366, -0.0185, -0.0489, -0.0916, -0.0062,
         0.0016,  0.0031, -0.0477, -0.0161, -0.0492,  0.0077, -0.0562, -0.0223,
        -0.0360, -0.0289, -0.0054, -0.0001, -0.0414, -0.0594, -0.0012,  0.0419,
        -0.0059,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1963, -0.0484, -0.0636, -0.0142, -0.0041, -0.0019, -0.0749, -0.0284,
         0.0145,  0.0274, -0.0059, -0.0075, -0.0009,  0.0132, -0.0121, -0.0601,
         0.0364, -0.0378, -0.2223,  0.0047,  0.0217, -0.0034, -0.0237,  0.0768,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0329, -0.0370, -0.0871, -0.0345, -0.0392, -0.1350, -0.0313, -0.0485,
         0.0239, -0.0460, -0.0754, -0.0468, -0.0818, -0.0325,  0.0553, -0.0173,
         0.0131, -0.0310,  0.0069,  0.0011, -0.0284,  0.0016, -0.0458, -0.0474,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.4395e-01, -4.3663e-02, -4.1291e-02, -5.6414e-02,  1.5990e-03,
        -2.7829e-03, -1.5174e-02, -1.2180e-02, -4.0616e-03, -8.7327e-03,
         1.5052e-03, -1.2087e-04,  1.8896e-02,  1.7594e-04, -3.7374e-02,
        -5.2145e-02, -5.4105e-02, -1.6795e-02, -1.2945e-02, -1.4728e-02,
        -3.9812e-02, -7.0143e-03, -2.4310e-02, -7.6512e-02, -7.0075e-03,
         3.5704e-02, -6.6863e-02, -4.5059e-03, -7.2967e-03,  2.4857e-02,
        -3.3254e-02,  2.1448e-02, -6.1054e-02,  5.1725e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0807,  0.0231,  0.0153,  0.0359, -0.0175,  0.0180,  0.0163,  0.0186,
        -0.0039, -0.0327,  0.0099,  0.0015, -0.0089, -0.3509, -0.0024, -0.0233,
        -0.1032, -0.0281, -0.0412, -0.0370,  0.0108, -0.0476, -0.0234, -0.0237,
        -0.0265,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0343, -0.0649, -0.0228,  0.0073, -0.0599,  0.0359, -0.0859, -0.0102,
        -0.0886, -0.3865,  0.0095, -0.0400, -0.0054, -0.0442, -0.0383,  0.0663,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0119, -0.1664, -0.0242, -0.0829, -0.0335,  0.0138, -0.2612, -0.0155,
         0.0296, -0.0054, -0.0106, -0.0190, -0.0320, -0.0170, -0.0228,  0.0128,
         0.0132, -0.0450, -0.0389, -0.0045,  0.0184,  0.0948,  0.0264,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.7246e-05,  2.1253e-01, -1.8865e-02,  2.1739e-01,  2.7221e-01,
        -3.3316e-02, -3.4280e-02, -2.5006e-03,  3.7762e-02,  2.5689e-02,
         2.5384e-02,  3.2544e-02, -4.8471e-02,  3.8997e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0462, -0.0461, -0.0466, -0.1149, -0.0120,  0.0365, -0.0211,  0.0310,
        -0.0341, -0.0230, -0.0634, -0.0491, -0.0711, -0.0159, -0.0022, -0.0346,
         0.0158, -0.1245, -0.0355, -0.0504, -0.0667, -0.0086,  0.0135, -0.0372,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0739, -0.0514, -0.1550, -0.0092, -0.1087, -0.0568, -0.3868, -0.0229,
        -0.0580, -0.0559, -0.0214,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0524, -0.0045, -0.1496, -0.0435, -0.0535, -0.0697, -0.1471,  0.0761,
        -0.0619,  0.0050,  0.0023,  0.0053, -0.0198,  0.0118, -0.0544,  0.0183,
        -0.0300, -0.0859, -0.0086, -0.0731,  0.0270,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.3197, -0.0179, -0.0201, -0.1553,  0.0050,  0.0188, -0.0330, -0.0977,
        -0.0271, -0.1284, -0.0358,  0.0488, -0.0235,  0.0690,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1416, -0.0177, -0.0039, -0.0328, -0.0524, -0.0799, -0.0260,  0.0019,
         0.0296, -0.0215,  0.0353, -0.0103, -0.3250, -0.0389,  0.0130,  0.0151,
         0.0258,  0.0265,  0.0043, -0.0305,  0.0679,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1316, -0.2610, -0.0164, -0.0811, -0.0474, -0.0790, -0.0182, -0.2053,
         0.0071, -0.0414,  0.0008, -0.0453,  0.0655,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0280, -0.0473,  0.0144,  0.0086, -0.0037, -0.0059, -0.0123, -0.0475,
        -0.0050,  0.0203,  0.0107,  0.0076,  0.0120,  0.0210, -0.0025, -0.0123,
        -0.0029,  0.0251, -0.0042,  0.0290, -0.0567, -0.0040, -0.0014,  0.0116,
        -0.0674, -0.0005, -0.1214, -0.0933, -0.0541, -0.0104, -0.0195,  0.0167,
         0.0080,  0.0036,  0.0330,  0.0109, -0.0429, -0.0061, -0.0069, -0.0033,
        -0.0142,  0.0109, -0.0198,  0.0384,  0.0244], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.9021e-02, -6.2609e-02, -1.9964e-02,  1.2976e-02, -2.9878e-03,
         1.6373e-04, -2.6869e-02, -1.6106e-02, -2.4024e-03,  9.9464e-03,
        -5.3847e-02,  1.2152e-02,  1.7476e-02,  1.5806e-03, -2.0157e-02,
        -1.3947e-02, -2.3570e-01, -1.3862e-02, -6.0412e-03, -1.7080e-02,
        -3.5715e-02, -2.1174e-02, -1.2325e-02,  1.2326e-02,  1.2942e-02,
        -4.8427e-03, -4.6967e-03, -7.1858e-03, -4.5339e-02, -1.5785e-02,
         9.0260e-03, -2.6296e-02,  8.3669e-02,  1.4379e-01,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.8314e-02,  5.8599e-02,  4.6816e-02,  4.9237e-02,  2.2829e-02,
         1.6125e-03, -1.2634e-02,  2.0615e-02, -7.7080e-02,  1.0426e-02,
         2.8458e-02,  2.4341e-03,  1.7624e-02,  3.7270e-05,  1.7978e-02,
        -2.5857e-02, -4.6788e-02, -2.5625e-02,  2.8220e-02,  1.0296e-01,
         1.7513e-01, -3.0653e-02,  1.0715e-02, -1.2380e-01,  5.8665e-03,
         3.9691e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1097, -0.2515, -0.0214, -0.0201,  0.0165,  0.0007, -0.0113, -0.0307,
        -0.0251,  0.0065, -0.0325, -0.0019, -0.0054, -0.0306, -0.0537, -0.0417,
        -0.1391,  0.0065, -0.0389, -0.0192, -0.0232, -0.0406,  0.0733,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0131, -0.0703,  0.0392, -0.0100,  0.0030, -0.0008, -0.0056, -0.0137,
        -0.0424,  0.0006, -0.0209,  0.0097, -0.0396, -0.0307, -0.2005,  0.0167,
         0.0403, -0.0310, -0.0378, -0.0187, -0.0018, -0.0034,  0.0091,  0.0285,
        -0.0041,  0.0029, -0.0436, -0.0756,  0.0359,  0.0158,  0.0104, -0.0254,
        -0.0007,  0.0055,  0.0272, -0.0179, -0.0172,  0.0049,  0.0052, -0.0206,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1482, -0.0753, -0.0110, -0.0345, -0.0438, -0.0225, -0.0599, -0.0493,
        -0.0177, -0.0160, -0.0055, -0.0108, -0.0333, -0.0223, -0.0291, -0.0319,
        -0.0279, -0.0155,  0.0084, -0.0198,  0.0101, -0.0163, -0.0062,  0.0311,
        -0.0075, -0.0117,  0.0084, -0.0339, -0.0285, -0.0105, -0.0768, -0.0326,
         0.0304, -0.0134,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.1142, -0.0771,  0.0300, -0.0099, -0.0323,  0.0317, -0.0148, -0.0216,
        -0.0010,  0.0231, -0.0123, -0.0123,  0.0080, -0.0232, -0.0757,  0.0209,
        -0.0146,  0.0009, -0.0341, -0.0005,  0.0032, -0.0074, -0.0069,  0.0122,
        -0.0276, -0.0523, -0.0329, -0.0303, -0.0402, -0.0053, -0.0131, -0.0038,
         0.0034, -0.0154, -0.0520, -0.0004, -0.0192, -0.0166, -0.0006, -0.0160,
        -0.0149,  0.0090, -0.0074, -0.0055, -0.0052, -0.0050, -0.0024,  0.0013,
        -0.0157, -0.0050,  0.0096, -0.0019], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0371, -0.0512,  0.0132,  0.0083, -0.0457,  0.0294, -0.0355,  0.0538,
        -0.0032,  0.0373, -0.0131, -0.0120, -0.0269, -0.1026, -0.0379, -0.0450,
        -0.0589, -0.0481, -0.0362, -0.0277, -0.0205, -0.0323, -0.0184, -0.0576,
        -0.0056, -0.0027,  0.0029, -0.0195, -0.0530,  0.0050, -0.0592,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0419,  0.0980, -0.0262,  0.0219,  0.0074,  0.0037,  0.0107, -0.0021,
         0.0284, -0.0319, -0.0006,  0.0591,  0.0438,  0.0037,  0.0476,  0.0428,
        -0.0658,  0.0348,  0.0805,  0.0848,  0.0242,  0.0393,  0.0832,  0.0119,
        -0.1056,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2287,  0.0263, -0.0685, -0.0489, -0.0271, -0.0505, -0.3477, -0.0607,
         0.0283, -0.0399,  0.0556,  0.0179,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0457, -0.3707, -0.0175, -0.0291, -0.0627,  0.0042,  0.0052, -0.0059,
        -0.0035,  0.0100, -0.0119,  0.0225, -0.0213, -0.0573, -0.0312, -0.0204,
         0.0109, -0.0105,  0.0215,  0.0039, -0.0129,  0.0305,  0.0186,  0.0215,
        -0.0135, -0.0719, -0.0204,  0.0112, -0.0337,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0153, -0.3968, -0.0089, -0.0363, -0.0306, -0.1063, -0.1541,  0.0699,
        -0.0705, -0.0475,  0.0298, -0.0342,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0301, -0.3933,  0.0297, -0.0202, -0.0068, -0.0665, -0.2143,  0.0298,
        -0.0881, -0.0136, -0.0352,  0.0724,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1536,  0.0685, -0.0276, -0.0105, -0.0865, -0.0143, -0.0144, -0.0126,
        -0.1222, -0.0607, -0.0005, -0.1630, -0.0149, -0.0268,  0.0769, -0.0589,
        -0.0060, -0.0187, -0.0196, -0.0438,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.7401e-02, -2.9315e-01, -7.4324e-03, -7.9203e-03, -3.4458e-02,
        -1.4694e-02, -3.8370e-02,  7.8204e-03,  9.7328e-03, -8.1937e-03,
        -1.5905e-02, -3.8705e-02, -2.6872e-02,  1.3228e-02,  2.9672e-02,
         1.7668e-02,  7.4621e-03,  1.3721e-02, -1.7477e-04, -2.8486e-02,
        -8.6990e-03, -5.1735e-03,  1.8982e-02, -9.7616e-03,  1.6526e-02,
         8.4413e-03, -2.7237e-02, -1.6381e-02, -1.7541e-02,  3.8708e-03,
        -6.6449e-03, -2.1073e-03, -3.4706e-02, -1.3798e-01, -2.1177e-02,
         9.3757e-03,  2.4331e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1582,  0.4751, -0.0097, -0.0433, -0.0491, -0.0318,  0.0652, -0.0720,
         0.0956,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0050,  0.0535,  0.0990, -0.0992, -0.0223, -0.0701, -0.0699, -0.0733,
         0.0049, -0.0361,  0.0007, -0.0358, -0.0644, -0.0720,  0.0149,  0.0265,
        -0.0131, -0.0130, -0.0574, -0.0481, -0.0133,  0.0179, -0.0109, -0.0116,
        -0.0179,  0.0249, -0.0243,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1428, -0.1612,  0.0108, -0.0108, -0.0186, -0.0080,  0.0118, -0.0096,
        -0.0334, -0.0402,  0.0234, -0.0209, -0.0203,  0.0155, -0.0167, -0.0037,
        -0.0178,  0.0131,  0.0267, -0.0175,  0.0594, -0.0133, -0.0207,  0.0006,
        -0.0094, -0.0188,  0.0010, -0.0154, -0.0677,  0.0051,  0.0100, -0.0143,
         0.1118, -0.0297,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0295, -0.0008,  0.0062,  0.0051, -0.0277, -0.0057,  0.0096,  0.0100,
        -0.0043, -0.0048, -0.0183, -0.0676, -0.0515, -0.0086,  0.0043, -0.0421,
        -0.0740,  0.0008,  0.0020,  0.0015,  0.0142,  0.0008, -0.0395, -0.0240,
         0.0006, -0.0125, -0.0176,  0.0082,  0.0100,  0.0129, -0.0093, -0.0387,
        -0.1075, -0.0233, -0.0087, -0.0289, -0.0728, -0.0102,  0.0214, -0.0008,
        -0.0038, -0.0710,  0.0101,  0.0131,  0.0659,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0432, -0.0826, -0.0759,  0.0129, -0.0205, -0.1141, -0.2039,  0.0083,
        -0.0062, -0.1050, -0.0289, -0.0271, -0.1333, -0.0437,  0.0148, -0.0796,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0319, -0.0440,  0.0090,  0.0083, -0.0121,  0.0507,  0.0727, -0.0476,
         0.0634,  0.0608,  0.0481,  0.1345,  0.0003,  0.0110,  0.0675, -0.0288,
         0.0290, -0.0107,  0.0107,  0.0766,  0.0068, -0.0013, -0.0017,  0.0458,
        -0.0244, -0.0010,  0.1013,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0139,  0.0189, -0.0768, -0.0126, -0.0248, -0.0070, -0.0284, -0.0254,
        -0.0201, -0.0026, -0.0075, -0.0087, -0.0146,  0.0019,  0.0020,  0.0033,
        -0.0089, -0.0015,  0.0119, -0.0226,  0.0308, -0.0339,  0.0113, -0.1092,
        -0.0015, -0.0162, -0.0259, -0.0462,  0.0056, -0.0576, -0.0255, -0.0141,
        -0.0507,  0.0244, -0.0177, -0.0185,  0.0013, -0.0394, -0.0163, -0.0061,
        -0.0089, -0.0388,  0.0032, -0.0171,  0.0294,  0.0370], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.9713e-02,  8.2199e-03, -1.7119e-02, -3.4523e-02,  6.3329e-03,
        -6.1628e-02,  8.7440e-05, -6.0869e-02, -5.0870e-03, -1.0550e-02,
        -2.1566e-02,  3.8820e-02,  1.3545e-02, -2.2911e-02, -2.8285e-02,
        -4.1054e-03, -2.6899e-02, -1.3447e-02, -5.0792e-02, -1.8479e-01,
         2.9471e-03, -4.5112e-02, -1.5156e-02, -9.9263e-03, -1.8957e-02,
        -7.0230e-02, -4.1146e-02,  1.1318e-01,  1.3722e-02,  1.0338e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.3676, -0.0499, -0.1672, -0.0545,  0.0099,  0.0787,  0.0786, -0.1559,
        -0.0043,  0.0308,  0.0025,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0184, -0.2157, -0.0389, -0.0740, -0.0232, -0.0150, -0.0434, -0.1840,
        -0.0144, -0.1504,  0.0016, -0.0074,  0.0038, -0.0172, -0.0199, -0.0349,
        -0.0439, -0.0229, -0.0157, -0.0553,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0390,  0.3976,  0.1586,  0.0942, -0.0319,  0.1066,  0.0635, -0.0106,
         0.0310,  0.0233, -0.0438,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1108, -0.1424, -0.0633, -0.0901, -0.0437,  0.0216, -0.0223, -0.0367,
        -0.0757, -0.0286, -0.0061,  0.0065, -0.0237, -0.0519, -0.0323, -0.0417,
        -0.0069, -0.0199,  0.0002, -0.0642,  0.1115,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0171,  0.0786,  0.0663, -0.0127,  0.0265,  0.2441, -0.2054, -0.0265,
         0.0674, -0.0134,  0.0204,  0.0617,  0.0034,  0.0062, -0.0021,  0.1204,
         0.0189,  0.0089,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0397, -0.0662, -0.1820,  0.0127,  0.0048, -0.0053, -0.0089,  0.0096,
        -0.0106,  0.0022, -0.0021,  0.0100, -0.0017, -0.0311, -0.0521,  0.0089,
        -0.0278, -0.0525, -0.0330, -0.0041, -0.0088, -0.0393, -0.0929, -0.0008,
        -0.0194, -0.0079, -0.0337, -0.0272, -0.0235, -0.0724, -0.0090, -0.0090,
        -0.0162,  0.0011, -0.0224, -0.0051, -0.0030, -0.0204,  0.0223,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0293, -0.3523,  0.0152, -0.0228, -0.0076, -0.0122,  0.0283,  0.0019,
         0.0303,  0.0140, -0.0232, -0.0377, -0.0286, -0.0099, -0.0215, -0.0225,
        -0.0592, -0.0068, -0.0030, -0.0382, -0.0067, -0.0042, -0.0032,  0.0079,
        -0.0032, -0.0053, -0.0214, -0.0011, -0.0572, -0.0678, -0.0040,  0.0030,
        -0.0326,  0.0178,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0852, -0.3105, -0.1402, -0.1978,  0.0326, -0.0552,  0.0749, -0.0310,
        -0.0728,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.8001e-03, -2.1114e-01, -4.9332e-02, -1.7721e-02,  1.7020e-02,
         1.2577e-02, -8.6643e-03,  9.3382e-05, -2.6478e-03, -2.4866e-02,
        -1.5376e-02, -1.0941e-02, -3.7282e-03, -6.4270e-03,  4.9443e-03,
        -6.8917e-02, -3.2788e-02, -7.7182e-02, -2.7627e-02, -2.6983e-02,
         2.3979e-02,  8.5380e-03, -5.2908e-02,  8.5961e-03, -4.1931e-02,
        -3.3202e-02, -3.1487e-02,  1.5399e-03,  1.5548e-02, -3.2205e-03,
        -9.0942e-03, -4.0886e-02,  5.5938e-03, -1.3565e-02,  4.6157e-03,
        -1.5385e-02,  1.8181e-02, -3.2666e-03,  4.6425e-03,  3.6046e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0202, -0.0494, -0.1202, -0.0228, -0.0232, -0.0054,  0.0110, -0.0103,
        -0.0328, -0.0682, -0.0029, -0.0041, -0.0070,  0.0033, -0.0011,  0.0127,
         0.0003, -0.0098, -0.0172, -0.0020, -0.0139, -0.0060,  0.0421, -0.0171,
         0.0133, -0.0199, -0.0371,  0.0047, -0.0352, -0.0966, -0.0097, -0.0346,
        -0.0010, -0.0025, -0.0101,  0.0100, -0.0301, -0.0217, -0.0016, -0.0049,
         0.0131, -0.0253,  0.0022, -0.0071, -0.0151,  0.0054,  0.0052, -0.0141,
         0.0173, -0.0148, -0.0038,  0.0408,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0140,  0.1907,  0.0440,  0.0135,  0.0415,  0.0616,  0.0326,  0.0152,
         0.0158, -0.0229,  0.0681,  0.0285,  0.0483,  0.0231,  0.0423,  0.0457,
         0.0198, -0.0310,  0.0563,  0.0717,  0.0311,  0.0823,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0404, -0.3788, -0.0429, -0.0099, -0.0690,  0.0216, -0.0783, -0.0128,
         0.0319,  0.0030, -0.0418, -0.0049, -0.0404, -0.0177,  0.0098, -0.0851,
        -0.0558,  0.0370, -0.0029, -0.0133, -0.0029,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.6208e-02, -5.0944e-02, -7.2852e-03, -1.3522e-02,  8.6278e-03,
         1.0781e-02, -2.5000e-02, -2.4921e-03, -1.3066e-02, -1.0015e-02,
        -2.0972e-03, -1.1303e-02,  5.4480e-03,  2.0307e-02,  7.1575e-03,
        -8.8773e-05,  1.2721e-02, -6.1139e-03, -8.8258e-03,  5.2031e-03,
         3.0360e-03, -6.7357e-03, -1.2072e-03, -8.3365e-03, -9.0137e-03,
         1.7948e-02, -3.0132e-02, -3.9670e-02, -1.3389e-02, -1.8216e-02,
        -7.6771e-02, -4.2549e-02, -1.1087e-02, -2.0323e-02, -6.8855e-03,
        -8.8732e-03, -8.3600e-04,  2.0788e-02,  6.2437e-03, -1.7841e-02,
        -1.0144e-01, -2.2020e-02, -1.6285e-02, -6.3306e-04, -3.9523e-02,
        -1.2905e-02, -2.5334e-03, -2.3426e-02,  8.2141e-04, -3.6636e-03,
        -3.8965e-02, -3.1009e-03, -1.7682e-02, -5.2797e-02,  5.7115e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0355, -0.0074,  0.0183, -0.0226, -0.0015, -0.0190, -0.0262,  0.0014,
        -0.0215,  0.0039,  0.0101,  0.0052, -0.0014, -0.0235, -0.0058, -0.0157,
        -0.0203, -0.0252, -0.0248, -0.0327, -0.0391, -0.0178, -0.0316,  0.0058,
         0.0118,  0.0225, -0.0271, -0.0930, -0.0352, -0.0320, -0.0297,  0.0019,
        -0.0280, -0.1016, -0.0045, -0.0309, -0.0116, -0.0381, -0.0371, -0.0209,
        -0.0115,  0.0051,  0.0086, -0.0301, -0.0026,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0987, -0.1600, -0.0461, -0.0588,  0.0332, -0.0356, -0.0153, -0.0046,
        -0.0555,  0.0473, -0.0061, -0.0316, -0.0046, -0.0550, -0.0556,  0.0429,
        -0.0352, -0.0733,  0.0356, -0.0114, -0.0138,  0.0254, -0.0145,  0.0398,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0894, -0.0034, -0.0039,  0.0049, -0.0008,  0.0271, -0.0124, -0.0260,
        -0.0108,  0.0087,  0.0136,  0.0100, -0.0049, -0.0612, -0.0389,  0.0236,
        -0.0678, -0.0337, -0.0254, -0.0475,  0.0255, -0.0072, -0.0457, -0.0635,
         0.0163, -0.0260, -0.0917, -0.0221, -0.0250, -0.0815, -0.0047, -0.0399,
        -0.0368,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0170, -0.3258, -0.0698, -0.0740,  0.0345, -0.0353,  0.0297, -0.0363,
        -0.1331,  0.0845,  0.0140, -0.0012,  0.0030, -0.0281,  0.0302,  0.0143,
         0.0141, -0.0400,  0.0151,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0167, -0.0095,  0.0221, -0.0562, -0.0766, -0.0189, -0.1229, -0.0160,
        -0.1123, -0.0127,  0.0164,  0.0712, -0.0149, -0.0037, -0.0121, -0.0103,
        -0.0220, -0.0705, -0.0147, -0.0055, -0.0467, -0.1978, -0.0227, -0.0169,
        -0.0106,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0444, -0.2828,  0.0057,  0.0136,  0.0016, -0.1287, -0.0839, -0.0179,
        -0.0965, -0.0386, -0.0278, -0.0244, -0.0259,  0.0271, -0.0594, -0.0534,
         0.0683,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0143,  0.0241, -0.0511,  0.0132,  0.0028, -0.0447,  0.0107, -0.0116,
         0.0301, -0.0537,  0.0088, -0.0215,  0.0195,  0.0177, -0.0140, -0.0258,
        -0.0174, -0.0964,  0.0107, -0.0585, -0.0100, -0.1098, -0.1514,  0.0002,
         0.0355,  0.0417, -0.0074, -0.0157, -0.0197, -0.0109,  0.0006,  0.0506,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0687, -0.0937, -0.0411, -0.0277, -0.0353, -0.0210, -0.0390, -0.1282,
        -0.0123, -0.0076,  0.0003, -0.0084,  0.0235, -0.0480, -0.0305, -0.0040,
        -0.0274, -0.0265, -0.0545, -0.0097, -0.0022, -0.0006, -0.0559, -0.1123,
        -0.0125, -0.0222,  0.0595, -0.0274,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0591, -0.0242, -0.0115,  0.0492,  0.0174, -0.0604, -0.2174, -0.0089,
        -0.0182,  0.0016,  0.0145, -0.0954, -0.0037, -0.0241,  0.0015, -0.0256,
        -0.0145, -0.0854,  0.0114, -0.0028, -0.0176, -0.0037,  0.0106, -0.0492,
        -0.0708, -0.0091,  0.0268, -0.0046, -0.0608,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0873, -0.0147, -0.0090, -0.0312, -0.1214, -0.0187,  0.0134, -0.0110,
        -0.0002,  0.0339, -0.0184, -0.0245, -0.0421, -0.1077,  0.0136, -0.0324,
        -0.0255, -0.1287, -0.0204, -0.0367, -0.0052, -0.0009, -0.0084, -0.0119,
        -0.0315, -0.0137, -0.0178, -0.0081,  0.0095,  0.0018, -0.0090, -0.0911,
        -0.0003,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0709, -0.1479, -0.0054, -0.0283, -0.0369, -0.0650,  0.0464, -0.0054,
        -0.0259, -0.0401, -0.0272, -0.0138,  0.0012, -0.0225, -0.0500, -0.0022,
        -0.0197, -0.0036, -0.1469, -0.0219, -0.0997, -0.0298, -0.0341, -0.0551,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0501, -0.1386, -0.1054, -0.0256, -0.0641, -0.0676, -0.0730, -0.0644,
        -0.0027, -0.0470, -0.0237, -0.0280, -0.0481,  0.0026, -0.0184, -0.0072,
         0.0064, -0.0279, -0.0102, -0.1012, -0.0230,  0.0511, -0.0140,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0332, -0.0836,  0.0240,  0.0271,  0.0014, -0.0447,  0.0092, -0.0328,
         0.0011, -0.0124,  0.0066, -0.0088, -0.0157,  0.0148,  0.0055,  0.0007,
        -0.0507, -0.0419, -0.0224, -0.0170,  0.0185, -0.0401,  0.0058, -0.0284,
        -0.0631, -0.0198, -0.0162, -0.0334, -0.0212, -0.0712, -0.0725, -0.0085,
        -0.0185,  0.0185, -0.0221, -0.0191, -0.0145,  0.0089, -0.0461,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0886, -0.0655, -0.0152, -0.0166, -0.1131, -0.0075, -0.0040,  0.0037,
        -0.0384,  0.0288, -0.0048, -0.0201,  0.0021,  0.0146, -0.0106, -0.0309,
         0.0119, -0.0045,  0.0023, -0.0212, -0.0620, -0.0219, -0.0195, -0.0321,
        -0.0530, -0.0518, -0.0049, -0.0146,  0.0052, -0.0276, -0.0221, -0.0248,
        -0.0145, -0.0116, -0.0060, -0.0062, -0.0120, -0.0568, -0.0037, -0.0060,
        -0.0019, -0.0235, -0.0142,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0662, -0.1762,  0.0043,  0.0150,  0.0168, -0.0112, -0.0209, -0.0123,
        -0.0427,  0.0080,  0.0233, -0.0024, -0.0165,  0.0056,  0.0053, -0.0222,
        -0.0148, -0.0376,  0.0087, -0.0078, -0.0030, -0.0193, -0.0209, -0.0082,
         0.0041, -0.0015,  0.0222,  0.0032, -0.0135,  0.0086,  0.0031, -0.0141,
        -0.0256,  0.0315,  0.0028, -0.0011, -0.0354, -0.0191, -0.0227, -0.0078,
        -0.0229, -0.0162, -0.0032, -0.0026,  0.0019, -0.0166, -0.0066, -0.0120,
         0.0080, -0.0106, -0.0039, -0.0053, -0.0387,  0.0019,  0.0214,  0.0163,
        -0.0262], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0101, -0.5003, -0.0748, -0.0056, -0.0455, -0.0165, -0.0252, -0.1045,
        -0.0246, -0.0620, -0.0857, -0.0049, -0.0174, -0.0230,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0771, -0.1417, -0.0759, -0.0179, -0.0129,  0.0074,  0.0152, -0.0090,
        -0.0006, -0.0035, -0.0126, -0.1099, -0.0686,  0.0189,  0.0012, -0.0663,
         0.0152, -0.0419, -0.0018, -0.0298, -0.0726, -0.0276, -0.0216, -0.0012,
         0.0128, -0.0043,  0.0086, -0.0096, -0.0261,  0.0880,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([0.1969, 0.2582, 0.0601, 0.2295, 0.1173, 0.0451, 0.0590, 0.0338, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0767, -0.0803, -0.0089, -0.0454, -0.0273, -0.1133, -0.0053,  0.0078,
         0.0021, -0.0110, -0.0252, -0.0278, -0.0393, -0.0876, -0.0013, -0.0011,
        -0.0153, -0.0012, -0.0120, -0.0062,  0.0007, -0.0086, -0.0080, -0.0299,
        -0.0594, -0.0166, -0.0456,  0.0425, -0.0666, -0.0673,  0.0105,  0.0055,
         0.0436,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.4346e-03, -1.1755e-01, -6.1484e-02,  8.7803e-03, -5.0748e-03,
         1.4341e-02,  1.9920e-02, -2.4551e-02,  6.2594e-04, -4.2366e-03,
        -4.5176e-04, -4.8556e-03, -1.2989e-02, -5.2622e-03,  1.0648e-02,
         2.6666e-03,  2.3984e-03,  1.1788e-02, -2.0235e-03,  3.3409e-03,
         5.5938e-03, -4.9467e-03,  2.6891e-04, -1.1327e-02,  1.3247e-02,
        -1.6022e-02, -1.8142e-02, -3.5999e-02,  1.0552e-02, -2.4171e-03,
        -8.9086e-03,  3.3613e-03, -1.6899e-02, -8.9066e-02, -1.0491e-03,
        -1.2900e-02, -1.1010e-02,  4.3119e-04, -1.1111e-02, -1.0789e-02,
        -9.1595e-03, -4.5922e-03, -1.5005e-03, -2.5125e-02, -8.8281e-03,
         1.2179e-02, -3.5161e-02,  3.0214e-03, -3.2854e-02, -3.7474e-02,
        -7.8521e-03, -1.0593e-02, -4.7706e-03, -5.6351e-03,  5.8172e-03,
         1.4521e-02, -5.2247e-04, -6.1755e-03, -9.8121e-03, -1.6923e-02,
        -1.2782e-02, -8.3735e-03, -6.3164e-02, -1.1216e-02,  3.1582e-03,
        -2.2004e-03, -2.1087e-03,  1.5279e-02,  8.0431e-05, -1.0545e-02,
        -1.8121e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0523, -0.0345, -0.0126,  0.0099, -0.0299, -0.0010, -0.1555, -0.0202,
        -0.0177,  0.0022, -0.0103, -0.0209,  0.0079, -0.0143, -0.0077, -0.0015,
        -0.0111,  0.0191, -0.0093, -0.0340, -0.0126, -0.0222,  0.0181, -0.0512,
         0.0240, -0.0314, -0.0326, -0.0152, -0.0037, -0.0160, -0.0253,  0.0021,
        -0.0363, -0.0440, -0.0096, -0.0046, -0.0123,  0.0111, -0.0125,  0.0284,
         0.0958,  0.0190,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0402, -0.1242, -0.0023, -0.0332, -0.0026, -0.0058, -0.0066,  0.0029,
        -0.0126, -0.0344, -0.0332, -0.0206, -0.0470, -0.0064, -0.0108,  0.0214,
         0.0192, -0.0464, -0.0509, -0.0294, -0.0177, -0.0037,  0.0113, -0.0016,
        -0.0292, -0.0497, -0.0235, -0.0115, -0.0671, -0.0275, -0.0022, -0.0304,
         0.0090, -0.0108, -0.0259, -0.0730,  0.0559,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0491,  0.3749,  0.0172,  0.1559,  0.0502,  0.0105,  0.0295,  0.0062,
        -0.0249,  0.0566,  0.0560,  0.0289,  0.0626,  0.0632, -0.0141,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1538,  0.0179,  0.0165,  0.0223,  0.0337,  0.0628,  0.0200,  0.0420,
         0.1402,  0.0218,  0.0019,  0.0607,  0.1568, -0.0461, -0.0465, -0.0164,
         0.0189,  0.0024,  0.0303, -0.0888,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.8073e-02,  9.4453e-04, -1.3635e-03, -2.2071e-02, -1.0078e-01,
         7.9489e-03, -3.1064e-02,  1.1735e-02,  1.6833e-02, -3.9713e-02,
        -2.2511e-02, -4.3475e-02, -6.2114e-03,  6.9760e-05, -2.0173e-02,
        -3.5935e-02,  2.0317e-02, -3.1874e-02,  1.6721e-02, -4.2456e-02,
         5.4897e-03,  1.0431e-04, -1.9875e-02, -1.7898e-02, -2.6296e-02,
        -2.8664e-02, -9.0761e-02, -2.2264e-03, -1.0048e-03, -2.8444e-02,
        -4.9540e-02, -5.9832e-03, -8.3461e-03, -3.5033e-02, -3.2358e-02,
        -4.9235e-02,  3.9498e-02, -4.9269e-03, -4.5301e-03,  1.3530e-02,
        -2.9809e-03,  6.4312e-03,  1.6570e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0343, -0.0120, -0.0023, -0.0113, -0.0283, -0.0396, -0.0208, -0.1240,
        -0.0355, -0.0222, -0.0288, -0.0056, -0.0640, -0.0413, -0.0567, -0.0225,
        -0.0019,  0.0145, -0.0113, -0.0024, -0.0477, -0.0141, -0.0144,  0.0237,
        -0.0033, -0.0420, -0.0422, -0.0010, -0.0417, -0.0322,  0.0026, -0.0390,
        -0.0135, -0.0292, -0.0322, -0.0203, -0.0118, -0.0099,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0361, -0.1645, -0.0771, -0.0673,  0.0125, -0.1257, -0.0375,  0.0072,
         0.0012, -0.0782, -0.0029, -0.0084, -0.0513, -0.0053, -0.0186, -0.0885,
        -0.0050, -0.0437,  0.0356,  0.0032, -0.1094, -0.0208,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0038, -0.0423,  0.0312,  0.0159,  0.0037, -0.0143, -0.0053,  0.0323,
         0.0197, -0.0736, -0.0007,  0.0010, -0.0089,  0.0081, -0.0127, -0.0232,
        -0.1420,  0.0004, -0.0125, -0.0660,  0.0052, -0.0445, -0.0212, -0.0201,
        -0.0113, -0.0094, -0.0128,  0.0260,  0.0121, -0.0228, -0.0584,  0.0115,
         0.0023,  0.0104, -0.0091, -0.1200,  0.0270,  0.0077, -0.0098, -0.0408,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0424, -0.0466,  0.0036, -0.0086, -0.0247, -0.0323,  0.0055,  0.0149,
        -0.0443, -0.0250, -0.0022, -0.0024, -0.0117, -0.1364, -0.0640, -0.0215,
        -0.0317, -0.0183, -0.0443, -0.0129, -0.0080, -0.0055, -0.0077, -0.0256,
        -0.0355, -0.1042, -0.0047, -0.0551, -0.0778, -0.0031,  0.0126,  0.0669,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.0394e-02, -3.0530e-01, -3.5414e-02,  2.6138e-02, -3.1232e-02,
         1.1827e-02, -1.9235e-02, -4.0230e-02, -1.0082e-01, -8.5287e-03,
        -1.1078e-02, -2.8268e-02, -9.7162e-05,  2.9082e-03,  2.5388e-03,
        -1.4385e-03, -5.6926e-02, -1.2666e-02, -1.1320e-02, -7.3235e-03,
        -2.4112e-03, -7.5255e-03,  1.3418e-02, -2.8773e-02,  5.3482e-03,
         4.3487e-02, -1.0516e-02, -1.1720e-02,  2.5476e-02,  3.7843e-03,
         1.2635e-02, -2.0651e-04, -6.3480e-03,  2.6108e-02, -1.8559e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0423,  0.0435,  0.1187,  0.1182,  0.0240,  0.0124,  0.1439,  0.0724,
         0.0972,  0.0541,  0.0318,  0.0431,  0.0050,  0.0551,  0.0305,  0.0206,
         0.0343,  0.0170, -0.0042,  0.0092,  0.0225,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1304, -0.2626, -0.0157, -0.0271, -0.0464, -0.0378, -0.0543, -0.0763,
        -0.1510, -0.0048,  0.0059,  0.0175,  0.0139,  0.0362, -0.0401, -0.0018,
        -0.0208, -0.0056,  0.0521,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1383, -0.1394, -0.0766, -0.0417, -0.0358, -0.0106,  0.0077,  0.0153,
        -0.0225, -0.0870, -0.0107,  0.0532, -0.0194, -0.1750,  0.0288, -0.0268,
         0.0348, -0.0101,  0.0099,  0.0564,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1106, -0.2104, -0.0462, -0.0878, -0.0229, -0.0076, -0.0481, -0.0527,
        -0.0022, -0.0221, -0.0137, -0.0220, -0.0026, -0.0411, -0.0054, -0.0456,
        -0.0858,  0.0071, -0.0226, -0.0376, -0.0163,  0.0644, -0.0250,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0330,  0.2197,  0.0030, -0.0245, -0.0057,  0.0369,  0.0231,  0.0358,
         0.0197,  0.1216,  0.0430,  0.0468,  0.0370,  0.1039,  0.0112, -0.0059,
        -0.0421, -0.0083,  0.0167,  0.0162,  0.0054, -0.0119,  0.0458,  0.0468,
        -0.0359,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.4599e-02,  2.7075e-01,  1.2777e-02,  1.0046e-01,  2.0168e-02,
         3.8612e-02, -3.1331e-02,  2.1465e-02, -4.4618e-02, -6.6512e-03,
         3.8520e-02, -5.9853e-03, -7.0143e-04,  3.1304e-03,  2.8037e-04,
         1.1628e-02,  2.0605e-02,  6.7488e-03, -9.7290e-03,  1.8696e-02,
         1.7254e-02,  2.0065e-02,  2.7538e-02,  3.5174e-02, -6.1857e-03,
         3.9194e-02,  4.1703e-02, -6.0834e-05, -6.8523e-03, -2.0986e-02,
        -1.7829e-02,  4.7166e-03, -4.5004e-03,  2.6814e-02,  2.3672e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1390,  0.0259, -0.0104, -0.0709, -0.0123,  0.0002, -0.0562, -0.0428,
        -0.1307, -0.0351,  0.0168,  0.0153,  0.0150, -0.0491, -0.0393, -0.0602,
        -0.0691, -0.1046, -0.0311,  0.0477, -0.0154,  0.0128,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0506,  0.0816, -0.0340,  0.0338, -0.0067,  0.1084,  0.0004, -0.0150,
         0.0266,  0.1344,  0.0930, -0.0308,  0.0482,  0.0098,  0.0528,  0.0564,
         0.0674, -0.0126,  0.0193, -0.0943, -0.0238,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.3612e-02, -3.5988e-01, -6.9163e-03, -3.0437e-02,  2.8204e-02,
         1.9692e-02, -4.5045e-02, -1.6111e-01, -2.2858e-02,  2.5395e-02,
        -2.5724e-02,  8.3738e-03, -2.1117e-03,  1.1873e-02, -2.2142e-02,
        -4.2964e-02,  1.1232e-01, -3.8783e-05, -4.1302e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0431, -0.0363, -0.0265, -0.0001, -0.0198, -0.0545,  0.0243,  0.0016,
         0.0253,  0.0002, -0.0106, -0.0117, -0.0101, -0.0348, -0.0024,  0.0159,
        -0.0214, -0.0531, -0.0039, -0.0265, -0.0004, -0.0125, -0.0332, -0.0064,
        -0.0133, -0.0151,  0.0169, -0.0443, -0.0202, -0.0844,  0.0246, -0.0131,
        -0.0364, -0.0822, -0.0142, -0.0028,  0.0179,  0.0004,  0.0048, -0.0073,
         0.0203,  0.0035,  0.0089, -0.0035,  0.0234,  0.0120,  0.0556],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0251, -0.0846, -0.0559, -0.0077,  0.0124, -0.0059, -0.0051,  0.0036,
        -0.0165,  0.0126, -0.0019,  0.0221, -0.0067, -0.0141, -0.0051, -0.0055,
        -0.1188,  0.0092, -0.0273, -0.0013, -0.0322, -0.0441, -0.0572, -0.1736,
        -0.0311,  0.0376, -0.0765, -0.0371,  0.0155,  0.0118,  0.0012,  0.0243,
        -0.0084,  0.0079,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.2272e-02, -4.1294e-01,  4.6816e-02,  2.5066e-03, -2.3453e-03,
        -2.3236e-02, -8.3579e-03, -1.5855e-02, -3.8104e-03, -5.9604e-03,
         2.8691e-02, -5.7343e-03, -2.7861e-02, -9.3866e-02,  2.2866e-02,
         2.8689e-02, -7.4950e-03,  3.9163e-03, -1.4184e-03, -3.8928e-02,
        -1.0499e-02, -2.2434e-03, -9.9936e-03, -2.4523e-03,  3.0244e-03,
        -1.0157e-02,  8.8060e-05, -6.4977e-03, -3.1944e-03, -7.4024e-03,
        -2.0257e-02, -1.3335e-03,  4.6022e-03, -9.4690e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0400,  0.2607, -0.1246, -0.0216,  0.0241,  0.0260, -0.0129, -0.0367,
         0.0316,  0.0279,  0.0341, -0.0075,  0.0681,  0.1239,  0.0709,  0.0341,
        -0.0066,  0.0488,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1364, -0.0077,  0.0238, -0.0490,  0.0201, -0.0154,  0.0134, -0.0056,
         0.0215,  0.1995,  0.0134,  0.0080,  0.0279,  0.0225,  0.0090,  0.0077,
         0.0475, -0.0189,  0.0293,  0.1142,  0.0096, -0.0352, -0.0028, -0.0181,
        -0.0003, -0.0458, -0.0975,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.0353e-01, -6.1421e-02, -2.1565e-02,  2.2390e-02,  1.9612e-02,
         3.4911e-02, -2.8816e-02, -2.3338e-02,  1.3495e-01, -8.3484e-03,
         1.8710e-02,  3.5764e-02,  1.6085e-01,  1.0758e-02,  5.0286e-03,
        -1.8913e-02,  1.7131e-02, -2.1528e-05, -1.5166e-02,  1.9435e-01,
         2.2120e-02,  4.3428e-04, -6.4289e-03,  2.7945e-04,  2.2388e-02,
         1.2773e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1511,  0.1694,  0.0036,  0.0447,  0.1850,  0.0424, -0.0288,  0.0169,
         0.0361,  0.0525,  0.0672,  0.0150,  0.0099, -0.0130,  0.0334,  0.0181,
         0.0094,  0.0162, -0.0191, -0.0428,  0.0255,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0323,  0.2872,  0.1002,  0.1963, -0.0219,  0.0193,  0.0870, -0.0027,
         0.0192, -0.0743,  0.0171,  0.0426,  0.0139, -0.0254,  0.0605,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1763,  0.0129,  0.0541, -0.0013,  0.0056,  0.0823,  0.0448,  0.0067,
         0.0115, -0.0253, -0.0059,  0.0526,  0.2256,  0.0763,  0.0788, -0.0304,
        -0.0706, -0.0389,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0072, -0.2128, -0.0790, -0.2039, -0.0195, -0.0096, -0.0260,  0.0193,
        -0.0563, -0.0573, -0.0584, -0.0068,  0.0216, -0.0419, -0.0186,  0.0009,
         0.0155,  0.0554,  0.0899,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0551, -0.0895,  0.0265, -0.0005,  0.0127, -0.0145,  0.0130, -0.0007,
        -0.0205, -0.0051, -0.0124,  0.0020, -0.0087,  0.0057, -0.0117, -0.0303,
        -0.0715, -0.0079,  0.0365,  0.0248, -0.0935, -0.0092, -0.0554, -0.0281,
        -0.0354, -0.0301, -0.0331, -0.0025, -0.0064, -0.0385, -0.0028, -0.0031,
        -0.0338, -0.0086, -0.0012, -0.0067, -0.0458, -0.0043, -0.0178,  0.0103,
         0.0184,  0.0178, -0.0267,  0.0208], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.3895e-01, -1.6431e-01, -4.6845e-02, -5.4235e-02, -2.1534e-02,
        -1.6029e-02, -5.0029e-03, -5.0989e-03, -3.6692e-02,  2.0477e-02,
        -2.7888e-02,  1.2627e-02,  2.4911e-02, -4.1347e-03,  3.5421e-02,
        -1.9602e-04, -5.3234e-03, -3.6742e-02,  1.0517e-03, -2.1056e-02,
        -4.8517e-02,  6.8521e-03,  1.0037e-02,  5.2891e-03,  1.1466e-02,
        -1.3287e-03,  1.1680e-04, -6.0964e-02,  2.3731e-02, -7.1995e-03,
        -1.2997e-02, -3.6741e-02,  8.3131e-03, -7.6958e-02,  1.0960e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0903, -0.2298, -0.0029, -0.0395, -0.0191, -0.0014, -0.0746, -0.0795,
        -0.0084,  0.0035,  0.0022, -0.0146, -0.1415,  0.0005,  0.0003, -0.0972,
        -0.0584,  0.0068, -0.0076,  0.0262,  0.0102, -0.0122,  0.0330,  0.0402,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0899, -0.2490, -0.0330,  0.0286,  0.0046, -0.0184,  0.0032,  0.0157,
        -0.0482,  0.0044,  0.0165,  0.0039,  0.0027, -0.0317, -0.0583,  0.0227,
         0.0107,  0.0158, -0.0516, -0.0032, -0.0110, -0.0265, -0.0060, -0.0181,
         0.0074, -0.0541, -0.0094, -0.0226, -0.0040,  0.0237, -0.0267, -0.0202,
        -0.0167,  0.0417,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0205, -0.0061,  0.0070, -0.0500,  0.0322,  0.0313, -0.1093, -0.0233,
        -0.0150, -0.0246, -0.0142, -0.0028, -0.0660, -0.0422, -0.0743, -0.2010,
        -0.0211,  0.0319, -0.0195, -0.0061,  0.0146, -0.0074, -0.0034, -0.0594,
        -0.0014, -0.0155, -0.0273, -0.0236, -0.0098, -0.0390,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.2833, -0.0319, -0.0605, -0.2797, -0.0140,  0.0111, -0.0173, -0.0342,
        -0.0593, -0.1417, -0.0006,  0.0048, -0.0616,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1062, -0.0910, -0.0160,  0.0024, -0.0244, -0.0697,  0.0492, -0.0249,
        -0.2581, -0.0350, -0.0385, -0.0575, -0.0825, -0.0155, -0.0942, -0.0098,
         0.0164,  0.0088,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1278,  0.2650,  0.0703, -0.0560,  0.1847, -0.0057,  0.0872,  0.2033,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0366, -0.1650, -0.0176,  0.0122, -0.0277, -0.0188, -0.0075,  0.0020,
         0.0037, -0.0043,  0.0062,  0.0040, -0.0030, -0.0072, -0.0009, -0.0037,
         0.0051,  0.0013, -0.0112,  0.0062, -0.0154, -0.0082, -0.0288, -0.0032,
        -0.0030, -0.0230, -0.0240, -0.0147, -0.1218, -0.0247,  0.0194, -0.0630,
        -0.0163, -0.0897, -0.0063, -0.0533, -0.0345, -0.0165,  0.0187,  0.0064,
        -0.0050,  0.0084, -0.0101,  0.0035,  0.0319,  0.0025, -0.0036],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0470,  0.0965,  0.2127,  0.0681,  0.0690,  0.0451,  0.0311,  0.0768,
         0.2144,  0.0013, -0.1381,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0561,  0.1808,  0.0539,  0.0784,  0.0195,  0.0058,  0.0268,  0.1982,
         0.0708,  0.0444,  0.0541,  0.0185,  0.0808,  0.0241,  0.0056, -0.0285,
        -0.0272,  0.0265,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0835, -0.0102,  0.2191, -0.0619, -0.0128, -0.0264,  0.0659,  0.0451,
         0.0962,  0.0534, -0.0492, -0.0230,  0.1093,  0.0247,  0.1194,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1695, -0.3305, -0.0392, -0.0235,  0.0461, -0.0418, -0.1078,  0.0396,
         0.0087, -0.0465, -0.0750, -0.0088, -0.0080,  0.0542,  0.0009,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1403, -0.0470,  0.0047,  0.0283,  0.0180,  0.0131,  0.0181, -0.0160,
        -0.0884, -0.0602, -0.0489, -0.0747, -0.0837,  0.0023,  0.0197, -0.0125,
        -0.0651,  0.0170,  0.0084,  0.0093, -0.0186, -0.0886,  0.0037,  0.0325,
        -0.0117, -0.0691,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.9640e-02,  1.4033e-03,  7.7311e-03, -6.3742e-02,  4.8251e-02,
         7.0688e-02, -1.1590e-02,  8.1396e-02,  4.2140e-01,  3.6816e-02,
         1.0093e-04,  3.7805e-03,  1.3973e-02, -1.3005e-01, -9.4384e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.4241e-02, -3.0278e-01, -5.5656e-03, -1.7605e-03, -7.3946e-02,
         1.2124e-03, -3.6613e-02, -1.3634e-03, -3.0057e-04, -1.2673e-01,
         2.0958e-02,  9.8236e-03, -1.4227e-02, -1.2008e-02, -7.2287e-03,
         8.7212e-03, -1.0797e-02,  2.3556e-02, -1.4486e-02,  1.5791e-02,
        -1.1835e-02, -3.1642e-02,  8.9162e-03,  3.0818e-02,  1.9735e-02,
         1.8495e-01,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0328,  0.0139,  0.0404, -0.0028, -0.0372, -0.0240, -0.0095, -0.0242,
        -0.0188, -0.0460, -0.1797, -0.0324, -0.0247, -0.1072, -0.0660, -0.0203,
         0.0176, -0.0347,  0.0118, -0.0280, -0.0314,  0.0022, -0.0075,  0.0040,
         0.1110, -0.0719,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0832, -0.2703, -0.0549, -0.0096, -0.0250,  0.0207, -0.0250, -0.0448,
         0.0148, -0.0212, -0.0155, -0.0110, -0.0460, -0.0221, -0.0165, -0.0365,
        -0.0134, -0.0044, -0.0188, -0.0088, -0.0086, -0.0184, -0.0076, -0.0225,
        -0.0456, -0.0386, -0.0222, -0.0035,  0.0114,  0.0018, -0.0092, -0.0100,
        -0.0047,  0.0064, -0.0088, -0.0181,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0278, -0.1299, -0.0175, -0.0358, -0.0241, -0.0198, -0.0260, -0.0634,
         0.0002,  0.0463, -0.0142,  0.0077,  0.0242,  0.0530, -0.0115, -0.0101,
         0.0093,  0.0011,  0.0119,  0.0094, -0.0085, -0.0274, -0.0126, -0.0241,
        -0.0243, -0.0615,  0.0264, -0.0114, -0.0605, -0.0120,  0.0266, -0.0184,
        -0.0471,  0.0259, -0.0173, -0.0527,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2740,  0.1513, -0.0344,  0.0451, -0.0037,  0.0842,  0.0305,  0.0182,
        -0.0018,  0.0257,  0.0560,  0.0243,  0.0657,  0.0362,  0.0260,  0.0251,
         0.0148,  0.0066, -0.0145, -0.0619,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0413, -0.1410, -0.0808, -0.0055,  0.0050, -0.0139,  0.0076, -0.0198,
        -0.0111, -0.0233,  0.0140, -0.0426,  0.0109, -0.0412, -0.0775, -0.0215,
        -0.0076, -0.0483, -0.0244, -0.0089, -0.0219, -0.0118,  0.0009, -0.0024,
        -0.0049, -0.0094,  0.0060, -0.0604,  0.0104, -0.0332,  0.0006, -0.0385,
        -0.0215, -0.0570,  0.0067,  0.0191,  0.0011,  0.0480,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0124, -0.0667, -0.0212, -0.1121, -0.0034, -0.0011,  0.0115, -0.0021,
         0.0091, -0.0060, -0.0492, -0.0451, -0.0025,  0.0159,  0.0463, -0.0716,
        -0.0180, -0.0627,  0.0127,  0.0156,  0.0166,  0.0053,  0.0168, -0.0030,
        -0.0365,  0.0172, -0.0110, -0.0223, -0.0616,  0.0045, -0.0214, -0.0083,
         0.0129,  0.0078, -0.0196,  0.0083,  0.0147,  0.0210,  0.0154,  0.0024,
        -0.0883], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0055, -0.4530, -0.0372,  0.0899, -0.0108,  0.0080, -0.0220, -0.0556,
        -0.0276,  0.0048, -0.0128, -0.0297, -0.0356, -0.0216, -0.0205, -0.0369,
         0.0013, -0.0375, -0.0116,  0.0297, -0.0173, -0.0310,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0676, -0.0011,  0.0037, -0.0080, -0.0247, -0.0161,  0.0091, -0.0010,
        -0.0182,  0.0100, -0.0107, -0.0563, -0.0095, -0.0304, -0.0240, -0.0372,
         0.0038, -0.0169, -0.0007, -0.0572, -0.1037,  0.0077,  0.0383, -0.0275,
        -0.0161,  0.0107,  0.0053, -0.0271, -0.1235, -0.0055, -0.0055, -0.0390,
        -0.0545,  0.0660,  0.0633,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1134, -0.0655, -0.1180, -0.0488, -0.1139,  0.0168,  0.0060,  0.0181,
        -0.0212,  0.0129, -0.0186, -0.0050, -0.0088, -0.0154,  0.0190, -0.0399,
         0.0077, -0.0254, -0.0871, -0.0261, -0.0187, -0.0196, -0.0094, -0.0238,
        -0.0065,  0.0100, -0.0469, -0.0240,  0.0535,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0619,  0.2690,  0.0697,  0.0531,  0.0352,  0.0685,  0.0279, -0.0152,
        -0.0062, -0.0245,  0.0098,  0.0309, -0.0159,  0.0030, -0.0099,  0.0265,
         0.0127,  0.0103,  0.0248, -0.0085,  0.0360,  0.0317,  0.0231, -0.0112,
        -0.1145,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1760,  0.0189, -0.0412,  0.0034, -0.2463, -0.0263, -0.0068, -0.0497,
        -0.0381, -0.0374,  0.0175, -0.0237, -0.0072,  0.0412, -0.0858, -0.0489,
        -0.0084, -0.0101, -0.0311, -0.0088, -0.0266,  0.0443, -0.0024,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0380, -0.0211, -0.0030, -0.0124,  0.0066,  0.0167, -0.0225, -0.0168,
        -0.0121, -0.0190, -0.0022, -0.0215, -0.0243, -0.2265,  0.0170, -0.0128,
         0.0013, -0.0143,  0.0128, -0.0027,  0.0159, -0.0704,  0.0186,  0.0052,
        -0.0151, -0.0283, -0.0196, -0.0356, -0.0777, -0.0261, -0.0178, -0.0130,
        -0.0499, -0.0317,  0.0098, -0.0026, -0.0172,  0.0171, -0.0245,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1272,  0.0859, -0.0590,  0.1498,  0.2795, -0.0068,  0.0225,  0.0808,
         0.1524, -0.0278, -0.0071, -0.0014,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.0511,  0.0598,  0.0848,  0.0069,  0.0031, -0.0003,  0.0367, -0.0316,
         0.0004,  0.0014, -0.0056, -0.0011, -0.0033, -0.0153,  0.0141,  0.0069,
         0.0075,  0.0268, -0.0032,  0.0308,  0.0159,  0.0167, -0.0087,  0.0108,
        -0.0277,  0.0176,  0.0626,  0.0848, -0.0082,  0.0330, -0.0066,  0.0135,
         0.0388, -0.0011,  0.0225,  0.0284,  0.0282,  0.0138, -0.0015,  0.0100,
        -0.0202, -0.0197, -0.0479, -0.0351,  0.0363,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0444,  0.4032,  0.0881,  0.1193,  0.0394,  0.0182, -0.0154,  0.0859,
         0.0931, -0.0697,  0.0098,  0.0135,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0184,  0.6912,  0.0438,  0.0148, -0.0433,  0.0605,  0.0894, -0.0385,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0383, -0.1648, -0.0361,  0.0087, -0.0093, -0.0066, -0.0129, -0.0005,
        -0.0089, -0.0168, -0.0250, -0.0024, -0.0032,  0.0108, -0.0020,  0.0068,
         0.0053, -0.0350,  0.0044, -0.0023, -0.0022,  0.0081,  0.0040, -0.0127,
         0.0006, -0.0061,  0.0111, -0.0167, -0.0137,  0.0225, -0.0060, -0.0104,
        -0.0021, -0.0038, -0.0059,  0.0038, -0.0028, -0.0127,  0.0021,  0.0017,
        -0.0109, -0.0023,  0.0048, -0.0085, -0.0234,  0.0710, -0.0553,  0.0127,
        -0.0261, -0.0578, -0.0315,  0.0153, -0.0145,  0.0214, -0.0179,  0.0066,
        -0.0149,  0.0034,  0.0029,  0.0438, -0.0058], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0021, -0.0404,  0.0115, -0.0238, -0.0785,  0.0066, -0.0288,  0.0077,
        -0.0012, -0.0124, -0.0057, -0.0477,  0.0007, -0.0449, -0.1096, -0.0250,
         0.0261,  0.0138, -0.0716, -0.0096, -0.0805, -0.0290, -0.0207,  0.0170,
        -0.0513, -0.0220,  0.0027, -0.0102, -0.0206, -0.0550, -0.0357, -0.0754,
         0.0124,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0318, -0.0686, -0.0968, -0.0382, -0.0019, -0.0348, -0.0235,  0.0194,
         0.0104, -0.0490,  0.0350, -0.0040, -0.0181, -0.0177, -0.0159,  0.1270,
        -0.0119, -0.0714, -0.1967, -0.0073,  0.0117,  0.0067,  0.0677,  0.0344,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1851, -0.0621, -0.0661, -0.0132, -0.0348, -0.0466, -0.0115, -0.0428,
         0.0112, -0.0505, -0.0528, -0.0607, -0.1182, -0.0261, -0.0098, -0.0247,
         0.0043, -0.0163,  0.0166,  0.0188, -0.0265,  0.0064, -0.0398,  0.0552,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0940, -0.0489, -0.0631, -0.0631,  0.0042, -0.0260,  0.0520, -0.0403,
         0.0048,  0.0357, -0.0021, -0.0062, -0.0023, -0.0031, -0.0218, -0.0648,
        -0.0427, -0.0199, -0.0077,  0.0053, -0.0694, -0.0120, -0.0275, -0.0858,
        -0.0123, -0.0017, -0.0042, -0.0155,  0.0202, -0.0149, -0.0165, -0.0082,
        -0.0483, -0.0556,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0736, -0.0441, -0.1103,  0.0033, -0.0111,  0.0527, -0.0118, -0.0150,
        -0.0037, -0.0323,  0.0008, -0.0734, -0.0006, -0.1194, -0.0609,  0.0793,
        -0.0617, -0.0204, -0.0057,  0.0183, -0.0239, -0.0758,  0.0338, -0.0600,
         0.0081,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0665, -0.0471,  0.0468,  0.0353,  0.0448,  0.0183,  0.0702,  0.0315,
         0.0717,  0.3975,  0.0112,  0.0229,  0.0465, -0.0009, -0.0183,  0.0705,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0282,  0.1625,  0.0119,  0.0373,  0.0497,  0.0588,  0.2742,  0.0201,
        -0.0076, -0.0335, -0.0003,  0.0060,  0.0309, -0.0045,  0.0794, -0.0583,
         0.0061,  0.0189,  0.0325, -0.0089, -0.0027, -0.0245, -0.0431,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0382,  0.1278,  0.0972,  0.1211,  0.1911,  0.0433,  0.0393,  0.0101,
         0.1001,  0.0374,  0.0526, -0.0130, -0.0556,  0.0733,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.0910,  0.0722,  0.0599,  0.1392,  0.0197,  0.0075,  0.0494, -0.0328,
         0.0525,  0.0316,  0.0612,  0.0837,  0.0176,  0.0038,  0.0209,  0.0061,
        -0.0002,  0.1149,  0.0127,  0.0797,  0.0230,  0.0106,  0.0089, -0.0011,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.8677e-03, -4.9834e-02, -1.4943e-01, -1.0922e-02, -1.2070e-01,
        -1.4894e-01, -2.8241e-01, -9.8474e-02, -7.0647e-02, -6.6596e-02,
         1.7676e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0438, -0.0751,  0.0048, -0.0515, -0.0461, -0.0804, -0.1706, -0.0323,
        -0.1086, -0.0179, -0.0201, -0.0202, -0.0588, -0.0351, -0.0264,  0.0089,
        -0.0135, -0.0778,  0.0412,  0.0327, -0.0343,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0388, -0.0702, -0.0012, -0.0616,  0.0280, -0.0300,  0.0349, -0.1294,
        -0.1460, -0.1907, -0.0598,  0.0430, -0.0348, -0.1317,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0660, -0.0286, -0.0003, -0.0347, -0.0411, -0.0559,  0.0156,  0.0746,
         0.0466, -0.0160, -0.0915, -0.0831, -0.2517, -0.0055, -0.0128,  0.0225,
        -0.0239,  0.0084, -0.0393,  0.0765,  0.0052,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0281, -0.3349, -0.0563, -0.0303, -0.0341, -0.1238, -0.0450, -0.1813,
         0.0242, -0.0503,  0.0041, -0.0426,  0.0450,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0098, -0.0283, -0.0104, -0.0038,  0.0142,  0.0099, -0.0175, -0.0546,
         0.0168, -0.0024, -0.0110, -0.0099, -0.0222,  0.0038, -0.0087,  0.0229,
         0.0192,  0.0125, -0.0045, -0.0228, -0.0056, -0.0051,  0.0062, -0.0100,
        -0.2053,  0.0031, -0.0728, -0.0697, -0.0390,  0.0244, -0.0268, -0.0186,
        -0.0108, -0.0005, -0.0056, -0.0019, -0.0264, -0.0042,  0.0016, -0.0134,
        -0.0157, -0.0063,  0.0148, -0.0692, -0.0381], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0078, -0.1340, -0.0065,  0.0060,  0.0003, -0.0109, -0.0119, -0.0096,
         0.0033, -0.0087, -0.0473, -0.0222,  0.0202,  0.0274, -0.0378, -0.0076,
        -0.1614,  0.0023, -0.0159,  0.0223, -0.0015, -0.0756,  0.0213, -0.0802,
         0.0065, -0.0206, -0.0164, -0.0223, -0.0635, -0.0181,  0.0176, -0.0349,
         0.0320,  0.0262,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0493,  0.0003,  0.0382,  0.0757,  0.0144,  0.0092,  0.0081,  0.0283,
        -0.1122,  0.0270,  0.0189, -0.0092, -0.0038,  0.0148, -0.0086, -0.0696,
        -0.0265, -0.0341,  0.0194,  0.1423,  0.1685,  0.0049,  0.0025, -0.0584,
         0.0501,  0.0058,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0256, -0.2449, -0.0331, -0.0294, -0.0172, -0.0703,  0.0111, -0.0616,
        -0.0334,  0.0059, -0.0085,  0.0004,  0.0044, -0.0173,  0.0340,  0.0029,
        -0.2423, -0.0085, -0.0243,  0.0208, -0.0015,  0.0615, -0.0412,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0316, -0.0593,  0.0022, -0.0021,  0.0072,  0.0149, -0.0042, -0.0198,
        -0.0315, -0.0096, -0.0002, -0.0002, -0.0145, -0.0307, -0.1984,  0.0669,
         0.0270, -0.0204, -0.0136, -0.0511,  0.0044,  0.0119,  0.0086,  0.0309,
        -0.0063,  0.0167, -0.0432, -0.0632, -0.0044,  0.0085,  0.0290,  0.0371,
         0.0030,  0.0285,  0.0191, -0.0115,  0.0039,  0.0101,  0.0414, -0.0128,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0279, -0.0728, -0.0207, -0.0315, -0.0600, -0.0141,  0.0049, -0.0294,
        -0.0315, -0.0248, -0.0223, -0.0153, -0.0245, -0.0236, -0.0429, -0.0378,
        -0.0266, -0.0120,  0.0085, -0.0150,  0.0194, -0.0382, -0.0283,  0.0012,
        -0.0014,  0.0008,  0.0147, -0.0528, -0.0010, -0.0212, -0.0864,  0.0217,
        -0.0686, -0.0982,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 2.9808e-02, -2.8900e-02, -1.2578e-02,  1.6652e-03, -1.7758e-02,
         2.6850e-02,  1.4393e-02, -1.3251e-02,  5.0615e-03,  3.8470e-03,
        -2.1035e-02, -8.4596e-03, -2.7707e-02, -1.2848e-02, -1.4860e-01,
        -2.4865e-02, -2.9060e-02,  2.3174e-03, -6.4760e-02, -1.6340e-03,
        -6.0421e-03, -1.9823e-04, -2.5470e-03,  3.6724e-03, -5.0783e-02,
        -9.1849e-02, -3.8643e-03, -1.5569e-02,  1.3310e-04, -3.8761e-02,
        -1.1413e-02,  3.5708e-03,  1.3823e-02,  1.1844e-02, -4.1537e-02,
        -1.1143e-02, -7.1787e-03, -2.3921e-02, -9.3135e-03,  2.8305e-04,
        -8.1952e-03,  7.4272e-03, -9.0090e-03,  3.6361e-03,  4.5538e-05,
         1.0123e-02,  1.3966e-02,  5.8214e-03, -1.0655e-02, -9.6430e-03,
        -3.0830e-02, -4.7811e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1144, -0.0133, -0.0365, -0.0381, -0.1184, -0.0196,  0.0059,  0.0207,
        -0.0182, -0.0369, -0.0028, -0.0080, -0.0124, -0.1210, -0.0311, -0.0736,
        -0.0512, -0.0411, -0.0388, -0.0092, -0.0194, -0.0264, -0.0308, -0.0553,
         0.0139, -0.0115,  0.0078, -0.0005, -0.0037, -0.0114, -0.0081,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0482, -0.1626,  0.0066,  0.0189,  0.0130,  0.0170,  0.0145,  0.0212,
        -0.0278,  0.0191, -0.0032, -0.0381, -0.0140, -0.0115, -0.0288, -0.0684,
         0.0161, -0.0141, -0.1022, -0.1092, -0.0380, -0.0082, -0.0868,  0.0164,
         0.0961,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0148, -0.0619, -0.0105,  0.0112, -0.0901, -0.2390, -0.3532, -0.0043,
        -0.0617, -0.0683, -0.0766, -0.0083,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0668, -0.2220, -0.0211, -0.0366, -0.0732, -0.0230,  0.0115, -0.0097,
        -0.0076, -0.0470,  0.0572, -0.0029, -0.0312, -0.0906, -0.0080, -0.0370,
        -0.0039,  0.0127,  0.0268,  0.0055, -0.0178, -0.0042, -0.0299, -0.0176,
         0.0167, -0.0900,  0.0014,  0.0007,  0.0275,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0812,  0.2244, -0.0444,  0.0052,  0.0021,  0.1587,  0.1871, -0.0352,
        -0.0960, -0.0271,  0.0425,  0.0962,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1928, -0.1422, -0.0741, -0.0819, -0.0099, -0.0223, -0.3228, -0.0273,
        -0.0430, -0.0393,  0.0080,  0.0363,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.1900e-01,  5.0331e-02, -3.9369e-02, -2.5420e-02, -5.6996e-02,
        -3.1436e-03, -2.7223e-02, -6.2328e-02, -1.1482e-01, -2.8650e-02,
         4.5382e-03, -6.9941e-02, -1.3000e-02, -1.1154e-04, -1.5102e-02,
        -9.9450e-02,  9.9269e-03, -1.1920e-02,  2.9906e-02,  1.1882e-01,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0369,  0.2712,  0.0242,  0.0459,  0.0961,  0.0459,  0.0096, -0.0174,
         0.0151,  0.0081,  0.0145,  0.0373,  0.0194, -0.0058,  0.0185, -0.0090,
         0.0176, -0.0027,  0.0054,  0.0207,  0.0083, -0.0131,  0.0139,  0.0130,
        -0.0049, -0.0099, -0.0028,  0.0238,  0.0103, -0.0019, -0.0219,  0.0021,
         0.0104,  0.0332,  0.0359, -0.0185,  0.0547,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.3739,  0.4354, -0.0100, -0.0344, -0.0084, -0.0248,  0.0206, -0.0799,
         0.0125,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0292, -0.0195,  0.2248, -0.0577, -0.0188, -0.0557, -0.0490, -0.0504,
         0.0127,  0.0007, -0.0154, -0.0306, -0.0625, -0.0802, -0.0151,  0.0020,
        -0.0030,  0.0312, -0.0161, -0.0410,  0.0081,  0.0029, -0.0122,  0.0061,
         0.0281,  0.1051, -0.0219,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0487, -0.2796, -0.0355, -0.0347, -0.0422, -0.0290,  0.0035,  0.0072,
        -0.0027, -0.0353, -0.0163, -0.0259, -0.0061, -0.0017, -0.0113, -0.0119,
        -0.0553,  0.0114,  0.0288,  0.0032,  0.0601, -0.0206, -0.0195,  0.0087,
         0.0076, -0.0221, -0.0049, -0.0182, -0.0909, -0.0057, -0.0089, -0.0087,
        -0.0272,  0.0063,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0389,  0.0157, -0.0021,  0.0019, -0.0171,  0.0121, -0.0011, -0.0093,
         0.0087,  0.0328, -0.0394, -0.1103, -0.0298, -0.0041, -0.0153, -0.0412,
        -0.0879,  0.0204, -0.0067,  0.0049,  0.0206, -0.0087, -0.0383, -0.0183,
         0.0051,  0.0014,  0.0058, -0.0040,  0.0151,  0.0091, -0.0053, -0.0486,
        -0.0876,  0.0250, -0.0152, -0.0254, -0.0486, -0.0111,  0.0115, -0.0022,
        -0.0111, -0.0410,  0.0110,  0.0098,  0.0207,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0772,  0.0228,  0.0173,  0.0458,  0.0617,  0.1224,  0.3008,  0.0261,
         0.0173,  0.0976,  0.0334,  0.0347,  0.0454, -0.0271, -0.0107, -0.0596,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0815, -0.0297,  0.1100,  0.0260,  0.0077,  0.0143,  0.0037,  0.0747,
         0.1246,  0.1606,  0.0280,  0.0697, -0.0041,  0.0199,  0.0288, -0.0038,
         0.0218,  0.0176,  0.0049,  0.0266,  0.0114,  0.0157, -0.0040, -0.0236,
         0.0187,  0.0130,  0.0558,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0800, -0.0124, -0.0338,  0.0027, -0.0057,  0.0073, -0.0049,  0.0020,
        -0.0207, -0.0078,  0.0042, -0.0028, -0.0114, -0.0006, -0.0075, -0.0052,
        -0.0092, -0.0083,  0.0224, -0.0222,  0.0005, -0.0125, -0.0032, -0.1020,
         0.0075,  0.0008, -0.0332, -0.0642, -0.0012, -0.0564,  0.0254, -0.0470,
        -0.0559, -0.0071, -0.0189, -0.0079, -0.0116, -0.0607, -0.0035,  0.0239,
        -0.0264, -0.0582, -0.0041,  0.0064,  0.0735, -0.0169], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1129,  0.0309, -0.0244,  0.0382, -0.0109, -0.0097, -0.0428, -0.0540,
         0.0100, -0.0118, -0.0221, -0.0197, -0.0091,  0.0066, -0.0147, -0.0054,
        -0.0188, -0.0090, -0.0366, -0.1445, -0.0218, -0.0491, -0.0116, -0.0234,
         0.0087,  0.0131, -0.0831,  0.0221, -0.0881,  0.0469,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1116,  0.0858,  0.1160,  0.0478,  0.0409, -0.0151, -0.1397,  0.1791,
        -0.0595, -0.2008, -0.0037,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0224, -0.3025, -0.1446, -0.0387, -0.0369, -0.0221, -0.0246, -0.1042,
         0.0231, -0.0236,  0.0192,  0.0003, -0.0448, -0.0010, -0.0401, -0.0576,
        -0.0195, -0.0307, -0.0051, -0.0388,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1617,  0.3619,  0.1631,  0.1157,  0.0068,  0.0313,  0.0261, -0.0115,
         0.0293,  0.0880, -0.0048,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0957,  0.2216,  0.0487,  0.1340, -0.0389, -0.0288,  0.0300,  0.0662,
         0.1002, -0.0048,  0.0096,  0.0137,  0.0067,  0.0044,  0.0266,  0.0811,
        -0.0086,  0.0004, -0.0190,  0.0072, -0.0539,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.1017e-01,  1.1942e-01,  5.5561e-02,  3.6038e-03,  1.0866e-01,
         2.0728e-01, -1.8594e-01, -6.3769e-02,  1.7935e-02, -2.5434e-02,
         3.3411e-02, -8.3121e-03, -9.8794e-03,  1.0950e-04, -1.5136e-03,
         2.4157e-02,  4.1551e-03, -2.0684e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.0741e-03, -5.1255e-02, -1.7703e-01,  8.6325e-03,  8.1646e-03,
        -1.4933e-03,  1.6325e-02, -6.3321e-05, -3.0447e-03, -7.9580e-03,
         3.6494e-03,  1.0125e-02,  3.0986e-02, -3.5437e-02, -2.8644e-02,
         9.9133e-03, -2.4803e-02, -5.9650e-02, -1.8421e-02, -3.9147e-02,
        -2.1479e-02, -5.9286e-02, -7.1605e-02, -2.1674e-02, -3.7634e-02,
        -7.7743e-03, -1.3086e-02, -1.2020e-02, -8.7305e-03, -6.4815e-02,
         3.0853e-03, -4.2266e-03, -2.4210e-03,  8.2366e-03, -1.3400e-02,
         1.4003e-04, -1.2443e-02,  4.5724e-02, -5.0400e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0248,  0.2114,  0.0100,  0.0371,  0.0141,  0.0219,  0.0018, -0.0398,
        -0.0122, -0.0060, -0.0137,  0.0565,  0.0056,  0.0180, -0.0290,  0.0230,
         0.0353,  0.0103, -0.0005,  0.0238, -0.0058,  0.0253, -0.0005, -0.0072,
         0.0009,  0.0212,  0.0267,  0.0194,  0.0755,  0.0820,  0.0115,  0.0196,
         0.0843,  0.0252,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.2116, -0.1597, -0.0966, -0.1347,  0.0513, -0.1207, -0.0281,  0.0734,
        -0.1238,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0967, -0.1121, -0.0256, -0.0520, -0.0016, -0.0568, -0.0042, -0.0080,
         0.0188,  0.0138, -0.0248, -0.0059, -0.0090, -0.0007,  0.0091, -0.0641,
        -0.0053, -0.1570, -0.0252, -0.0232,  0.0103, -0.0021,  0.0183,  0.0011,
        -0.0215, -0.0128, -0.0168,  0.0003,  0.0039, -0.0119, -0.0018, -0.0313,
         0.0035, -0.0114,  0.0402,  0.0024,  0.0086, -0.0168,  0.0349, -0.0361,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.4393e-02, -6.8643e-02, -8.0523e-02, -1.8705e-02,  2.2634e-03,
        -8.2413e-03,  2.6620e-02, -1.7430e-03, -1.0885e-02, -4.4790e-02,
         2.1331e-03,  5.9307e-03, -6.1791e-03, -7.8624e-03, -3.2649e-02,
         2.6219e-05, -1.0936e-02, -1.6219e-02, -2.5130e-02,  4.5444e-02,
         1.6918e-02, -5.6740e-04,  5.1286e-03,  1.0842e-02, -1.3811e-02,
        -1.3684e-02, -3.6517e-02, -3.7459e-04, -4.0206e-02, -6.6339e-02,
         7.4835e-03,  6.9145e-03, -3.8385e-03, -8.2094e-03,  5.2672e-03,
         7.6472e-04, -2.3293e-02, -4.8525e-02,  1.0397e-02,  1.1818e-02,
         1.7605e-02,  2.0908e-02, -4.4666e-03, -1.5377e-02,  3.0393e-04,
         1.4629e-02,  5.7125e-04,  3.4314e-02,  1.2962e-02, -1.9494e-02,
         5.8970e-02,  1.8479e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0193,  0.2109,  0.0167,  0.0141, -0.0019,  0.0160,  0.0533,  0.0073,
         0.0019,  0.0034, -0.0615,  0.0861,  0.0945,  0.0168,  0.0491,  0.0682,
         0.0455, -0.0517,  0.0809,  0.0254,  0.0145,  0.0610,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0410,  0.3388, -0.0058,  0.0366, -0.0137,  0.0075, -0.0007,  0.0454,
         0.0467,  0.0395,  0.1091,  0.0124,  0.0316,  0.0296, -0.0137,  0.0492,
        -0.0515,  0.0563, -0.0017,  0.0093,  0.0597,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1034, -0.1338, -0.0070, -0.0162,  0.0119, -0.0066, -0.0043,  0.0250,
        -0.0011,  0.0093, -0.0026, -0.0010,  0.0112,  0.0066, -0.0004, -0.0049,
         0.0117,  0.0094,  0.0059, -0.0004,  0.0057,  0.0090,  0.0111,  0.0228,
        -0.0008,  0.0064, -0.0451, -0.0588, -0.0063, -0.0048, -0.0266, -0.0337,
        -0.0041, -0.0025, -0.0139,  0.0006,  0.0136,  0.0053,  0.0172, -0.0074,
        -0.1263,  0.0024,  0.0118, -0.0153, -0.0367, -0.0002, -0.0233, -0.0209,
         0.0172,  0.0186, -0.0066,  0.0010,  0.0051, -0.0191, -0.0272],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0723, -0.0184,  0.0195, -0.0253,  0.0096, -0.0072, -0.0653, -0.0288,
        -0.0221,  0.0028,  0.0141, -0.0124, -0.0016, -0.0220, -0.0234, -0.0119,
        -0.0143, -0.0430,  0.0032, -0.0266, -0.0092, -0.0104, -0.0248,  0.0084,
         0.0179,  0.0011, -0.0021, -0.0527,  0.0033, -0.0186, -0.0359, -0.0017,
        -0.0002, -0.0708,  0.0232, -0.0194, -0.0020, -0.0421, -0.0328, -0.0364,
        -0.0145,  0.0368,  0.0058,  0.0468, -0.0393,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0761,  0.2600,  0.0528,  0.1063,  0.0023,  0.0013,  0.0245, -0.0098,
         0.0335, -0.0382, -0.0201,  0.0419,  0.0039,  0.0314,  0.0450, -0.0715,
         0.0117,  0.0134, -0.0226,  0.0110,  0.0418, -0.0251, -0.0295,  0.0262,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1072,  0.0158,  0.0118,  0.0349, -0.0043, -0.0085, -0.0214, -0.0451,
         0.0113,  0.0015,  0.0098,  0.0015, -0.0001, -0.0444, -0.0307, -0.0238,
        -0.0635, -0.0380, -0.0322, -0.0519, -0.0310, -0.0182, -0.0347, -0.0241,
        -0.0323, -0.0564, -0.0974, -0.0177, -0.0282, -0.0412, -0.0015, -0.0489,
         0.0110,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1346, -0.1907, -0.0530, -0.0472,  0.0189,  0.0386,  0.0117, -0.0134,
        -0.1640, -0.0204,  0.0248, -0.0085,  0.0526, -0.0377, -0.0290, -0.0360,
        -0.0113,  0.0405, -0.0672,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2228, -0.0108,  0.0303, -0.0401, -0.0711, -0.0183, -0.0682, -0.0486,
        -0.1230, -0.0043,  0.0024, -0.0210, -0.0281, -0.0092, -0.0011,  0.0005,
        -0.0013, -0.0324, -0.0096,  0.0030, -0.0496, -0.1377, -0.0040, -0.0214,
         0.0412,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0214, -0.2244, -0.0465, -0.0442, -0.0927, -0.0576, -0.2413, -0.0572,
        -0.0189, -0.0365, -0.0026, -0.0090, -0.0345, -0.0229, -0.0485,  0.0223,
         0.0197,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.1417, -0.0230,  0.0389, -0.0057,  0.0012,  0.0159, -0.0050,  0.0436,
        -0.0024,  0.0739, -0.0003,  0.0227,  0.0157,  0.0289,  0.0172,  0.0180,
         0.0446,  0.1060, -0.0190,  0.0399, -0.0062,  0.0599,  0.1466,  0.0032,
         0.0050,  0.0084,  0.0109,  0.0155, -0.0141, -0.0130, -0.0497, -0.0039,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.0053e-01,  1.3058e-01,  4.9335e-02,  2.3662e-02,  1.4778e-02,
         2.1937e-02,  4.0076e-02,  7.3299e-02,  7.5618e-03,  1.0071e-03,
         1.5176e-02, -1.3742e-02,  1.4742e-01,  2.2865e-02,  2.9267e-02,
        -2.6153e-02,  2.1404e-02,  1.7154e-02,  5.1197e-02,  7.7651e-03,
         1.2817e-02, -7.8110e-03,  6.4598e-02,  5.6122e-02,  1.0126e-02,
        -1.5419e-02,  1.8120e-02, -7.8708e-05,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1590, -0.0675, -0.0352, -0.0303, -0.0224, -0.0270, -0.1084, -0.0057,
        -0.0297, -0.0742, -0.0081, -0.0620, -0.0091,  0.0046, -0.0040, -0.0491,
        -0.0102, -0.0629, -0.0240, -0.0085, -0.0038,  0.0084,  0.0043, -0.0256,
        -0.0225,  0.0162, -0.0179,  0.0353,  0.0641,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0184, -0.0261,  0.0023,  0.0287,  0.1168,  0.0102,  0.0469,  0.0162,
        -0.0024,  0.0046, -0.0214,  0.0272,  0.0485,  0.0939, -0.0285,  0.0399,
         0.0613,  0.1507,  0.0366,  0.0119,  0.0165, -0.0029,  0.0211,  0.0032,
         0.0139,  0.0129,  0.0122,  0.0302,  0.0362, -0.0195,  0.0092, -0.0047,
        -0.0252,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1385, -0.1948, -0.0051, -0.0412,  0.0054, -0.0085, -0.0126, -0.0395,
        -0.0142, -0.0159, -0.0068, -0.0071, -0.0138, -0.0075, -0.0169, -0.0087,
        -0.0007, -0.0139, -0.1146,  0.0171, -0.0690, -0.0246,  0.1175, -0.1060,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0869, -0.1799, -0.1381,  0.0043, -0.0281, -0.0443, -0.0689, -0.0996,
         0.0093, -0.0463, -0.0315, -0.0159, -0.0194, -0.0146,  0.0009, -0.0018,
         0.0051, -0.0271, -0.0108, -0.1213,  0.0091,  0.0049,  0.0319,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0419, -0.0944,  0.0080, -0.0200, -0.0102, -0.0724, -0.0044, -0.0005,
        -0.0061, -0.0289,  0.0291, -0.0083, -0.0198, -0.0078, -0.0204, -0.0081,
        -0.0578, -0.0468, -0.0297, -0.0250,  0.0281, -0.0516,  0.0211, -0.0103,
        -0.0326, -0.0251, -0.0170, -0.0252, -0.0249, -0.0815,  0.0051, -0.0052,
        -0.0153, -0.0156, -0.0176, -0.0311, -0.0221, -0.0046, -0.0263,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0136, -0.0343,  0.0262, -0.0009, -0.1082, -0.0315,  0.0206, -0.0029,
        -0.0259,  0.0047,  0.0036,  0.0050, -0.0111, -0.0289, -0.0081, -0.0749,
        -0.0237, -0.0512,  0.0193,  0.0026, -0.0623, -0.0279, -0.0079, -0.0180,
        -0.0331, -0.0685, -0.0275, -0.0103,  0.0038, -0.0159, -0.0174, -0.0247,
        -0.0092, -0.0034, -0.0170,  0.0076, -0.0016, -0.0743,  0.0083, -0.0169,
        -0.0118,  0.0228, -0.0126,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.2027e-02, -2.1550e-01,  7.5020e-03,  7.9736e-03, -1.0254e-02,
        -2.9234e-03, -2.1251e-03, -9.3170e-03, -2.8137e-02,  2.1876e-02,
         2.3181e-02,  3.3380e-03, -1.7535e-02,  3.6282e-03, -2.6761e-03,
        -1.9337e-02, -2.6491e-02, -3.1665e-02,  4.4543e-03, -9.7329e-04,
         7.2190e-03, -5.8269e-03, -6.7425e-03, -1.6400e-03,  3.6958e-04,
         1.5455e-03,  2.7812e-02, -4.1088e-03,  8.6751e-03, -5.3616e-03,
         9.5300e-03, -1.0420e-02, -2.0794e-02,  3.6201e-02, -5.0209e-03,
         1.0980e-04,  3.3604e-03, -3.4666e-03, -5.2645e-02,  1.7856e-02,
        -2.9688e-02, -7.6354e-02, -8.8494e-03, -1.2946e-02, -4.8401e-03,
        -1.6639e-02, -5.5615e-03, -1.8252e-03,  6.0115e-03,  1.3388e-02,
        -1.1138e-02, -2.2219e-02, -3.7199e-02,  6.1955e-03,  1.7384e-02,
        -3.1803e-02, -1.8341e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0092, -0.3590, -0.0972, -0.0068, -0.0686, -0.0466, -0.0739, -0.1274,
        -0.0252,  0.0064, -0.0457,  0.0171, -0.0581,  0.0588,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0784,  0.1479, -0.0304, -0.0440, -0.0113,  0.0104,  0.0096,  0.0046,
        -0.0081,  0.0112,  0.0308,  0.1494,  0.0276,  0.0253,  0.0357,  0.0070,
         0.0074,  0.0373, -0.0074,  0.0540,  0.0992,  0.0214,  0.0110,  0.0034,
        -0.0082,  0.0030, -0.0074,  0.0088, -0.0797,  0.0200,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0845,  0.2234,  0.0304,  0.2452,  0.0297, -0.0837,  0.1006,  0.2026,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([-0.0501, -0.2376, -0.0549,  0.0274, -0.0241, -0.0687, -0.0089,  0.0203,
         0.0027, -0.0161,  0.0045, -0.0253, -0.0376, -0.0300, -0.0059, -0.0166,
         0.0056, -0.0138, -0.0033, -0.0119, -0.0033,  0.0092, -0.0122,  0.0277,
        -0.0348,  0.0173, -0.0278,  0.0128, -0.0509, -0.0282, -0.0212,  0.0580,
         0.0312,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1015e-02,  2.4064e-01,  4.6991e-02,  4.3409e-02, -1.8590e-02,
         2.9286e-02,  1.7625e-02,  3.9773e-02, -3.7578e-03,  1.4266e-02,
        -3.2179e-03,  4.1915e-03,  2.3159e-03,  5.9625e-03, -1.4251e-02,
         1.0814e-02, -1.2081e-03, -8.3947e-03,  8.2453e-03, -4.3065e-03,
        -3.0003e-03, -9.7307e-04,  9.7140e-03,  1.8696e-04, -1.5367e-02,
         1.3671e-02,  2.3874e-02,  5.4853e-02, -5.4720e-03, -2.8534e-03,
         7.5296e-03,  3.0588e-03,  1.3107e-02,  5.7976e-02,  1.3014e-02,
         1.5700e-02,  5.0303e-03,  4.2038e-03,  2.7815e-03,  6.8307e-04,
        -2.3821e-03, -1.3396e-03,  4.8132e-03,  2.0998e-02, -3.5072e-03,
        -6.7519e-03,  1.0801e-02,  2.1836e-03,  5.3878e-03, -4.8010e-03,
         9.0218e-03,  6.8947e-03, -3.4892e-05,  2.5781e-03, -1.6286e-03,
        -1.0567e-02,  1.6446e-03,  8.8278e-04,  2.3284e-03,  1.3149e-02,
         7.2907e-03,  5.1013e-03,  5.6098e-02, -2.6245e-03,  1.9324e-03,
        -1.3542e-03,  4.7304e-03, -1.5544e-02,  1.9698e-03,  9.4077e-03,
        -9.3903e-04], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0544, -0.0335, -0.0153,  0.0033, -0.0320, -0.0099, -0.1427, -0.0082,
         0.0030,  0.0171,  0.0126, -0.0220, -0.0108,  0.0078, -0.0085, -0.0265,
        -0.0030, -0.0015, -0.0035, -0.0655, -0.0095,  0.0099,  0.0101, -0.0673,
         0.0490, -0.0250, -0.0343, -0.0014,  0.0151, -0.0072, -0.0026, -0.0213,
        -0.0379, -0.0681, -0.0176, -0.0409, -0.0468, -0.0228,  0.0030,  0.0098,
        -0.0057, -0.0140,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0959, -0.1374, -0.0084, -0.0207,  0.0014,  0.0074,  0.0186, -0.0059,
        -0.0283, -0.0703, -0.0212, -0.0306, -0.0651, -0.0016,  0.0044, -0.0035,
         0.0016, -0.0345, -0.0921, -0.0179, -0.0166, -0.0058,  0.0018, -0.0092,
        -0.0233, -0.0541, -0.0270, -0.0131, -0.0419, -0.0137, -0.0226, -0.0107,
         0.0178, -0.0094, -0.0075, -0.0189,  0.0396,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1579, -0.3687, -0.0691, -0.0673, -0.0259,  0.0821, -0.0264,  0.0195,
        -0.0136, -0.0523, -0.0161, -0.0310, -0.0600, -0.0081,  0.0019,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0141,  0.1167, -0.0053, -0.0596,  0.0586,  0.1037,  0.0702,  0.0338,
         0.1831,  0.0172, -0.0177, -0.0034,  0.0879, -0.0029, -0.0879,  0.0168,
         0.0185, -0.0548,  0.0152, -0.0326,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0112, -0.0189, -0.0070,  0.0013, -0.1326, -0.0202, -0.0148, -0.0128,
        -0.0378, -0.0158, -0.0217, -0.0143, -0.0618, -0.0127, -0.0266, -0.0127,
         0.0026, -0.0181, -0.0051, -0.0207, -0.0074, -0.0003, -0.0073, -0.0060,
         0.0103, -0.0312, -0.0590, -0.0104, -0.0267, -0.0070, -0.0372, -0.0342,
        -0.0090, -0.0322, -0.0226, -0.0201, -0.0318, -0.0032, -0.0121, -0.0478,
         0.0654,  0.0467, -0.0036,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1042, -0.0034,  0.0357, -0.0251, -0.0173, -0.0347, -0.0206, -0.1385,
        -0.0198, -0.0064, -0.0220, -0.0036, -0.0344, -0.0630, -0.0655, -0.0253,
         0.0084,  0.0124,  0.0085, -0.0117, -0.0137, -0.0077,  0.0290,  0.0123,
         0.0123,  0.0304,  0.0086, -0.0010, -0.0806, -0.0114, -0.0085, -0.0268,
        -0.0155, -0.0218, -0.0221, -0.0009,  0.0095,  0.0273,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0590,  0.1680,  0.0644,  0.0911, -0.0195,  0.0901,  0.0119, -0.0682,
         0.0257,  0.1126,  0.0134,  0.0114,  0.0473,  0.0110,  0.0106,  0.0204,
         0.0420,  0.0363, -0.0013,  0.0246, -0.0593,  0.0118,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0725, -0.1361, -0.0202, -0.0267, -0.0015, -0.0064,  0.0079, -0.0080,
         0.0067, -0.0675, -0.0294, -0.0120, -0.0104, -0.0015, -0.0142, -0.0221,
        -0.0925, -0.0276, -0.0285, -0.0417, -0.0333,  0.0014, -0.0343, -0.0239,
        -0.0202, -0.0163, -0.0041,  0.0124, -0.0007, -0.0409, -0.0128,  0.0038,
        -0.0133, -0.0162, -0.0162, -0.0597, -0.0041,  0.0033, -0.0097, -0.0398,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0340,  0.0783,  0.0136, -0.1004,  0.0269, -0.0108,  0.0377, -0.0139,
        -0.0450, -0.0223,  0.0026,  0.0153, -0.0260, -0.1358, -0.0386, -0.0172,
         0.0087, -0.0193, -0.0371, -0.0141, -0.0238, -0.0148, -0.0301, -0.0266,
        -0.0102, -0.0751,  0.0089, -0.0384, -0.0415,  0.0020, -0.0089,  0.0220,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0473e-01, -3.2965e-01, -8.9842e-03, -1.7671e-02, -1.7653e-02,
         1.3086e-02,  1.1833e-02, -3.9964e-02, -7.1963e-02, -7.6660e-03,
        -2.1669e-04, -2.7659e-02, -1.5600e-02,  3.5759e-03,  1.6604e-02,
         1.4172e-02, -6.0871e-02,  1.6269e-02, -4.5015e-03,  2.3567e-02,
        -3.6604e-03, -1.2099e-02,  4.4510e-03, -2.1145e-02, -2.6637e-03,
         2.9366e-03, -1.2686e-02, -5.4763e-04,  4.5358e-03, -1.0556e-02,
         7.1475e-04,  2.0844e-03, -2.1668e-03, -5.4258e-02,  5.9259e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0984, -0.1402, -0.0892, -0.0441, -0.0578, -0.0384, -0.1111, -0.0821,
        -0.0273, -0.1100,  0.0130, -0.0270, -0.0354, -0.0513, -0.0303, -0.0020,
         0.0005, -0.0062, -0.0067, -0.0083, -0.0208,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1123, -0.2859,  0.0794, -0.0810, -0.0634, -0.0323, -0.0231, -0.0522,
        -0.1224,  0.0163,  0.0122,  0.0051,  0.0039,  0.0566, -0.0012,  0.0121,
        -0.0132, -0.0261,  0.0011,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1914, -0.1252, -0.0564, -0.0127, -0.0454, -0.0043,  0.0004,  0.0224,
        -0.0387, -0.0799, -0.1107,  0.0178, -0.0306, -0.1291,  0.0115, -0.0350,
        -0.0116, -0.0181, -0.0310,  0.0278,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0013, -0.2690, -0.0311, -0.1155,  0.0183, -0.0051, -0.0736, -0.0291,
        -0.0655, -0.0474, -0.0147,  0.0049,  0.0044, -0.0103,  0.0042, -0.0098,
        -0.1019, -0.0080, -0.0174, -0.0551,  0.0146,  0.0345,  0.0642,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0324, -0.2398, -0.0828,  0.0136, -0.0020, -0.0370, -0.0471, -0.0354,
        -0.0321, -0.1522, -0.0204, -0.0526, -0.0400, -0.0874,  0.0105, -0.0137,
         0.0034,  0.0191, -0.0139, -0.0046, -0.0234, -0.0053, -0.0072,  0.0068,
        -0.0173,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0481,  0.1718,  0.0166,  0.0019,  0.0059, -0.0024, -0.0163,  0.0506,
         0.0128,  0.0136,  0.0479,  0.0155, -0.0037, -0.0052,  0.0084, -0.0581,
         0.0046, -0.0097, -0.0038,  0.0311,  0.0212,  0.0927,  0.0403,  0.0532,
        -0.0067,  0.0106,  0.0016,  0.0287,  0.0059,  0.0039, -0.0040, -0.0387,
        -0.0127,  0.0885, -0.0633,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0556, -0.0004,  0.0425, -0.0686, -0.0067, -0.0491, -0.0295, -0.0309,
        -0.1850, -0.0153, -0.0369,  0.0079,  0.0086, -0.0525, -0.0021, -0.1434,
        -0.0330, -0.0903, -0.0540, -0.0239,  0.0377, -0.0259,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1996, -0.0590,  0.0050,  0.0039, -0.0570, -0.0738,  0.0121,  0.0016,
        -0.0416, -0.1505, -0.0114, -0.0327, -0.0124, -0.0394, -0.0228, -0.0738,
        -0.0189, -0.0182,  0.0394,  0.0298, -0.0971,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0007, -0.2359, -0.0404,  0.0493, -0.0313, -0.0243, -0.0648, -0.1360,
         0.0394, -0.0299, -0.0079, -0.0105, -0.0299, -0.0571, -0.0189, -0.1181,
         0.0327, -0.0204,  0.0526,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.2246e-01, -3.8039e-02, -4.4344e-02, -5.1573e-03, -2.4715e-03,
        -7.4711e-02,  4.7456e-03, -2.6144e-03, -4.3178e-03,  3.6661e-03,
        -1.6020e-04, -1.0674e-03, -5.4648e-03, -2.5358e-02, -2.4216e-03,
        -1.7210e-03, -1.3648e-03, -6.8104e-03, -8.6383e-03, -1.5233e-02,
        -9.2105e-03,  2.3999e-02, -2.0171e-02,  1.7939e-04, -4.3279e-02,
        -1.5984e-02, -7.0416e-05, -3.7808e-02, -2.3080e-02, -5.5787e-02,
         1.2417e-02, -4.3174e-02, -4.2196e-02, -1.2775e-01,  1.0043e-02,
        -4.2810e-03,  1.1321e-02,  6.7770e-03, -1.5158e-02, -2.1378e-02,
        -4.8758e-03,  1.1625e-02, -2.5801e-03,  9.5391e-03, -1.8978e-02,
         2.8654e-02,  2.8919e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.5579e-02, -1.7913e-01, -2.9445e-02, -7.7370e-03,  1.7350e-02,
         2.0510e-02,  9.0164e-03, -9.4717e-03, -3.0491e-02,  5.3601e-03,
         1.7997e-02,  1.3098e-02,  2.7752e-03,  1.2566e-02,  3.2130e-03,
        -1.7269e-02, -3.3177e-02,  1.4227e-02, -1.2795e-02, -5.0217e-03,
        -2.7414e-02,  3.6180e-03, -6.9427e-02, -1.7358e-01, -8.5188e-03,
         5.9233e-02, -2.9085e-02, -2.3731e-02, -2.3498e-05,  2.5084e-02,
        -5.6507e-03, -3.5988e-03, -4.2015e-03, -9.0607e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1074, -0.2676, -0.0111,  0.0298, -0.0234, -0.0251, -0.0214, -0.0005,
        -0.0053, -0.0065, -0.0007,  0.0137, -0.0356, -0.0878, -0.0045,  0.0187,
         0.0100,  0.0458, -0.0072, -0.0514, -0.0228, -0.0135, -0.0170,  0.0232,
        -0.0082,  0.0030, -0.0093, -0.0166, -0.0039, -0.0294, -0.0106,  0.0117,
         0.0013, -0.0562,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0361, -0.3754,  0.0160,  0.0246, -0.0342, -0.0621,  0.0112, -0.0070,
        -0.0172, -0.1106, -0.0805, -0.0283, -0.0355, -0.0758, -0.0089, -0.0237,
        -0.0420,  0.0110,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0272,  0.0002, -0.0594,  0.0802, -0.0322,  0.0178, -0.0152,  0.0087,
        -0.1230, -0.1329,  0.0173, -0.0064,  0.0436, -0.0217, -0.0275, -0.0149,
        -0.0148,  0.0522, -0.0027, -0.1067,  0.0211,  0.0047, -0.0514,  0.0090,
         0.0186, -0.0571,  0.0334,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.8492e-03,  5.7722e-03,  4.8456e-03, -1.2503e-02, -1.8297e-02,
        -3.0830e-02,  1.8610e-04, -1.3757e-02, -1.1582e-01,  7.3602e-02,
         4.3092e-02, -1.6981e-02, -2.2620e-01, -4.9943e-02, -1.0481e-02,
         2.9635e-02,  4.0350e-03, -3.9595e-02,  1.6631e-03, -1.7321e-01,
        -4.2979e-03, -1.2188e-03, -1.6243e-02, -1.2086e-02,  2.5997e-02,
         6.0869e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0868, -0.2878,  0.0364, -0.0486, -0.0464,  0.0221,  0.0076, -0.0642,
        -0.0365, -0.0552, -0.0900, -0.0042, -0.0198, -0.0211, -0.0284, -0.0450,
        -0.0133, -0.0131, -0.0140, -0.0216,  0.0380,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0214,  0.2485,  0.1426,  0.1412,  0.0373,  0.0069,  0.1264,  0.0254,
         0.0167, -0.1141,  0.0183,  0.0371, -0.0099, -0.0478,  0.0064,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1204,  0.0008,  0.1016,  0.0198, -0.0839,  0.1003,  0.0033,  0.0246,
         0.0462, -0.0068,  0.0587,  0.0597,  0.2136,  0.0564, -0.0506, -0.0229,
        -0.0060,  0.0245,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0774, -0.1809, -0.0287, -0.1339, -0.0365,  0.0086, -0.0098, -0.0445,
        -0.0160, -0.0967, -0.0819, -0.0483,  0.0106, -0.0699,  0.0237,  0.0019,
         0.0166, -0.0374,  0.0769,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0425, -0.1198, -0.0537,  0.0292,  0.0059, -0.0184,  0.0188,  0.0309,
        -0.0035,  0.0056, -0.0009,  0.0016, -0.0172,  0.0056, -0.0050, -0.0011,
        -0.0044,  0.0057,  0.0102, -0.0040, -0.0833,  0.0007, -0.0553, -0.0237,
        -0.0291, -0.0339, -0.0377,  0.0208, -0.0062, -0.0280, -0.0100, -0.0248,
        -0.0443, -0.0220, -0.0128, -0.0122, -0.0502,  0.0069, -0.0181,  0.0106,
         0.0172,  0.0301,  0.0238, -0.0141], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.0191e-02, -2.2998e-01, -3.9499e-02, -2.6085e-02,  6.9528e-03,
         2.5175e-04, -1.2484e-02,  1.8766e-02, -5.0294e-02, -6.4531e-03,
        -4.5190e-02, -1.3086e-02,  1.6496e-02, -4.8869e-03,  1.6411e-02,
        -3.4727e-03, -2.4169e-02, -4.8153e-02,  9.9072e-03,  5.8164e-03,
        -4.5478e-02,  3.5674e-03,  6.7948e-03, -8.2483e-05,  1.6374e-02,
         2.2238e-03,  5.2058e-03, -5.9197e-02,  2.9506e-02, -5.9230e-03,
        -1.7126e-02, -5.9008e-02, -7.9077e-04, -7.2459e-02, -4.7724e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0707,  0.3540, -0.0014,  0.0295,  0.0401,  0.0209,  0.0951,  0.0789,
         0.0183, -0.0007, -0.0019,  0.0365,  0.0856, -0.0007,  0.0047,  0.0483,
        -0.0093, -0.0036, -0.0238, -0.0038, -0.0046,  0.0239, -0.0306, -0.0129,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.9370e-02,  2.3977e-01,  2.2503e-02, -2.4833e-02,  1.5860e-02,
         1.2739e-02, -6.9221e-03,  5.6861e-03,  6.2122e-02,  1.3238e-02,
         2.3759e-02, -7.5171e-03, -7.5086e-03,  6.2154e-02,  5.3264e-02,
         2.1841e-04, -1.8672e-03,  1.0227e-02,  1.1856e-02,  2.3538e-03,
        -5.1744e-03,  4.7076e-02,  4.5106e-03,  7.4117e-04,  1.1967e-02,
         3.0744e-02,  3.3096e-02,  2.3734e-02,  1.7562e-02, -2.7250e-03,
         7.3998e-02,  1.6884e-02,  1.0545e-02,  7.7479e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.0936e-02,  3.2147e-05, -9.2327e-03, -7.6199e-02,  2.2984e-02,
         3.0241e-02, -1.5287e-01, -2.5306e-02, -4.4758e-03, -1.8780e-02,
        -1.3842e-02,  2.3272e-02, -4.7074e-02, -2.4275e-02, -2.6929e-02,
        -1.7736e-01, -2.2954e-04,  7.3371e-03, -9.5458e-03,  1.8404e-03,
         2.8771e-02, -2.3290e-02, -1.2168e-02, -2.2895e-02, -4.0176e-02,
        -2.8094e-02, -3.8064e-02,  1.0240e-02, -5.3542e-02,  9.9936e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0555, -0.1807, -0.0412, -0.2625, -0.0258, -0.0019, -0.0271, -0.0190,
        -0.0078, -0.2122, -0.0659,  0.0080, -0.0925,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0532, -0.0321, -0.0516,  0.0168, -0.0289, -0.0016, -0.0535, -0.0504,
        -0.1900, -0.0428, -0.0459, -0.0595,  0.0236, -0.0325, -0.1280, -0.0632,
         0.0917,  0.0345,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0966, -0.1573, -0.2741, -0.0494, -0.1940, -0.0646,  0.0209, -0.1430,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0359, -0.1481, -0.0333,  0.0035, -0.0288, -0.0351, -0.0062, -0.0119,
         0.0067,  0.0047, -0.0058,  0.0097, -0.0063,  0.0112, -0.0021, -0.0003,
         0.0050, -0.0014,  0.0044,  0.0005,  0.0223,  0.0196, -0.0174, -0.0260,
        -0.0019, -0.0254, -0.0063,  0.0046, -0.0820, -0.0015,  0.0148, -0.0583,
        -0.0205, -0.0496, -0.0355, -0.0384, -0.0215, -0.0088, -0.0176, -0.0066,
        -0.0093, -0.0171, -0.0020,  0.0116,  0.0100,  0.0794,  0.0311],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.4300e-01,  1.7042e-01,  1.9655e-01,  2.2505e-01,  3.8166e-02,
        -1.1395e-01,  9.6101e-03, -4.0468e-02, -3.2004e-02,  3.0691e-02,
        -9.1014e-05,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0549, -0.0550, -0.0399, -0.0527, -0.0197, -0.0595, -0.0551, -0.2248,
        -0.0747, -0.0949, -0.0201, -0.0234, -0.1150, -0.0149,  0.0399,  0.0008,
        -0.0045, -0.0501,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0702, -0.0790,  0.1620,  0.0242,  0.0049,  0.0358,  0.0560, -0.0040,
         0.0470,  0.0648, -0.0286,  0.0853,  0.1608, -0.1561,  0.0213,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1917, -0.2935,  0.0271,  0.0030,  0.0384, -0.0357, -0.1075,  0.0024,
         0.0543, -0.0568, -0.0667,  0.0063,  0.0349, -0.0132, -0.0686,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1021,  0.0051,  0.0215,  0.0368,  0.0140, -0.0032,  0.0351, -0.0205,
        -0.0334, -0.0268, -0.0154, -0.0583, -0.0937,  0.0116, -0.0261, -0.0253,
        -0.0587,  0.0804,  0.0334, -0.0231, -0.0660, -0.0654,  0.0101,  0.0512,
        -0.0035, -0.0795,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1469, -0.0860, -0.0156, -0.0368, -0.0710, -0.1154,  0.0011, -0.1352,
        -0.1529, -0.0316,  0.0119, -0.0452, -0.0599,  0.0631,  0.0275,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0815, -0.3478, -0.0490, -0.0058, -0.1169, -0.0452, -0.0306, -0.0009,
        -0.0147, -0.0926,  0.0111, -0.0027, -0.0111, -0.0181,  0.0072, -0.0018,
        -0.0058, -0.0250, -0.0446,  0.0022, -0.0147,  0.0266, -0.0249,  0.0131,
         0.0012,  0.0049,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1247, -0.0064,  0.0181,  0.0185, -0.0257, -0.0202, -0.0114, -0.0207,
        -0.0184, -0.0384, -0.2003, -0.0317, -0.0195, -0.0523, -0.0795, -0.0274,
        -0.0299, -0.0509, -0.0146, -0.0281, -0.0078, -0.0034,  0.0029, -0.0177,
         0.0875,  0.0442,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.0574, -0.2581, -0.0061,  0.0364,  0.0195,  0.0034, -0.0098, -0.0032,
        -0.0065, -0.0217, -0.0007, -0.0145, -0.0933, -0.0018, -0.0236, -0.0224,
        -0.0071,  0.0531, -0.0300, -0.0180,  0.0174, -0.0222,  0.0037, -0.0654,
        -0.0217, -0.0003, -0.0448, -0.0144,  0.0142, -0.0013, -0.0055, -0.0129,
         0.0191, -0.0058, -0.0089,  0.0558,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0115, -0.1136, -0.0520, -0.0267, -0.0439, -0.0024, -0.0677, -0.0912,
        -0.0016,  0.0196, -0.0398,  0.0016, -0.0252,  0.0214, -0.0016, -0.0062,
         0.0127, -0.0104,  0.0012,  0.0400, -0.0089, -0.0200, -0.0174, -0.0036,
        -0.0474, -0.0797,  0.0143, -0.0202, -0.0692, -0.0036, -0.0030, -0.0295,
        -0.0634,  0.0089, -0.0114,  0.0091,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2017,  0.3070,  0.0836,  0.0132,  0.0080, -0.0037,  0.0332,  0.0132,
         0.0456, -0.0244,  0.0159, -0.0116,  0.0281, -0.0162, -0.0269,  0.0126,
         0.0422,  0.0176, -0.0273, -0.0680,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1077, -0.0230,  0.0067,  0.0028,  0.0168, -0.0302, -0.0243, -0.0081,
        -0.0154, -0.0236,  0.0181, -0.0191, -0.0163,  0.0005, -0.0730, -0.0014,
         0.0058, -0.0350, -0.0048, -0.0236, -0.0525, -0.0030, -0.0116,  0.0083,
        -0.0335, -0.0354, -0.0178, -0.0003, -0.0036, -0.0452, -0.0223, -0.0187,
         0.0197, -0.1030, -0.0116,  0.0199,  0.1355, -0.0018,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0155, -0.0669, -0.0253, -0.0810, -0.0164,  0.0083, -0.0057, -0.0003,
         0.0227, -0.0004, -0.0597, -0.0673, -0.0101,  0.0112, -0.0095, -0.0553,
         0.0471, -0.0448,  0.0036,  0.0229,  0.0313,  0.0206,  0.0336,  0.0381,
         0.0062, -0.0025,  0.0071, -0.0263, -0.0863,  0.0122, -0.0103, -0.0038,
         0.0289, -0.0030, -0.0267,  0.0022, -0.0091,  0.0282,  0.0208,  0.0167,
        -0.0122], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1150,  0.3369, -0.0131,  0.0311,  0.0239,  0.0177,  0.0382,  0.0701,
         0.0209,  0.0151,  0.0268,  0.0591,  0.0393,  0.0171,  0.0017,  0.0541,
         0.0156,  0.0463, -0.0017,  0.0117,  0.0379,  0.0069,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0740,  0.0012, -0.0077,  0.0041,  0.0039, -0.0547, -0.0145,  0.0024,
        -0.0255,  0.0076, -0.0223, -0.0175, -0.0100, -0.0486, -0.0195, -0.0534,
        -0.0175, -0.0216,  0.0082, -0.0473, -0.0776,  0.0137,  0.0130, -0.0394,
        -0.0102, -0.0014, -0.0166, -0.0299, -0.0354, -0.0023, -0.0193, -0.0184,
        -0.0208, -0.1492,  0.0914,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0399, -0.0287, -0.0866, -0.0402, -0.0937, -0.0327,  0.0096, -0.0256,
        -0.0237, -0.0202, -0.0524, -0.0046, -0.0112, -0.0123, -0.0036, -0.0355,
         0.0426, -0.0146, -0.0850,  0.0060, -0.0470, -0.0150, -0.0061,  0.0072,
        -0.0083, -0.0032, -0.0457, -0.0581,  0.1409,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0142,  0.2761,  0.0939,  0.0666,  0.0156,  0.0458,  0.0320,  0.0266,
         0.0169,  0.0365,  0.0169,  0.0484, -0.0263, -0.0420,  0.0058,  0.0235,
         0.0159,  0.0148,  0.0275, -0.0098,  0.0035, -0.0014, -0.0176, -0.0162,
         0.1063,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0978,  0.0403,  0.0113, -0.0326, -0.1669, -0.0384, -0.0399, -0.1127,
        -0.0907, -0.0353,  0.0012,  0.0098,  0.0023,  0.0023, -0.0579, -0.1229,
        -0.0251, -0.0175,  0.0071, -0.0070, -0.0319,  0.0052,  0.0438,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1387e-01,  2.8323e-02, -5.7589e-03, -9.9323e-03, -2.2841e-02,
         2.0116e-02,  1.5294e-02, -1.1527e-02, -5.2956e-04, -1.1477e-04,
         1.3909e-02, -1.0755e-02, -3.0238e-02, -6.8934e-03, -7.5628e-03,
        -2.6683e-02, -5.9924e-03, -2.2191e-02, -1.3416e-02, -1.3036e-02,
        -9.1087e-03,  1.8511e-02, -3.9671e-03, -7.0128e-03, -3.7323e-02,
        -5.8567e-02, -1.5915e-01,  3.2009e-03, -7.5171e-02, -3.2204e-02,
        -1.8472e-02,  7.3001e-03, -4.8999e-02, -3.5795e-02, -6.1427e-03,
         2.3756e-02,  2.4390e-02,  3.3060e-03,  4.8636e-02,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0594,  0.1088, -0.0032,  0.0497,  0.2776,  0.0745,  0.0206,  0.1098,
         0.2398,  0.0381,  0.0005,  0.0181,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0698, -0.0976, -0.1023, -0.0003, -0.0091, -0.0585, -0.0349,  0.0298,
        -0.0115, -0.0197, -0.0013,  0.0085, -0.0058,  0.0093, -0.0130,  0.0132,
        -0.0150, -0.0404,  0.0090, -0.0258, -0.0067, -0.0227,  0.0279, -0.0254,
         0.0051, -0.0090,  0.0245, -0.0331,  0.0116,  0.0035,  0.0112, -0.0486,
        -0.0357,  0.0001, -0.0264, -0.0033, -0.0425,  0.0007,  0.0050, -0.0038,
         0.0201,  0.0033,  0.0404,  0.0029, -0.0117,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0676, -0.4810, -0.0231, -0.0198,  0.0335, -0.0456, -0.0263, -0.0558,
        -0.1391, -0.0410, -0.0324,  0.0348,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2678, -0.3520,  0.0620, -0.1194, -0.0391, -0.0887, -0.0353, -0.0357,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.2455e-02,  1.4617e-01,  2.0958e-02, -1.3839e-02,  1.5626e-02,
         1.4815e-02, -4.4233e-03,  7.0009e-04,  6.3892e-03,  6.1402e-02,
         5.0789e-03, -7.4175e-03,  2.9612e-03,  1.1009e-02,  2.1684e-03,
         3.8213e-04, -3.8106e-05,  1.2753e-02,  1.3953e-02,  6.0016e-03,
         1.9060e-02, -4.3610e-03,  7.0719e-03,  1.6222e-02, -4.3091e-03,
         1.9877e-02,  7.0444e-03,  9.0812e-04,  1.5781e-02, -1.1973e-02,
         1.9540e-02,  9.3224e-03,  1.8498e-02,  3.1417e-04, -2.5813e-03,
         1.4224e-03,  2.4522e-02, -1.0108e-02,  8.4294e-04,  1.6891e-02,
         5.7498e-04,  1.0557e-02,  1.3827e-02,  3.1813e-03,  9.4044e-03,
        -8.5450e-02,  3.9243e-02,  5.0143e-03,  5.1279e-02,  5.1767e-02,
         9.6732e-03, -5.5248e-03,  1.8384e-02, -1.1541e-02,  1.7725e-03,
        -2.1911e-03,  1.3817e-02, -1.1310e-02, -6.3618e-03,  1.2171e-02,
         7.7661e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1920, -0.0225,  0.0198, -0.0190, -0.0858, -0.0226, -0.0085, -0.0170,
        -0.0092,  0.0110,  0.0018, -0.0155, -0.0087, -0.0229, -0.0685,  0.0023,
        -0.0189, -0.0122, -0.0260, -0.0149, -0.0622, -0.0212, -0.0050,  0.0251,
        -0.0058, -0.0237,  0.0004, -0.0079, -0.0094, -0.0388, -0.0216,  0.0882,
        -0.0917,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.1979e-04, -1.2297e-01, -1.2225e-01, -4.7599e-02, -3.4276e-02,
        -1.8566e-02, -5.2127e-02,  1.8297e-02, -2.4146e-03,  1.6234e-02,
         1.4258e-02, -5.4268e-03, -2.4329e-03, -1.1390e-02, -2.0344e-02,
         5.7596e-02,  1.0098e-02, -5.1737e-02, -1.8553e-01,  7.1733e-03,
         5.9684e-02, -3.6430e-03,  8.7036e-02, -4.8799e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0145, -0.1170, -0.0587, -0.0420, -0.0428, -0.0948, -0.0344, -0.0168,
         0.0206, -0.1267, -0.0438, -0.0581, -0.1123, -0.0428, -0.0181, -0.0168,
         0.0137, -0.0107, -0.0071,  0.0173, -0.0099, -0.0084,  0.0492,  0.0233,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1249, -0.0072, -0.0171, -0.0152, -0.0156, -0.0031, -0.0269, -0.0026,
        -0.0193, -0.0210,  0.0156, -0.0204, -0.0117, -0.0548, -0.0413, -0.0566,
        -0.0251, -0.0075, -0.0394, -0.0124, -0.0256, -0.0008, -0.0262, -0.0266,
        -0.0256,  0.0048, -0.0117, -0.0088, -0.0066,  0.0096, -0.0200,  0.0009,
        -0.0244,  0.2711,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.1445e-02,  3.9421e-03,  2.4119e-02,  2.8271e-02, -7.4198e-02,
         1.1305e-02,  4.1673e-03,  4.6106e-02, -5.7886e-02, -1.2724e-02,
         1.4267e-03, -2.3715e-02,  1.1733e-02, -2.3943e-01, -3.4317e-02,
         1.7614e-02, -9.3440e-02, -1.7061e-04, -2.9764e-02, -2.1347e-02,
        -1.3528e-02, -1.5188e-01,  1.0102e-02,  2.8732e-02,  1.8632e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0713, -0.0114, -0.0316, -0.0356, -0.0352,  0.0299, -0.1282, -0.0424,
        -0.0745, -0.3178, -0.0101, -0.0262, -0.0308, -0.0262, -0.0760, -0.0529,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0236,  0.1701,  0.0166,  0.0724,  0.0375,  0.0393,  0.2083, -0.0052,
        -0.0583, -0.0076, -0.0119,  0.0060,  0.0446,  0.0220,  0.0632,  0.0191,
         0.0312,  0.0153,  0.0200,  0.0059,  0.0218, -0.0286,  0.0715,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0747,  0.0595,  0.0701,  0.1001,  0.2299,  0.0671, -0.0819, -0.0352,
         0.0222, -0.0247, -0.0046, -0.0216, -0.1637,  0.0447,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.0808,  0.0733,  0.0723,  0.1811,  0.0352, -0.0378,  0.0583, -0.0118,
         0.0456, -0.0012,  0.0403,  0.0759,  0.0230, -0.0018, -0.0087,  0.0322,
         0.0076,  0.0885,  0.0088,  0.0738,  0.0113, -0.0020,  0.0102, -0.0183,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1248, -0.1032, -0.1692, -0.0193, -0.1316, -0.0604, -0.1955, -0.0027,
        -0.1142, -0.0134, -0.0656,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0199,  0.0027,  0.0029, -0.0146, -0.0004, -0.0345, -0.1984, -0.0590,
        -0.1739, -0.0265, -0.0360,  0.0030, -0.0351, -0.0566, -0.0449, -0.0148,
        -0.0169, -0.0788,  0.0535,  0.0399,  0.0875,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0607, -0.0716, -0.0470, -0.1537, -0.0485,  0.0396,  0.0123, -0.2080,
        -0.0580, -0.0896,  0.0349,  0.0809,  0.0402, -0.0550,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0335,  0.0576,  0.0316,  0.0372,  0.0536, -0.0327,  0.0011,  0.0383,
        -0.0333,  0.0380,  0.0630,  0.1126,  0.2352,  0.0460,  0.0659,  0.0343,
        -0.0024, -0.0402, -0.0019, -0.0135, -0.0282,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0189,  0.1823, -0.0121,  0.0935,  0.0349,  0.1788,  0.0576,  0.1653,
         0.0747,  0.0679,  0.0532, -0.0322, -0.0288,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.5794e-02, -3.6727e-02,  3.0873e-03, -8.4248e-03, -7.9513e-03,
        -1.3109e-02, -1.9648e-02, -3.9539e-02, -2.7080e-02, -1.0090e-03,
        -1.9264e-02, -2.4218e-02,  4.4523e-03,  8.7234e-06, -5.1556e-03,
         5.2485e-03,  2.7187e-03,  4.6691e-03, -2.2019e-03, -1.9477e-02,
        -4.0663e-02,  2.4565e-03,  1.1011e-02, -4.1463e-02, -1.4720e-01,
         1.1983e-02, -1.6217e-01, -8.0288e-02, -6.8464e-02, -8.1851e-03,
        -6.7821e-03, -1.0657e-02,  1.8083e-02,  1.2256e-02, -2.6515e-03,
         1.2766e-02, -1.8356e-02,  3.4288e-03, -5.3111e-03, -2.8727e-03,
        -1.3748e-02, -1.6341e-04, -1.6631e-03,  4.7074e-02,  5.1309e-04],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0571, -0.2069, -0.0354,  0.0099, -0.0072,  0.0206,  0.0051, -0.0207,
        -0.0211,  0.0057,  0.0097,  0.0264,  0.0007,  0.0132, -0.0369,  0.0069,
        -0.0944,  0.0012, -0.0259, -0.0106, -0.0017, -0.0803,  0.0051, -0.0364,
         0.0141,  0.0133, -0.0235, -0.0034, -0.0556,  0.0145, -0.0049,  0.0021,
        -0.1136,  0.0160,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0979,  0.1355,  0.0663, -0.0048,  0.0109,  0.0086,  0.0071,  0.0181,
         0.0598,  0.0026,  0.0035,  0.0126, -0.0068, -0.0280,  0.0049,  0.0024,
         0.0063,  0.0125,  0.0359,  0.0953,  0.1423, -0.0191,  0.0065, -0.0340,
        -0.0579, -0.1204,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0382, -0.4302, -0.0103, -0.0419, -0.0081, -0.0285,  0.0276, -0.0372,
        -0.0427,  0.0060, -0.0245,  0.0032,  0.0158,  0.0025, -0.0595, -0.0341,
        -0.1327,  0.0038,  0.0011,  0.0021, -0.0118,  0.0217, -0.0164,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1161, -0.1195,  0.0158, -0.0219,  0.0073,  0.0179, -0.0225, -0.0383,
        -0.0636, -0.0134,  0.0177,  0.0131, -0.0126, -0.0119, -0.1351,  0.0207,
         0.0124, -0.0391, -0.0308, -0.0139,  0.0083,  0.0096,  0.0057,  0.0170,
        -0.0049, -0.0061, -0.0238, -0.0812,  0.0143,  0.0019,  0.0143,  0.0079,
         0.0012,  0.0062,  0.0065, -0.0110, -0.0059,  0.0072,  0.0095, -0.0136,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0388, -0.0932, -0.0267, -0.0307, -0.0619, -0.0187,  0.0093, -0.0825,
         0.0174, -0.0128, -0.0104, -0.0254, -0.0375,  0.0005, -0.0332, -0.0515,
        -0.0119, -0.0004, -0.0187, -0.0325,  0.0227, -0.0419, -0.0379,  0.0225,
        -0.0247,  0.0129,  0.0013, -0.0743, -0.0033, -0.0216, -0.0153, -0.0027,
         0.0816, -0.0232,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0491, -0.0583, -0.0045, -0.0200, -0.0064,  0.0182, -0.0020, -0.0218,
        -0.0016,  0.0164, -0.0121, -0.0107,  0.0124, -0.0202, -0.0774, -0.0037,
        -0.0600, -0.0047, -0.0349,  0.0017, -0.0073,  0.0009, -0.0221,  0.0096,
        -0.0251, -0.0725,  0.0383, -0.0037,  0.0283,  0.0135, -0.0165,  0.0097,
         0.0104, -0.0578, -0.0303,  0.0003, -0.0202, -0.0257, -0.0160, -0.0116,
        -0.0094, -0.0081, -0.0078, -0.0086,  0.0023,  0.0158,  0.0123,  0.0060,
        -0.0039, -0.0145, -0.0419, -0.0132], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0369, -0.0121, -0.0171,  0.0188, -0.0413, -0.0037,  0.0369,  0.0351,
        -0.0082,  0.0046,  0.0035, -0.0016, -0.0081, -0.1063, -0.0388, -0.0554,
        -0.0868, -0.1105, -0.0342, -0.0177, -0.0190, -0.0155, -0.0444, -0.0723,
         0.0007,  0.0055,  0.0240, -0.0177,  0.0408,  0.0224, -0.0603,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0125, -0.1436,  0.0248, -0.0126,  0.0143,  0.0216, -0.0129,  0.0287,
        -0.0174,  0.0070, -0.0246, -0.0411, -0.0014, -0.0663, -0.0487, -0.0316,
         0.0508, -0.0459, -0.0960, -0.1296, -0.0566, -0.0315, -0.0473,  0.0190,
         0.0142,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0989,  0.0238,  0.0592,  0.0018,  0.0174,  0.1948,  0.3204, -0.0560,
         0.0677,  0.0441,  0.0950, -0.0209,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1021, -0.2966,  0.0631,  0.0264, -0.0310, -0.0060, -0.0041, -0.0036,
        -0.0041, -0.0542,  0.0622, -0.0096, -0.0088, -0.0684, -0.0115, -0.0043,
         0.0014,  0.0025,  0.0045,  0.0099, -0.0129,  0.0105, -0.0055,  0.0100,
         0.0258, -0.0425, -0.0090,  0.0274, -0.0822,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0055, -0.3547,  0.1236,  0.0363,  0.0724, -0.0505, -0.2512,  0.0318,
         0.0506, -0.0028, -0.0029,  0.0178,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0655, -0.2827, -0.0803, -0.0516,  0.0010, -0.0928, -0.1560,  0.0419,
        -0.0396, -0.0401, -0.0829, -0.0656,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0347, -0.1394,  0.1324,  0.0673,  0.1162, -0.0223,  0.0467,  0.0251,
         0.0846,  0.0237, -0.0184,  0.0748,  0.0125,  0.0094,  0.0019,  0.1236,
         0.0024, -0.0233,  0.0322,  0.0092,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0467,  0.2830, -0.0089,  0.0150,  0.0414,  0.0146,  0.0335, -0.0384,
        -0.0011,  0.0035,  0.0119,  0.0370,  0.0285,  0.0211, -0.0228, -0.0103,
         0.0250,  0.0016, -0.0137,  0.0344,  0.0180,  0.0034, -0.0191, -0.0196,
        -0.0142, -0.0022, -0.0085,  0.0133,  0.0191, -0.0119, -0.0074, -0.0007,
         0.0086,  0.0671,  0.0372, -0.0175,  0.0398,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0972,  0.5728,  0.0529, -0.0648,  0.0541, -0.0398, -0.0269, -0.0682,
         0.0233,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0436, -0.0252, -0.0339, -0.0719, -0.0256, -0.0854, -0.0453, -0.1102,
         0.0121, -0.0104,  0.0286, -0.0025, -0.0380, -0.1517, -0.0196, -0.0008,
        -0.0147, -0.0340, -0.0395, -0.0542,  0.0087,  0.0123,  0.0060, -0.0049,
        -0.0420,  0.0456,  0.0334,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0733, -0.2288,  0.0288, -0.0239, -0.0071, -0.0323, -0.0006, -0.0049,
        -0.0134, -0.0122,  0.0080, -0.0289, -0.0017,  0.0143,  0.0044, -0.0185,
        -0.0313,  0.0076, -0.0070,  0.0138,  0.0005, -0.0519, -0.0495,  0.0057,
        -0.0141,  0.0154, -0.0028, -0.0221, -0.1896, -0.0123,  0.0033, -0.0292,
         0.0124,  0.0303,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0223, -0.0042,  0.0076, -0.0108, -0.0214, -0.0069, -0.0009,  0.0113,
         0.0193,  0.0481, -0.0299, -0.0461, -0.0412, -0.0157,  0.0142, -0.0309,
        -0.0593,  0.0286, -0.0136, -0.0157,  0.0231,  0.0031, -0.0254, -0.0165,
         0.0170,  0.0106,  0.0047,  0.0177,  0.0038,  0.0211,  0.0364, -0.0451,
        -0.0964, -0.0037, -0.0214, -0.0051, -0.0591, -0.0237,  0.0105, -0.0063,
        -0.0057, -0.0470, -0.0166,  0.0095, -0.0226,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0348, -0.0242, -0.1003, -0.0712, -0.0189, -0.0939, -0.1415,  0.0384,
        -0.0208, -0.1333, -0.0727, -0.0516, -0.1192, -0.0143, -0.0471, -0.0180,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0650, -0.0423,  0.0289, -0.0348, -0.0364,  0.0077,  0.0092,  0.0583,
         0.0970,  0.0586,  0.0121,  0.1298,  0.0107,  0.0126,  0.0464, -0.0178,
         0.0069,  0.0191, -0.0086,  0.0225,  0.0023, -0.0348,  0.0236,  0.0240,
         0.0284,  0.1360,  0.0265,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0750,  0.0646, -0.0230, -0.0146, -0.0226,  0.0016, -0.0119, -0.0099,
        -0.0207,  0.0060, -0.0025, -0.0104, -0.0157, -0.0002, -0.0091,  0.0036,
        -0.0100,  0.0108,  0.0145, -0.0422, -0.0282, -0.0253, -0.0205, -0.1039,
         0.0008, -0.0242, -0.0388, -0.0549, -0.0195, -0.0448,  0.0102, -0.0246,
        -0.0344, -0.0087, -0.0028, -0.0093, -0.0016, -0.0297, -0.0011,  0.0243,
        -0.0076, -0.0423,  0.0067, -0.0172,  0.0275,  0.0221], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2573,  0.0543,  0.0105, -0.0123,  0.0105, -0.0378,  0.0090, -0.0459,
         0.0013, -0.0194,  0.0087, -0.0057, -0.0104, -0.0044,  0.0354,  0.0065,
        -0.0118, -0.0185, -0.0505, -0.2029, -0.0158,  0.0268, -0.0140, -0.0079,
         0.0005, -0.0170, -0.0530, -0.0230,  0.0038, -0.0250,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1417, -0.0132,  0.1031,  0.0736,  0.0857,  0.0404,  0.1121,  0.1534,
         0.0699,  0.1175,  0.0894,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0303, -0.1993, -0.0601, -0.0215, -0.0687, -0.0164, -0.0203, -0.1548,
        -0.0143, -0.0432,  0.0050, -0.0174, -0.0167, -0.0311, -0.0078, -0.0721,
        -0.0155, -0.0361,  0.0290, -0.1404,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1446, -0.2409, -0.1414, -0.0544,  0.0600, -0.0314, -0.1431,  0.0476,
        -0.0234, -0.0493, -0.0639,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.7174e-02, -2.3832e-01, -1.0301e-01, -1.6256e-01,  3.7630e-02,
         1.4153e-02, -1.5567e-02, -2.9590e-02, -7.2156e-02,  2.2111e-02,
        -1.6287e-02, -3.2452e-02, -6.8760e-03,  8.3796e-03, -2.6710e-03,
        -8.1754e-02,  1.5540e-02, -1.0362e-03, -5.9229e-03, -2.1936e-05,
         9.6790e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1707, -0.1071, -0.0436, -0.0325, -0.0207, -0.2662,  0.0757, -0.0272,
        -0.0228,  0.0866, -0.0062, -0.0334, -0.0038,  0.0015, -0.0055, -0.0270,
        -0.0256, -0.0438,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0879, -0.0618, -0.0859, -0.0271, -0.0306, -0.0081,  0.0035,  0.0018,
        -0.0102, -0.0176,  0.0171,  0.0029, -0.0320, -0.0567, -0.0477, -0.0263,
        -0.0193, -0.0514, -0.0237, -0.0019, -0.0245, -0.0325, -0.0402, -0.0002,
        -0.0417,  0.0131, -0.0343, -0.0154, -0.0151, -0.0231,  0.0033, -0.0095,
         0.0021, -0.0149, -0.0122,  0.0017,  0.0061, -0.0299,  0.0666,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0468, -0.3030,  0.0009, -0.0321, -0.0056, -0.0057,  0.0049, -0.0202,
        -0.0270,  0.0058, -0.0046, -0.0861, -0.0319, -0.0150,  0.0150, -0.0173,
        -0.0276, -0.0206,  0.0064, -0.0087,  0.0160, -0.0302, -0.0100,  0.0037,
        -0.0178, -0.0006, -0.0212, -0.0236, -0.0254, -0.0654, -0.0064, -0.0047,
         0.0359, -0.0543,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.1524,  0.4035,  0.0816,  0.1147, -0.0222,  0.0354, -0.0697, -0.0314,
         0.0891,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.9432e-02, -2.1960e-01, -4.1904e-02, -4.5506e-02, -1.9749e-02,
        -3.5757e-02, -6.9621e-03,  8.0919e-03, -9.5988e-04,  2.3685e-02,
        -1.8433e-02, -3.2705e-03,  1.8402e-02,  1.2113e-03,  4.1526e-03,
        -3.4549e-02,  7.0936e-03, -1.6199e-01, -2.1751e-02,  1.7686e-04,
        -6.1467e-03,  5.2532e-03,  1.2189e-02,  8.3665e-03, -2.3299e-02,
        -1.8391e-02,  5.1758e-03,  1.7746e-02,  4.6606e-03, -4.3245e-03,
        -9.6192e-03, -9.9730e-03,  8.8092e-03, -1.0899e-03, -9.9780e-03,
         1.5136e-02,  6.4704e-03, -3.2527e-04,  7.9172e-02, -1.1988e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0255, -0.0216, -0.1635, -0.0361, -0.0162, -0.0185,  0.0064, -0.0157,
        -0.0277, -0.0812, -0.0054,  0.0045,  0.0035, -0.0021,  0.0052, -0.0018,
        -0.0136,  0.0120, -0.0408, -0.0028, -0.0083, -0.0022,  0.0201,  0.0219,
         0.0121, -0.0060, -0.0632,  0.0100, -0.0606, -0.0630, -0.0256, -0.0026,
        -0.0033, -0.0070, -0.0050, -0.0046, -0.0299, -0.0428,  0.0026,  0.0010,
         0.0156,  0.0063, -0.0007,  0.0035, -0.0110,  0.0109,  0.0059,  0.0129,
        -0.0060,  0.0089,  0.0185,  0.0037,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1224, -0.1834, -0.0550, -0.0060, -0.0754, -0.0116, -0.0368, -0.0037,
        -0.0150, -0.0244, -0.0272, -0.0369, -0.0358, -0.0119, -0.0283, -0.0422,
        -0.0255,  0.0116, -0.1426, -0.0442, -0.0039, -0.0563,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1539, -0.4318, -0.0600, -0.0075,  0.0113, -0.0168, -0.0197, -0.0161,
         0.0105,  0.0087, -0.0185, -0.0395, -0.0260, -0.0344,  0.0028, -0.0259,
        -0.0390,  0.0042,  0.0629, -0.0006, -0.0098,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0191, -0.1551, -0.0081,  0.0187, -0.0025, -0.0026, -0.0115,  0.0155,
        -0.0103, -0.0191,  0.0041, -0.0076,  0.0130,  0.0127, -0.0077, -0.0076,
         0.0102, -0.0095, -0.0329, -0.0029,  0.0008, -0.0013, -0.0047,  0.0059,
        -0.0025,  0.0180, -0.0352, -0.0258,  0.0139,  0.0032, -0.0660, -0.0492,
        -0.0096, -0.0125, -0.0087, -0.0007,  0.0211,  0.0003,  0.0021, -0.0011,
        -0.1138, -0.0049, -0.0012, -0.0227, -0.0293, -0.0055, -0.0289, -0.0326,
         0.0046,  0.0163, -0.0126,  0.0118, -0.0218,  0.0151, -0.0259],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0580, -0.0010,  0.0313, -0.0086,  0.0120, -0.0027, -0.0681, -0.0113,
        -0.0089,  0.0029, -0.0053, -0.0051, -0.0140, -0.0090, -0.0319, -0.0056,
        -0.0239, -0.0193, -0.0621, -0.0617, -0.0455, -0.0210, -0.0226, -0.0129,
         0.0135,  0.0026, -0.0192, -0.0544, -0.0186, -0.0102, -0.0249, -0.0030,
        -0.0176, -0.0599, -0.0090, -0.0224, -0.0119, -0.0312, -0.0290, -0.0413,
        -0.0140, -0.0048,  0.0214,  0.0166, -0.0296,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0337,  0.2152,  0.0941,  0.0874, -0.0608,  0.0332,  0.0171, -0.0196,
         0.0407,  0.0219,  0.0040,  0.0399,  0.0170,  0.0403,  0.0334,  0.0297,
        -0.0051,  0.0338,  0.0018,  0.0112,  0.0472,  0.0513,  0.0488,  0.0127,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0467, -0.0074, -0.0005,  0.0212,  0.0130,  0.0283, -0.0128, -0.0124,
         0.0083, -0.0086,  0.0042,  0.0055, -0.0640, -0.0574,  0.0034, -0.0039,
        -0.1072, -0.0358, -0.0438, -0.0694, -0.0396,  0.0184, -0.0472, -0.0820,
         0.0340, -0.0421, -0.0467, -0.0011, -0.0261, -0.0711, -0.0156, -0.0152,
         0.0071,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0233, -0.2903, -0.0101, -0.0741, -0.0234, -0.0025, -0.0493,  0.0195,
        -0.2711, -0.0284, -0.0071,  0.0194,  0.0291, -0.0567,  0.0029, -0.0428,
        -0.0336,  0.0126, -0.0038,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0719,  0.0177,  0.0233, -0.0182, -0.0906, -0.0059, -0.0832, -0.0556,
        -0.1258,  0.0090, -0.0200, -0.0252, -0.0123,  0.0135,  0.0095, -0.0186,
        -0.0055, -0.0516, -0.0157,  0.0106, -0.0296, -0.1787, -0.0571,  0.0253,
        -0.0256,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0194, -0.3182, -0.0496, -0.0497, -0.0877, -0.0554, -0.1419, -0.0020,
        -0.0625, -0.0501,  0.0234, -0.0179, -0.0237, -0.0096, -0.0223,  0.0474,
        -0.0193,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 9.5514e-02, -2.7878e-03, -2.4268e-02,  5.9698e-04, -5.0024e-03,
         9.0504e-03,  1.2334e-03, -1.8140e-02,  6.3243e-03, -4.8930e-02,
        -3.1321e-02, -1.1034e-02,  2.5180e-03, -5.5503e-03,  1.0043e-03,
         7.0244e-03, -4.7028e-02, -1.5303e-01, -3.5485e-03, -4.9676e-02,
        -2.3716e-02, -8.1008e-02, -1.3500e-01, -1.0192e-03,  1.4677e-04,
        -6.9751e-03, -6.0364e-02, -4.8430e-02, -4.1130e-02, -2.0079e-02,
        -4.6599e-03,  5.3888e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0065,  0.0851,  0.0636, -0.0040,  0.0149, -0.0013,  0.0389,  0.2077,
        -0.0081,  0.0019, -0.0009,  0.0013, -0.0281, -0.0202,  0.0312, -0.0251,
         0.0198,  0.0062,  0.0804,  0.0499,  0.0140,  0.0050,  0.0730,  0.0841,
         0.0125,  0.0123, -0.0252,  0.0789,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.8396e-02,  9.1155e-02,  8.3940e-05,  2.9897e-02,  3.5585e-03,
         3.4085e-02,  1.8424e-01,  3.3282e-02,  4.2669e-02,  8.6594e-02,
        -1.2928e-02,  1.2731e-02,  1.2487e-02,  5.3359e-02,  7.8129e-03,
         6.7510e-02,  1.4849e-02,  7.2907e-02,  1.1272e-03,  1.4899e-03,
         1.5580e-02,  3.9099e-03,  1.2086e-02, -1.0127e-03,  3.4364e-02,
         8.6623e-03, -1.2147e-04, -4.2647e-02, -3.0454e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0475, -0.0340, -0.0122, -0.0336, -0.1301,  0.0155, -0.0184,  0.0274,
        -0.0084,  0.0066,  0.0114, -0.0257, -0.0164, -0.1076,  0.0286, -0.0462,
         0.0098, -0.1643, -0.0141, -0.0517,  0.0290,  0.0061, -0.0258, -0.0172,
        -0.0061, -0.0025, -0.0163, -0.0019, -0.0022,  0.0008, -0.0003,  0.0183,
        -0.0640,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0428, -0.2653, -0.0145, -0.0055, -0.0402, -0.0413, -0.0248, -0.0337,
        -0.0026, -0.0435, -0.0071, -0.0023, -0.0070, -0.0208, -0.0144,  0.0082,
        -0.0481, -0.0125, -0.1426, -0.0062, -0.0768, -0.0501,  0.0317, -0.0580,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0387, -0.1587, -0.1494, -0.0114, -0.0275, -0.0165, -0.0390, -0.0588,
         0.0060, -0.0537, -0.0248, -0.0368, -0.0166, -0.0259, -0.0177,  0.0020,
         0.0008, -0.0250, -0.0088, -0.1726, -0.0547, -0.0056,  0.0488,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0391, -0.1157,  0.0003, -0.0044, -0.0107, -0.0352,  0.0176,  0.0028,
        -0.0290, -0.0380, -0.0175, -0.0146, -0.0267, -0.0057,  0.0085, -0.0283,
        -0.0510, -0.0250, -0.0070,  0.0425, -0.0030, -0.0326, -0.0011, -0.0056,
        -0.0623, -0.0320, -0.0096, -0.0147, -0.0254, -0.0571, -0.0670, -0.0135,
        -0.0231, -0.0187, -0.0209, -0.0054,  0.0023, -0.0590, -0.0272,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0536, -0.0926, -0.0309, -0.0081, -0.1681, -0.0095, -0.0050, -0.0034,
        -0.0043,  0.0158,  0.0015, -0.0015,  0.0083, -0.0038,  0.0161, -0.0221,
        -0.0116,  0.0100,  0.0031, -0.0188, -0.0265, -0.0117,  0.0059, -0.0091,
        -0.1152, -0.0775,  0.0057,  0.0018, -0.0014, -0.0073, -0.0154, -0.0216,
        -0.0342, -0.0065, -0.0048,  0.0049, -0.0123, -0.0725, -0.0108,  0.0100,
        -0.0034, -0.0037, -0.0496,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.2919e-02, -1.8437e-01, -1.1969e-03, -1.0669e-03, -5.0828e-03,
        -1.1351e-02, -3.0169e-02, -1.4882e-02, -6.4680e-02, -1.2688e-02,
         2.6810e-03, -3.0968e-03, -1.3499e-02,  3.4823e-03, -4.0508e-03,
        -8.1838e-03, -1.8059e-02, -4.8507e-02,  6.0387e-05, -1.9417e-03,
         1.0631e-02,  2.0198e-03, -1.2808e-02, -1.1540e-03, -6.4281e-03,
         9.5084e-03, -7.5349e-04,  2.2148e-05, -1.3153e-02, -1.4605e-02,
        -1.9556e-02, -9.5970e-03, -2.6576e-02,  1.0379e-02, -6.5237e-03,
        -6.4336e-03, -1.3723e-02, -4.2599e-03, -4.8824e-02, -2.1911e-02,
        -2.5445e-02, -6.0527e-02, -1.1596e-02, -9.5179e-03, -4.6460e-03,
         1.3103e-03, -7.2943e-03, -7.7588e-03, -1.7242e-02, -6.3328e-03,
        -1.3076e-02, -7.8328e-03,  1.9546e-02, -2.2896e-03, -8.7349e-03,
         1.9275e-02, -6.6747e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0474, -0.3248, -0.0753, -0.0281, -0.0596, -0.0392, -0.0496, -0.0409,
         0.0842, -0.0517, -0.0932, -0.0480, -0.0548, -0.0033,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1071,  0.1367,  0.0405,  0.0173, -0.0134, -0.0090,  0.0106,  0.0128,
         0.0205,  0.0064,  0.0099,  0.1557,  0.0317,  0.0176,  0.0170,  0.0255,
         0.0123,  0.0310, -0.0082,  0.0529,  0.0552,  0.0164,  0.0225, -0.0034,
        -0.0133, -0.0268, -0.0085, -0.0180,  0.0612, -0.0385,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([0.0195, 0.1127, 0.1886, 0.2778, 0.2329, 0.0708, 0.0919, 0.0058, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0232, -0.0755, -0.0251, -0.0408, -0.0523, -0.0088, -0.0048,  0.0064,
         0.0018, -0.0160, -0.0468, -0.0487, -0.0846, -0.1157,  0.0033, -0.0055,
         0.0092,  0.0012,  0.0095, -0.0097, -0.0036,  0.0006, -0.0151,  0.0050,
        -0.0411, -0.0189, -0.0366, -0.0040, -0.0561, -0.1649, -0.0265, -0.0276,
         0.0110,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0107, -0.1301, -0.0428,  0.0093, -0.0026, -0.0033,  0.0003, -0.0227,
         0.0161, -0.0058,  0.0029,  0.0082, -0.0066,  0.0004,  0.0027,  0.0053,
         0.0042,  0.0065, -0.0052,  0.0022, -0.0006,  0.0075, -0.0033,  0.0038,
         0.0177, -0.0110, -0.0229, -0.0707,  0.0100, -0.0120, -0.0093, -0.0133,
        -0.0203, -0.0139, -0.0212,  0.0110, -0.0310,  0.0036, -0.0056,  0.0041,
         0.0026,  0.0071, -0.0033, -0.0399,  0.0149, -0.0088, -0.0041,  0.0003,
        -0.0113, -0.0041,  0.0010, -0.0108, -0.0039, -0.0107, -0.0102,  0.0055,
        -0.0054,  0.0093, -0.0090, -0.0174, -0.0084, -0.0054, -0.1346, -0.0163,
        -0.0036, -0.0063, -0.0080,  0.0039, -0.0025,  0.0265,  0.0241],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0320, -0.0047, -0.0054,  0.0013, -0.0200, -0.0548, -0.1433, -0.0294,
        -0.0055,  0.0296,  0.0024, -0.0356,  0.0050,  0.0014,  0.0045,  0.0131,
        -0.0018,  0.0011, -0.0194, -0.0607, -0.0002,  0.0269,  0.0270, -0.0739,
         0.0227, -0.0329, -0.0047,  0.0020,  0.0065, -0.0054, -0.0086, -0.0283,
        -0.0513, -0.0787,  0.0036,  0.0025, -0.0710, -0.0305, -0.0036, -0.0055,
         0.0358,  0.0075,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1426, -0.1675,  0.0101,  0.0398,  0.0326, -0.0161,  0.0084, -0.0397,
         0.0201, -0.0295, -0.0365, -0.0191, -0.0238, -0.0177,  0.0250,  0.0094,
         0.0023, -0.0313, -0.0623, -0.0161, -0.0052, -0.0060, -0.0008, -0.0042,
         0.0113, -0.0337, -0.0254, -0.0086, -0.0242,  0.0016, -0.0077,  0.0058,
        -0.0117, -0.0149, -0.0162,  0.0654,  0.0076,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0506,  0.5329, -0.0080,  0.0695,  0.0457, -0.0060,  0.0211,  0.0091,
        -0.0642,  0.0249, -0.0020,  0.0358,  0.0545, -0.0036, -0.0720,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0927, -0.1004,  0.0495,  0.0268, -0.0205, -0.1209, -0.0466, -0.0321,
        -0.1850, -0.0400,  0.0124, -0.0125, -0.0150, -0.0146,  0.0160,  0.0266,
        -0.0250, -0.1359,  0.0219, -0.0054,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0201,  0.0092, -0.0300, -0.0011, -0.0598, -0.0230, -0.0190, -0.0107,
        -0.0449, -0.0333, -0.0326, -0.0686, -0.0173, -0.0181, -0.0272, -0.0247,
        -0.0137, -0.0308,  0.0061, -0.0359, -0.0050, -0.0052,  0.0012,  0.0046,
         0.0280, -0.0212, -0.1066, -0.0051, -0.0132, -0.0106, -0.0661, -0.0179,
        -0.0100, -0.0267, -0.0166, -0.0485,  0.0114, -0.0036, -0.0133,  0.0039,
         0.0498,  0.0048,  0.0008,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0080, -0.0014,  0.0011, -0.0219, -0.0277, -0.0439, -0.0449, -0.1539,
        -0.0380, -0.0186, -0.0408,  0.0037, -0.0360, -0.0390, -0.0960, -0.0118,
        -0.0183,  0.0057, -0.0044, -0.0420, -0.0693, -0.0072, -0.0076,  0.0379,
        -0.0160, -0.0006,  0.0146, -0.0158, -0.0441, -0.0133,  0.0060, -0.0188,
        -0.0057, -0.0206, -0.0321, -0.0054, -0.0261,  0.0018,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0780, -0.1357, -0.0762, -0.1530, -0.0229, -0.0913,  0.0286,  0.0332,
        -0.0047, -0.0964, -0.0088, -0.0173, -0.0493, -0.0230,  0.0112,  0.0163,
        -0.0175, -0.0357,  0.0064,  0.0232,  0.0378,  0.0334,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0037,  0.1109,  0.0160, -0.0084,  0.0533,  0.0068,  0.0053,  0.0162,
         0.0365,  0.0725,  0.0191,  0.0099,  0.0183, -0.0109,  0.0247,  0.0246,
         0.0927,  0.0217,  0.0259,  0.0456,  0.0182, -0.0138,  0.0266,  0.0332,
        -0.0027,  0.0014,  0.0032,  0.0040, -0.0058,  0.0190,  0.0278, -0.0196,
         0.0068,  0.0130,  0.0219,  0.0670, -0.0155,  0.0058, -0.0174,  0.0545,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0700, -0.0228, -0.0070, -0.0002, -0.0078,  0.0173,  0.0034, -0.0420,
        -0.0406, -0.0212, -0.0080, -0.0090, -0.0008, -0.0987, -0.0597, -0.0058,
         0.0059,  0.0172, -0.0343, -0.0070, -0.0290,  0.0026, -0.0409,  0.0162,
         0.0122, -0.1171, -0.0069, -0.0741, -0.1130, -0.0302, -0.0470,  0.0319,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0315, -0.3202, -0.0046,  0.0262, -0.0115,  0.0125,  0.0038, -0.0326,
        -0.0737, -0.0176,  0.0210, -0.0352, -0.0432,  0.0022,  0.0017, -0.0056,
        -0.0485, -0.0016,  0.0092,  0.0111,  0.0050, -0.0253, -0.0015, -0.0625,
        -0.0041,  0.0205, -0.0267, -0.0043,  0.0047, -0.0319,  0.0024, -0.0240,
        -0.0070, -0.0366, -0.0300,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0790, -0.0489, -0.1153, -0.0827, -0.0336, -0.0020, -0.1196, -0.1380,
         0.0008, -0.0976, -0.0183, -0.0279,  0.0207, -0.0605, -0.0265,  0.0241,
        -0.0080, -0.0200,  0.0164,  0.0411, -0.0190,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0006, -0.3622,  0.0095, -0.0897, -0.0704, -0.0533, -0.0305, -0.0878,
        -0.1031,  0.0196, -0.0043,  0.0322, -0.0073,  0.0383, -0.0285, -0.0139,
        -0.0114,  0.0084, -0.0291,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1918, -0.2152, -0.0652, -0.0135, -0.0172, -0.0119,  0.0168,  0.0204,
        -0.0305, -0.0277, -0.1220,  0.0004, -0.0222, -0.1077, -0.0148, -0.0508,
        -0.0045, -0.0313,  0.0010, -0.0350,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0456, -0.3141, -0.0352, -0.0682, -0.0291, -0.0349, -0.0065, -0.0335,
        -0.0330, -0.0389, -0.0123,  0.0008,  0.0194, -0.0270, -0.0455, -0.0243,
        -0.0524,  0.0044, -0.0102, -0.0356,  0.0058,  0.0547,  0.0686,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0138, -0.2011, -0.0378, -0.0248,  0.0337, -0.0137, -0.0038, -0.0183,
        -0.0566, -0.1120, -0.0328, -0.0853, -0.0214, -0.1173,  0.0267, -0.0063,
         0.0149,  0.0225, -0.0074,  0.0532, -0.0385,  0.0120, -0.0113, -0.0223,
        -0.0126,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.5552e-02,  3.4343e-01, -3.4057e-03,  1.4576e-02, -9.4072e-04,
         1.3968e-02, -2.1279e-02, -4.1422e-03, -1.2463e-02, -2.2629e-02,
         6.4369e-02,  1.1696e-02,  1.3503e-03,  1.4701e-02, -3.1097e-03,
        -5.1977e-03,  2.8763e-03,  2.3034e-04, -2.2866e-02,  2.5938e-02,
         4.9183e-02,  5.7959e-02,  4.3777e-02,  2.5296e-02,  1.1977e-02,
         1.4034e-02,  1.7657e-02, -1.6491e-02, -2.8665e-03, -2.2449e-02,
        -5.1663e-03, -2.6618e-02, -5.2523e-03, -1.3219e-03, -5.5233e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0284,  0.0290,  0.0332, -0.1222,  0.0155,  0.0066, -0.0043, -0.0260,
        -0.1187,  0.0168,  0.0772,  0.0415,  0.0748, -0.0313, -0.0335, -0.0772,
        -0.0033, -0.1160, -0.0594,  0.0056, -0.0076, -0.0719,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0795, -0.0670,  0.0210, -0.0289,  0.0757, -0.0437,  0.0081, -0.0219,
        -0.0378, -0.1329, -0.0030, -0.0457,  0.0102, -0.0306,  0.0410, -0.0645,
        -0.1180,  0.0290, -0.0598,  0.0374, -0.0441,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0215, -0.2965, -0.0443,  0.0209,  0.0012, -0.0088, -0.0706, -0.1581,
        -0.0171,  0.0204, -0.0750, -0.0233, -0.0131, -0.0698, -0.0238, -0.0601,
        -0.0125,  0.0141,  0.0488,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1203,  0.0012, -0.0190, -0.0034,  0.0022, -0.0783, -0.0156, -0.0058,
         0.0125, -0.0047, -0.0144, -0.0100,  0.0021, -0.0163,  0.0082,  0.0014,
        -0.0102, -0.0386,  0.0014, -0.0162, -0.0153, -0.0339, -0.0162,  0.0032,
        -0.0248, -0.0353,  0.0213, -0.0587, -0.0467, -0.1284, -0.0201, -0.0324,
        -0.0164, -0.0544, -0.0005, -0.0088,  0.0190,  0.0049,  0.0283, -0.0095,
         0.0158,  0.0006,  0.0049,  0.0048, -0.0008, -0.0013,  0.0121],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.1985e-02,  2.6967e-01,  3.8029e-02,  8.8121e-03, -1.5859e-04,
        -2.1762e-03, -4.9301e-04, -3.6083e-02,  4.0433e-02, -2.0483e-02,
         6.7357e-03, -5.5299e-03, -9.6550e-03,  1.0630e-02,  2.1188e-02,
         9.9385e-05,  8.0908e-02,  2.7461e-03,  5.1669e-03,  1.8159e-04,
         2.8590e-03,  3.9582e-02,  6.6218e-02,  8.1616e-02,  9.8838e-04,
        -3.0475e-02, -2.2157e-02,  3.4408e-03,  1.6680e-02,  2.4543e-02,
         2.6373e-02, -1.4226e-02,  3.6684e-02,  2.2995e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0013, -0.3148, -0.0257,  0.0169, -0.0146, -0.0407, -0.0155, -0.0013,
        -0.0108, -0.0128,  0.0111, -0.0098, -0.0251, -0.0904,  0.0071,  0.0038,
         0.0009,  0.0323,  0.0386, -0.1198,  0.0136, -0.0172, -0.0177, -0.0294,
        -0.0033, -0.0055, -0.0144, -0.0191,  0.0018, -0.0161,  0.0044,  0.0160,
        -0.0307, -0.0172,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.1186, -0.2620, -0.0395,  0.0153, -0.0012, -0.0779,  0.0270,  0.0385,
         0.0271, -0.0729, -0.0617, -0.0187, -0.0301, -0.1242, -0.0099, -0.0331,
         0.0343,  0.0079,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0939,  0.0092,  0.0586, -0.0355,  0.0263, -0.0268,  0.0289,  0.0238,
         0.0649,  0.2273, -0.0121, -0.0023,  0.0057, -0.0016, -0.0005,  0.0285,
         0.0066,  0.0189,  0.0371,  0.1280,  0.0022,  0.0123,  0.0375, -0.0126,
         0.0062, -0.0171, -0.0757,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0500,  0.0485,  0.0128,  0.0024,  0.0378, -0.0203,  0.0205,  0.0402,
         0.1052,  0.0178,  0.0172, -0.0196,  0.2194,  0.0757, -0.0018, -0.0034,
        -0.0182,  0.0228, -0.0079,  0.0753,  0.0169,  0.0299,  0.0362, -0.0003,
         0.0526,  0.0472,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1964,  0.2656,  0.0255,  0.0059,  0.0977,  0.0216,  0.0142,  0.0304,
         0.0140,  0.0369,  0.0238,  0.0221, -0.0018, -0.0019, -0.0096,  0.0644,
         0.0200,  0.0077,  0.0260,  0.0431, -0.0711,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0813, -0.2462, -0.1121, -0.1009,  0.0146, -0.0585, -0.1066, -0.0181,
        -0.0374,  0.0254, -0.0262, -0.0577, -0.0446,  0.0447, -0.0256,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1198, -0.0131,  0.0207, -0.0022, -0.0405,  0.1122,  0.0526,  0.0739,
        -0.0129,  0.0291,  0.0560,  0.0237,  0.1780,  0.0263, -0.0045, -0.0284,
         0.1366,  0.0698,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0481, -0.2170, -0.0504, -0.0757, -0.0047, -0.0197, -0.0846, -0.0535,
        -0.0028, -0.0950, -0.0875, -0.0558,  0.0232, -0.0595, -0.0029,  0.0088,
        -0.0237, -0.0192,  0.0679,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0388, -0.0868, -0.0515, -0.0075, -0.0124, -0.0094,  0.0340,  0.0134,
        -0.0045, -0.0050, -0.0020,  0.0102, -0.0125,  0.0074,  0.0006, -0.0090,
        -0.0809, -0.0024, -0.0111,  0.0258, -0.0476, -0.0076, -0.0582, -0.0249,
        -0.0419, -0.0408, -0.0479,  0.0134, -0.0252, -0.0150, -0.0079, -0.0338,
        -0.0259,  0.0112,  0.0009, -0.0037, -0.0462,  0.0062, -0.0027, -0.0140,
         0.0102,  0.0207, -0.0264, -0.0424], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.7795e-02, -2.1823e-01, -4.8938e-02, -2.8757e-02, -1.2513e-02,
         1.8870e-02, -4.3038e-02,  2.5904e-03, -6.3780e-02,  1.7991e-02,
         1.0973e-03,  1.8226e-02,  1.2935e-02,  9.8876e-03,  3.3258e-02,
        -1.3228e-02, -7.7885e-03, -8.5323e-02,  2.6828e-02, -1.1436e-02,
        -2.5343e-02, -4.8160e-05,  1.4798e-02, -1.9558e-02, -2.9937e-03,
         1.2933e-02,  1.6830e-02, -6.5730e-02,  2.6827e-02, -4.8813e-03,
        -1.7432e-02, -1.1104e-02,  5.1962e-03, -2.7515e-02,  3.6299e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0311, -0.3046, -0.0328, -0.0354, -0.0060, -0.0137, -0.0926, -0.0456,
         0.0250, -0.0008,  0.0303, -0.0280, -0.0420, -0.0041,  0.0328, -0.0550,
        -0.0121, -0.0333, -0.0443,  0.0087, -0.0215, -0.0104,  0.0328, -0.0572,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.0762e-01, -1.4318e-01, -2.1616e-03, -3.3426e-02, -1.5645e-03,
         1.2059e-02,  1.1866e-03, -3.4308e-02, -8.1430e-02, -2.5344e-02,
        -5.9322e-03, -1.7106e-02, -6.5048e-05, -2.1380e-02, -8.4561e-02,
        -9.9262e-03,  9.8907e-03, -3.3046e-03, -9.1078e-02,  1.4350e-02,
         6.2960e-03, -1.8199e-02,  1.7746e-02, -1.3783e-02, -7.7866e-03,
        -3.5016e-02, -1.8992e-02, -2.3345e-02, -2.0386e-02,  4.6650e-03,
        -7.5011e-02, -1.7392e-02,  1.3592e-02, -2.7919e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.8507e-02,  1.0321e-02,  2.8938e-02, -5.5474e-02,  2.1592e-02,
         1.1198e-03, -1.3024e-01, -3.8462e-02,  1.8545e-02, -1.9513e-02,
        -2.5123e-03,  1.1718e-02, -9.8127e-02, -1.5657e-02, -8.0505e-02,
        -1.3668e-01, -2.0244e-02, -7.7450e-03,  9.8518e-03, -2.3010e-03,
        -1.2771e-04,  2.8123e-03,  1.4495e-02, -5.1892e-02, -7.2267e-03,
        -1.1470e-03, -2.1589e-02, -4.2397e-02, -6.5960e-02, -2.4310e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0592,  0.0393,  0.1448,  0.2571,  0.0305,  0.0859,  0.0052,  0.0446,
         0.0276,  0.1239,  0.0308, -0.0766, -0.0745,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0222, -0.0254,  0.1111,  0.0898, -0.0237, -0.0566,  0.0091, -0.1096,
        -0.1580, -0.0216,  0.0032, -0.0798, -0.0116, -0.0716, -0.0889, -0.0345,
         0.0477,  0.0353,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0460,  0.1915,  0.2571,  0.1641,  0.2044, -0.0678,  0.0588, -0.0104,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1537e-01, -2.0759e-01, -9.3338e-03,  2.2553e-02, -1.5671e-02,
        -3.3603e-02, -1.1477e-02, -2.9763e-02,  2.6054e-02,  2.8492e-02,
         4.2150e-03,  2.6763e-02, -1.0767e-02,  8.8889e-03, -5.9259e-03,
        -4.3290e-04, -1.2033e-03,  5.3331e-04, -3.3556e-03, -5.0740e-04,
        -8.9203e-04, -1.5072e-02, -3.7969e-02, -7.8037e-03,  3.4341e-04,
        -2.8728e-02,  5.6641e-04, -8.7729e-03, -5.9024e-02, -2.6113e-02,
        -5.8788e-03, -3.9766e-02, -2.6090e-02, -1.9610e-02, -3.3730e-03,
        -2.9746e-02, -2.4155e-02, -1.9017e-02,  1.1499e-02,  8.0599e-04,
        -2.4858e-03, -1.2991e-04, -6.3494e-03,  4.9196e-03,  2.0042e-02,
        -2.4991e-02, -1.3350e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1332, -0.1871, -0.2450, -0.1691, -0.0686,  0.0264,  0.0496,  0.0191,
        -0.0807,  0.0017, -0.0195,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0465,  0.2238,  0.0430,  0.0083,  0.0194,  0.0736,  0.0677,  0.1352,
         0.0509,  0.0459,  0.0243,  0.0144,  0.0561, -0.0216,  0.0441, -0.0252,
        -0.0039, -0.0961,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0439,  0.0408,  0.3912, -0.0273,  0.0140,  0.0044,  0.0502,  0.0348,
         0.0588,  0.0502,  0.0139,  0.0368, -0.0286, -0.0320, -0.1730,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0432, -0.4471, -0.0547, -0.0064, -0.0334, -0.0374, -0.1299, -0.0202,
        -0.0166, -0.0342, -0.0973, -0.0135,  0.0011, -0.0185,  0.0466,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0143, -0.0152, -0.0212,  0.0087,  0.0459, -0.0274, -0.0296, -0.0762,
        -0.0733, -0.0181, -0.0180, -0.0232, -0.1088,  0.0355, -0.0423, -0.0742,
        -0.0702, -0.0178,  0.0021,  0.0132, -0.0431, -0.1173, -0.0249,  0.0365,
        -0.0312, -0.0120,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0899,  0.0466,  0.0641,  0.0012,  0.0114, -0.0161, -0.0106,  0.1690,
         0.2724,  0.0718, -0.0365, -0.0340,  0.0800, -0.0844,  0.0120,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0498, -0.2043, -0.0245,  0.0038, -0.0774, -0.0204, -0.0392,  0.0210,
        -0.0159, -0.1656,  0.0223, -0.0028, -0.0309,  0.0068, -0.0125, -0.0026,
         0.0033,  0.0223, -0.0457, -0.0309, -0.0202,  0.0244,  0.0258,  0.0331,
        -0.0078,  0.0868,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0074, -0.0092,  0.0232, -0.0165, -0.0453, -0.0414,  0.0150, -0.0186,
        -0.0153, -0.0480, -0.2200, -0.0181, -0.0328, -0.0748, -0.0621, -0.0458,
         0.0112, -0.0399,  0.0377, -0.0515, -0.0100, -0.0331, -0.0221, -0.0157,
         0.0483, -0.0368,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0556, -0.2641, -0.0053,  0.0057,  0.0032,  0.0293, -0.0085,  0.0098,
        -0.0153, -0.0397, -0.0311, -0.0181, -0.0545, -0.0087, -0.0185,  0.0073,
        -0.0069, -0.0116, -0.0391,  0.0053, -0.0052, -0.0241, -0.0154, -0.0483,
        -0.0615, -0.0160, -0.0238,  0.0007, -0.0016,  0.0018, -0.0143, -0.0286,
         0.0245,  0.0117, -0.0377,  0.0471,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0395, -0.1051, -0.0708, -0.0188, -0.0481, -0.0295, -0.0539, -0.0773,
        -0.0027,  0.0350, -0.0101, -0.0035,  0.0071,  0.0203, -0.0083, -0.0062,
         0.0044, -0.0036,  0.0071,  0.0148, -0.0027, -0.0306, -0.0237, -0.0474,
        -0.0368, -0.0538, -0.0095, -0.0199, -0.0532,  0.0166,  0.0052, -0.0322,
        -0.0541,  0.0116,  0.0027,  0.0336,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1408,  0.2831, -0.0139, -0.0009, -0.0264,  0.0188,  0.0373, -0.0625,
        -0.0071, -0.0406,  0.0281,  0.0009,  0.0391, -0.0111, -0.1063, -0.0424,
         0.0361,  0.0247,  0.0007,  0.0792,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.2855e-01, -6.1911e-02,  2.9539e-03, -1.9971e-02, -3.7202e-03,
         6.2371e-04, -4.9156e-02, -4.2276e-02, -8.1690e-03, -2.8612e-02,
         1.6027e-03, -2.9711e-02, -8.2617e-03, -1.3550e-02, -7.8382e-02,
        -2.9589e-02, -1.2785e-02, -3.6144e-02,  4.6611e-04, -5.1714e-04,
         1.6863e-02, -9.2215e-03,  3.4371e-03,  3.0445e-03, -5.7390e-02,
        -1.7270e-04,  7.9476e-03, -4.0642e-03,  8.2516e-03, -4.8163e-02,
        -1.5769e-02, -4.0186e-02, -8.6368e-03, -5.5736e-02,  5.8434e-03,
         1.3051e-02, -3.7486e-03, -4.1525e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0911, -0.0633,  0.0191, -0.0554, -0.0110, -0.0080, -0.0027, -0.0024,
        -0.0278, -0.0082, -0.0424, -0.0994,  0.0142,  0.0075,  0.0530, -0.0920,
        -0.0165, -0.0820, -0.0246, -0.0030, -0.0076,  0.0052,  0.0058,  0.0136,
         0.0158, -0.0150, -0.0032, -0.0092, -0.0659,  0.0058, -0.0068, -0.0029,
         0.0313,  0.0044, -0.0237,  0.0087, -0.0103,  0.0125, -0.0058,  0.0195,
         0.0035], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0756, -0.3782, -0.0070, -0.0405, -0.0202, -0.0175, -0.0291, -0.0275,
        -0.0260, -0.0071, -0.0227, -0.0180, -0.0438, -0.0604,  0.0198, -0.0632,
        -0.0299, -0.0452, -0.0074, -0.0271, -0.0042, -0.0295,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0839,  0.0053,  0.0099, -0.0073,  0.0064, -0.0436, -0.0259,  0.0037,
        -0.0268,  0.0051,  0.0049, -0.0712, -0.0229, -0.0266,  0.0195, -0.0544,
        -0.0027, -0.0302, -0.0100, -0.0626, -0.1662,  0.0297, -0.0107, -0.0311,
        -0.0153, -0.0011,  0.0048, -0.0159, -0.1022,  0.0080, -0.0093, -0.0184,
        -0.0373,  0.0098,  0.0175,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1209,  0.0366,  0.0692,  0.0780,  0.0379,  0.0233,  0.0088, -0.0161,
         0.0696,  0.0114,  0.0279,  0.0179,  0.0173,  0.0112, -0.0093,  0.0443,
        -0.0437,  0.0514,  0.0792,  0.0048, -0.0031,  0.0149, -0.0049,  0.0241,
         0.0222,  0.0078,  0.0192, -0.0076, -0.1174,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0529, -0.2991, -0.0859, -0.0779, -0.0372, -0.0645, -0.0229,  0.0374,
         0.0348,  0.0016, -0.0090, -0.0013, -0.0024, -0.0122,  0.0011, -0.0574,
        -0.0135,  0.0184,  0.0013, -0.0132,  0.0503, -0.0249, -0.0243, -0.0046,
         0.0517,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0390,  0.0457, -0.0408, -0.0076, -0.1426,  0.0110, -0.0325, -0.0646,
        -0.0551,  0.0175,  0.0342,  0.0219, -0.0049,  0.0386, -0.0605, -0.1177,
        -0.0254, -0.0178, -0.0265, -0.0174,  0.0063, -0.0971,  0.0752,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0856,  0.0039,  0.0056,  0.0209,  0.0354, -0.0058,  0.0051, -0.0070,
        -0.0076, -0.0010,  0.0002, -0.0083, -0.0182, -0.1739,  0.0011, -0.0273,
        -0.0025, -0.0205,  0.0044,  0.0076,  0.0080,  0.0117, -0.0101,  0.0212,
        -0.0215, -0.0239, -0.0860, -0.0103, -0.0363, -0.0173, -0.0150, -0.0243,
        -0.0437, -0.0146, -0.0167,  0.0450, -0.0360,  0.0647, -0.0521,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0354,  0.0077,  0.1244, -0.1208, -0.2956, -0.0116, -0.0042, -0.0434,
        -0.1327,  0.0972,  0.0742,  0.0529,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 1.6744e-01, -7.8908e-02, -2.4357e-01, -8.1173e-03, -2.2309e-03,
        -4.8726e-03, -2.2458e-02,  4.7824e-03,  2.6721e-03, -7.7046e-03,
         2.3975e-03, -6.8439e-04, -1.8545e-03,  7.2230e-03, -9.1382e-03,
         4.6094e-04,  3.6845e-03, -4.2435e-02,  7.1880e-04, -9.9867e-03,
         3.9722e-04, -3.7883e-02, -6.1021e-03, -6.9300e-03, -4.1353e-03,
         6.0340e-04,  3.5189e-02,  1.5924e-02, -7.7956e-02,  1.9168e-02,
        -2.3954e-04,  8.0059e-03, -3.4231e-02, -1.7611e-02, -1.6569e-02,
        -1.3235e-02, -2.6690e-02, -8.0385e-03, -1.0724e-03, -2.0254e-03,
         4.6932e-03,  7.4522e-03,  5.0939e-03, -4.4197e-03,  2.5006e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1520, -0.3272, -0.0426, -0.0629,  0.0764,  0.0579,  0.1041, -0.0298,
        -0.0004,  0.0312, -0.0389, -0.0766,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0094,  0.5507,  0.0386,  0.0892,  0.1036,  0.1095,  0.0695, -0.0295,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0187, -0.0855, -0.0334,  0.0002, -0.0200, -0.0107, -0.0113, -0.0145,
        -0.0063, -0.0206, -0.0197, -0.0162,  0.0004, -0.0061, -0.0087,  0.0012,
         0.0032, -0.0382, -0.0292, -0.0189, -0.0126,  0.0015, -0.0092, -0.0162,
        -0.0079, -0.0097,  0.0080, -0.0142, -0.0270,  0.0258, -0.0145, -0.0069,
         0.0002,  0.0007,  0.0085, -0.0066,  0.0064,  0.0081, -0.0025,  0.0092,
        -0.0072, -0.0113, -0.0116, -0.0016,  0.0137,  0.0273, -0.0669, -0.0241,
        -0.0938, -0.0472, -0.0040,  0.0021, -0.0243, -0.0131, -0.0142,  0.0073,
        -0.0097,  0.0040, -0.0182, -0.0074,  0.0325], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0817, -0.0537,  0.0190, -0.0310, -0.1025,  0.0594, -0.0254, -0.0230,
        -0.0182,  0.0080,  0.0039, -0.0065, -0.0020, -0.0167, -0.0942,  0.0026,
         0.0037,  0.0022, -0.0427, -0.0317, -0.0716, -0.0621, -0.0304, -0.0087,
        -0.0506, -0.0279, -0.0184, -0.0112, -0.0224, -0.0277,  0.0104, -0.0223,
         0.0083,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0666, -0.1071, -0.1659, -0.0697, -0.0177, -0.0187, -0.0721, -0.0072,
         0.0029, -0.0311,  0.0073, -0.0086, -0.0149, -0.0048, -0.0145,  0.0337,
        -0.0097, -0.0458, -0.1870, -0.0055,  0.0460,  0.0103,  0.0461, -0.0066,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0661, -0.0716, -0.0386, -0.0253, -0.0268, -0.1452, -0.0128, -0.0291,
         0.0227, -0.0801, -0.0775, -0.0858, -0.0330, -0.0047, -0.0263, -0.0048,
         0.0209, -0.0137,  0.0459,  0.0192, -0.0096,  0.0144,  0.1061, -0.0198,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0026, -0.0316, -0.0454, -0.0386, -0.0019, -0.0166,  0.0205, -0.0913,
        -0.0248,  0.0145,  0.0237, -0.0012,  0.0030,  0.0025, -0.0244, -0.0733,
        -0.0493,  0.0081,  0.0291, -0.0190, -0.0466,  0.0255, -0.0328, -0.0930,
        -0.0044, -0.0259, -0.0579, -0.0218,  0.0167,  0.0068, -0.0171,  0.0028,
        -0.0888,  0.0382,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0201,  0.0143,  0.0365,  0.0255, -0.0739,  0.0443,  0.0151,  0.0258,
        -0.0146,  0.0124, -0.0445, -0.0152,  0.0226, -0.1933, -0.0303,  0.0031,
        -0.1008,  0.0349, -0.0305, -0.0419, -0.0268, -0.1196,  0.0072,  0.0054,
         0.0414,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0262, -0.0093, -0.0176, -0.0096, -0.0547,  0.0042, -0.1173, -0.0769,
        -0.0859, -0.3095,  0.1018, -0.0288, -0.0352,  0.0475, -0.0540, -0.0216,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0981,  0.1241,  0.0099,  0.0722,  0.0218,  0.0208,  0.3527,  0.0169,
        -0.0267, -0.0234, -0.0030, -0.0021,  0.0429,  0.0163,  0.0230, -0.0113,
         0.0093,  0.0219,  0.0125, -0.0044, -0.0069, -0.0304, -0.0494,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0762,  0.0963,  0.1536,  0.1487,  0.1903,  0.0220,  0.0130, -0.0319,
         0.0303, -0.0481, -0.0123, -0.0536, -0.1035, -0.0203,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0385, -0.0381, -0.0512, -0.1350,  0.0054, -0.0063, -0.0237,  0.0146,
        -0.0398, -0.0327, -0.0650, -0.0508, -0.0502, -0.0097,  0.0024, -0.0065,
        -0.0467, -0.1357, -0.0706, -0.0974, -0.0363,  0.0102, -0.0281, -0.0051,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.9442e-02, -1.1460e-01, -1.3703e-01, -2.7018e-02,  4.6132e-02,
        -8.2211e-02, -2.8485e-01, -9.7960e-02, -8.8350e-02,  2.1397e-04,
        -3.2192e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0371,  0.0353, -0.0007, -0.0289, -0.0290, -0.0671, -0.1544,  0.0125,
        -0.1159, -0.0442, -0.0551,  0.0504, -0.0291, -0.0237, -0.0259,  0.0172,
        -0.0351, -0.0747,  0.0516, -0.0221,  0.0900,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.3012, -0.0627, -0.0600, -0.0134, -0.0557, -0.0394, -0.0441, -0.1949,
        -0.0210, -0.1183, -0.0729,  0.0088,  0.0051, -0.0024,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0840,  0.0356, -0.0079, -0.0132, -0.0137, -0.0161, -0.0079, -0.0649,
        -0.0186,  0.0285,  0.0069,  0.0745,  0.2184,  0.0259,  0.0307,  0.0906,
        -0.1125,  0.0210, -0.0202, -0.0191,  0.0898,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0594, -0.3167,  0.0009, -0.0229,  0.0424, -0.1509,  0.0100, -0.1786,
        -0.0261, -0.0271, -0.0441,  0.0676,  0.0533,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0066, -0.0306,  0.0121, -0.0004,  0.0088,  0.0082, -0.0214, -0.0243,
         0.0204,  0.0011,  0.0112, -0.0250,  0.0194,  0.0006, -0.0138, -0.0041,
         0.0133,  0.0091,  0.0170, -0.0110,  0.0152,  0.0176,  0.0262,  0.0153,
        -0.1002,  0.0067, -0.1340, -0.0701, -0.0624,  0.0089, -0.0299, -0.0407,
         0.0158,  0.0121,  0.0163, -0.0155, -0.0140, -0.0069,  0.0072, -0.0171,
        -0.0061,  0.0115,  0.0095, -0.0679,  0.0147], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0129, -0.0333, -0.0501, -0.0146,  0.0286,  0.0087, -0.0069, -0.0127,
        -0.0037, -0.0053, -0.0553,  0.0036, -0.0042,  0.0359, -0.0911, -0.0321,
        -0.1640, -0.0519, -0.0205, -0.0131, -0.0104, -0.0700, -0.0015, -0.0699,
        -0.0153,  0.0106, -0.0152,  0.0063, -0.0426, -0.0230,  0.0003,  0.0080,
        -0.0080,  0.0707,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1160,  0.1212,  0.0681,  0.0693,  0.0361,  0.0293,  0.0225,  0.0073,
         0.0152,  0.0193,  0.0092,  0.0046,  0.0044,  0.0194,  0.0014, -0.0068,
        -0.0013,  0.0203,  0.0095,  0.1190,  0.2285, -0.0101,  0.0270, -0.0121,
        -0.0204,  0.0015,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0669, -0.2327, -0.0094, -0.0474,  0.0341,  0.0008, -0.0239, -0.1039,
        -0.0480, -0.0369, -0.0201,  0.0081, -0.0135,  0.0027,  0.0300, -0.0630,
        -0.1322, -0.0051, -0.0181, -0.0016, -0.0427,  0.0365,  0.0223,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0664, -0.0887,  0.0184, -0.0114, -0.0003,  0.0005, -0.0176, -0.0170,
        -0.0518, -0.0029,  0.0162, -0.0085, -0.0013,  0.0134, -0.0753,  0.0431,
        -0.0020, -0.0424, -0.0730, -0.0559, -0.0140, -0.0004, -0.0028,  0.0419,
        -0.0039, -0.0107, -0.0207, -0.0960,  0.0035, -0.0063,  0.0180, -0.0043,
         0.0111,  0.0046, -0.0019, -0.0251,  0.0075,  0.0012, -0.0436,  0.0765,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0201, -0.0880, -0.0553, -0.0182, -0.0560,  0.0017, -0.0011, -0.0590,
         0.0132, -0.0075, -0.0218,  0.0028, -0.0520, -0.0257, -0.0528, -0.0527,
        -0.0409, -0.0139, -0.0124, -0.0294,  0.0037, -0.0462, -0.0375, -0.0080,
        -0.0174,  0.0029,  0.0097, -0.0585, -0.0125, -0.0067, -0.0484, -0.0080,
        -0.0857,  0.0301,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-5.2753e-02,  6.5786e-03, -2.6168e-04,  6.9303e-03, -5.2893e-02,
         3.7766e-02,  8.1728e-03, -3.2272e-05,  3.5597e-03,  7.2331e-03,
        -7.3478e-03, -9.7519e-03,  1.8046e-02, -2.0051e-02, -1.6213e-01,
         6.6951e-03, -1.5394e-02, -1.4287e-02, -5.3860e-02,  5.8549e-03,
        -1.6989e-02,  3.9464e-03,  3.8300e-03,  1.6631e-02,  6.0533e-03,
        -6.2210e-02,  4.0316e-03, -1.3295e-02, -6.3168e-03, -1.2661e-02,
        -1.3636e-02, -5.4317e-03,  5.7624e-03, -1.2914e-02, -3.4609e-02,
        -9.3889e-03, -3.9142e-03, -2.9956e-02, -2.0807e-02, -3.8946e-02,
        -3.3487e-02, -1.3267e-02, -2.5009e-02, -1.5315e-02, -1.4471e-02,
         2.6099e-03,  1.4846e-02,  1.0715e-02, -8.3832e-03,  4.2946e-03,
        -2.3967e-02, -2.2711e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1155, -0.0314, -0.0657, -0.0376, -0.0901, -0.0376,  0.0159,  0.0252,
         0.0037, -0.0231,  0.0043,  0.0295, -0.0026, -0.0667, -0.0109, -0.0737,
        -0.0979, -0.0311, -0.0339, -0.0121, -0.0256, -0.0184, -0.0272, -0.0629,
         0.0003,  0.0040,  0.0066,  0.0065, -0.0018,  0.0115,  0.0267,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1648, -0.0896,  0.0552,  0.0277, -0.0261,  0.0151, -0.0134,  0.0197,
        -0.0414,  0.0168, -0.0306, -0.0190,  0.0212, -0.0116, -0.0269, -0.0681,
         0.0158, -0.0366, -0.0681, -0.0768, -0.0268, -0.0079, -0.0331,  0.0362,
        -0.0519,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0444,  0.0042,  0.0542, -0.0108,  0.0999,  0.1772,  0.2275,  0.0904,
         0.0748,  0.0909,  0.0589, -0.0668,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0721, -0.2250,  0.0021, -0.0799, -0.0747, -0.0025,  0.0113, -0.0039,
         0.0132, -0.0153,  0.0046, -0.0208, -0.0402, -0.0975,  0.0334, -0.0308,
         0.0077, -0.0051,  0.0159,  0.0158, -0.0233,  0.0116, -0.0181, -0.0107,
         0.0247, -0.0999,  0.0056,  0.0026, -0.0318,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0241, -0.3071,  0.0582,  0.0593, -0.0054, -0.0886, -0.1901,  0.0640,
        -0.0358,  0.0622,  0.0568, -0.0484,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0223, -0.3150,  0.0211, -0.0689, -0.0324, -0.0752, -0.2050, -0.0090,
        -0.0891, -0.0282, -0.0711, -0.0628,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0115,  0.0666, -0.1454, -0.0047, -0.0965,  0.0180, -0.0317, -0.1005,
        -0.1323, -0.0122, -0.0128, -0.0720, -0.0293,  0.0164,  0.0479, -0.0582,
        -0.0233, -0.0082,  0.1072, -0.0055,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0228, -0.2404, -0.0139, -0.0167, -0.1542, -0.0455, -0.0322, -0.0021,
        -0.0155,  0.0020, -0.0013, -0.0411, -0.0345, -0.0091,  0.0206,  0.0094,
        -0.0229, -0.0100, -0.0058, -0.0175, -0.0226,  0.0023, -0.0101,  0.0147,
         0.0140,  0.0223,  0.0106, -0.0204, -0.0089,  0.0065,  0.0034, -0.0040,
        -0.0265, -0.0659, -0.0047,  0.0328, -0.0129,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0378,  0.6856, -0.0109, -0.0601,  0.0536, -0.0303, -0.0325,  0.0604,
         0.0287,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0317,  0.0240,  0.1003, -0.1276, -0.0283, -0.0103, -0.0583, -0.0810,
        -0.0068, -0.0025,  0.0094,  0.0134, -0.0725, -0.0765,  0.0006, -0.0443,
        -0.0294,  0.0226, -0.0500, -0.0689,  0.0189,  0.0140, -0.0079,  0.0394,
        -0.0040,  0.0189,  0.0388,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0506,  0.2830, -0.0162,  0.0355,  0.0138,  0.0187, -0.0107, -0.0090,
        -0.0016,  0.0060,  0.0182,  0.0048,  0.0101, -0.0210,  0.0073, -0.0272,
         0.0628, -0.0042,  0.0170,  0.0060, -0.0196,  0.0531,  0.0280,  0.0060,
        -0.0017, -0.0068,  0.0107,  0.0164,  0.0655,  0.0125,  0.0090, -0.0378,
         0.0408,  0.0683,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 4.7652e-02,  6.3495e-03,  1.4467e-03,  3.9775e-03, -2.1566e-02,
        -1.1802e-02, -3.3438e-03,  5.4620e-05,  1.4090e-02,  1.4422e-02,
        -1.2171e-02, -7.0402e-02, -3.5877e-02, -1.1191e-02,  1.4760e-04,
        -5.7984e-03, -6.1843e-02,  5.7331e-02, -1.1781e-03, -4.8581e-03,
         1.1999e-02,  2.7186e-03, -4.9222e-02, -2.6625e-02, -2.0369e-02,
         4.4814e-03, -6.4909e-03, -5.3537e-03,  7.0652e-03,  2.5080e-04,
         6.5043e-03, -4.8834e-02, -1.1565e-01, -4.7218e-03,  6.4515e-03,
        -2.1642e-02, -5.5290e-02,  6.0579e-03,  7.5895e-03,  6.5337e-03,
         4.7860e-03, -3.5615e-02,  1.5453e-02,  3.6977e-02,  1.0782e-01,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0793, -0.0054, -0.0228, -0.0241, -0.0653, -0.1059, -0.1816, -0.0464,
        -0.0359, -0.1025, -0.0378, -0.0479, -0.0940,  0.0486, -0.0384,  0.0639,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1848,  0.0202,  0.0528,  0.0419, -0.0247,  0.0609, -0.0138,  0.0716,
         0.0803,  0.0595,  0.0455,  0.0145,  0.0231,  0.0454,  0.0288, -0.0077,
         0.0116, -0.0109,  0.0093,  0.0149, -0.0080, -0.0056,  0.0246,  0.0131,
         0.0311, -0.0668, -0.0288,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0247,  0.0088, -0.0469, -0.0112, -0.0007, -0.0150, -0.0546, -0.0031,
        -0.0427, -0.0076,  0.0027, -0.0002, -0.0087, -0.0035, -0.0157, -0.0125,
         0.0038,  0.0117,  0.0376, -0.0341,  0.0052, -0.0292, -0.0032, -0.0687,
         0.0009,  0.0038, -0.0453, -0.0656,  0.0012, -0.0900, -0.0045, -0.0446,
        -0.0450, -0.0170, -0.0131, -0.0302, -0.0089, -0.0380, -0.0002,  0.0071,
        -0.0026, -0.0267, -0.0024, -0.0161, -0.0625,  0.0222], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1468,  0.0051, -0.0043, -0.0106, -0.0023, -0.0414, -0.0042, -0.0619,
         0.0094,  0.0186, -0.0002,  0.0498,  0.0112, -0.0032, -0.0150,  0.0088,
        -0.0275, -0.0195, -0.0537, -0.1559, -0.0237, -0.0187,  0.0057, -0.0315,
         0.0057, -0.0333, -0.1059, -0.0763,  0.0336, -0.0166,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1121,  0.0708, -0.2448, -0.0236,  0.0653,  0.1110, -0.0035,  0.1846,
         0.0877, -0.0645,  0.0321,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1411, -0.2357, -0.0671, -0.0093, -0.0427, -0.0171, -0.0449, -0.1754,
         0.0219, -0.0621,  0.0102, -0.0358, -0.0180, -0.0025,  0.0086, -0.0559,
        -0.0065, -0.0326, -0.0117, -0.0010,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1139,  0.4524,  0.0931,  0.0924, -0.0372,  0.0750,  0.0706,  0.0038,
         0.0169,  0.0222, -0.0224,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0244,  0.2615,  0.1334,  0.1253,  0.0443,  0.0409,  0.0033,  0.0178,
         0.0580, -0.0192,  0.0255, -0.0263, -0.0023,  0.0504,  0.0119,  0.0266,
        -0.0027,  0.0014, -0.0067,  0.0495, -0.0685,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1892,  0.1163,  0.0693,  0.0372, -0.0383,  0.2456, -0.0650, -0.0354,
         0.0482, -0.0177, -0.0113,  0.0100, -0.0210, -0.0107,  0.0173,  0.0397,
        -0.0136, -0.0143,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.6056e-02, -7.8350e-02, -1.6487e-01, -5.1249e-02, -4.2651e-02,
        -2.1144e-03,  8.2431e-04,  6.2130e-03, -9.7077e-03, -6.4310e-03,
         1.9914e-02,  4.6176e-04, -1.3596e-04, -5.0346e-02, -4.4673e-02,
        -1.3158e-02, -2.6551e-02, -4.3941e-02, -3.8216e-02, -3.2026e-02,
        -2.2977e-02, -3.5171e-02, -6.2839e-02, -6.2898e-03, -2.3256e-02,
         1.1360e-02, -3.0978e-02, -2.1358e-02, -1.1680e-02, -5.0363e-02,
        -2.7448e-03, -1.0598e-02, -6.1851e-03,  4.0284e-03, -1.3788e-02,
        -4.8685e-03, -6.2509e-03, -3.7803e-03,  3.5925e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.2632e-02, -3.2161e-01, -1.3973e-02, -1.7408e-02, -2.4338e-02,
         4.4964e-05, -4.5637e-03, -7.0946e-03,  3.4681e-02, -4.4986e-03,
        -1.7072e-02, -3.6588e-02, -4.4732e-04, -8.5786e-03,  1.2066e-02,
        -4.1948e-02, -7.3612e-02, -2.3510e-02, -6.0100e-03, -1.5665e-02,
        -1.4340e-02, -1.9946e-02,  1.5977e-02,  2.0768e-03, -4.3605e-03,
         3.1173e-03, -1.8330e-02,  8.6049e-03, -4.8907e-02, -5.4111e-02,
        -2.5826e-04, -9.8724e-03, -5.1786e-02,  4.1965e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.2191,  0.2954,  0.2035,  0.0980,  0.0034,  0.1086,  0.0291,  0.0120,
         0.0308,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0055,  0.1808,  0.0318,  0.0542,  0.0354,  0.1147,  0.0129,  0.0270,
        -0.0117,  0.0149,  0.0318,  0.0086, -0.0268, -0.0034,  0.0184,  0.0336,
         0.0016,  0.0766, -0.0024,  0.0207,  0.0148, -0.0095,  0.0037,  0.0078,
         0.0312,  0.0050,  0.0148,  0.0042, -0.0233, -0.0029, -0.0019,  0.0115,
        -0.0056,  0.0058, -0.0043, -0.0194, -0.0158, -0.0188,  0.0152,  0.0715,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.9950e-02, -5.7930e-02, -1.4941e-01, -2.0067e-02, -1.7075e-06,
         1.9919e-02, -7.0881e-03, -3.9897e-03, -1.3823e-02, -5.4400e-02,
        -1.5476e-03,  8.2223e-03,  4.6319e-03,  7.2990e-03,  1.1583e-02,
         3.2002e-03,  8.3422e-04,  1.2477e-02, -4.3160e-02,  8.8623e-03,
        -4.9122e-03, -6.2745e-03,  1.4878e-02, -1.8568e-02,  1.0897e-02,
        -1.0904e-02, -5.8842e-03,  5.3846e-03, -5.9877e-02, -5.5159e-02,
        -1.0706e-02, -3.6585e-02,  7.1695e-03, -5.8377e-03,  7.5638e-03,
         1.5117e-02, -2.1426e-02, -4.7612e-02,  7.5557e-03, -2.4412e-02,
         3.2515e-02, -5.4152e-03,  4.5897e-03,  6.3772e-03,  1.2676e-03,
         5.1033e-03,  9.3831e-03,  3.3048e-03,  1.1477e-02, -2.9458e-02,
        -3.2944e-02, -3.3043e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1266, -0.2177, -0.0445, -0.0082,  0.0593,  0.0021, -0.0436,  0.0034,
        -0.0242, -0.0066,  0.0143, -0.0362, -0.0767, -0.0049, -0.0320, -0.0239,
        -0.0509, -0.0477, -0.0628, -0.0030,  0.0711, -0.0402,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.6460e-04,  5.2175e-01,  6.6204e-02,  3.6149e-03,  2.1684e-02,
        -7.9316e-04,  1.2687e-02,  5.3597e-02,  5.3509e-04, -1.8388e-03,
         5.2388e-02,  2.8915e-02,  2.6227e-02,  1.4338e-02,  2.6794e-02,
         4.9406e-02, -1.1875e-02, -7.1563e-04,  1.6675e-02, -5.4933e-02,
        -3.4866e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0047, -0.0644, -0.0203, -0.0220,  0.0322, -0.0023, -0.0225, -0.0131,
        -0.0148, -0.0018, -0.0040,  0.0100,  0.0021, -0.0014,  0.0027,  0.0014,
         0.0065,  0.0061, -0.0161, -0.0058, -0.0015,  0.0076,  0.0055,  0.0029,
         0.0074,  0.0069, -0.0296, -0.0275, -0.0102, -0.0118, -0.0838, -0.0618,
        -0.0162, -0.0190, -0.0146,  0.0009,  0.0143,  0.0042, -0.0016,  0.0084,
        -0.1371, -0.0053, -0.0139, -0.0280, -0.0449, -0.0053, -0.0243, -0.0239,
         0.0068,  0.0289, -0.0236, -0.0016,  0.0003, -0.0049, -0.0612],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0281, -0.0167,  0.0280, -0.0162, -0.0076, -0.0200, -0.0511,  0.0054,
         0.0172, -0.0032,  0.0232, -0.0044, -0.0096, -0.0317, -0.0103, -0.0159,
         0.0056, -0.0109, -0.0078, -0.0193, -0.0467, -0.0236, -0.0334, -0.0023,
         0.0317, -0.0021,  0.0042, -0.1113,  0.0020, -0.0202, -0.0116, -0.0216,
        -0.0292, -0.0688,  0.0427, -0.0333, -0.0138, -0.0446, -0.0339, -0.0494,
        -0.0088,  0.0072,  0.0010,  0.0200,  0.0047,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1351,  0.1165,  0.0794,  0.0620, -0.0059,  0.0126,  0.0228,  0.0270,
         0.0458,  0.0031,  0.0216,  0.0434,  0.0176,  0.0430,  0.0422, -0.0596,
         0.0255,  0.0935,  0.0134,  0.0079,  0.0638, -0.0062,  0.0485, -0.0038,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0036,  0.0463, -0.0133,  0.0139, -0.0198,  0.0077, -0.0280, -0.0310,
         0.0121,  0.0134, -0.0013,  0.0035, -0.0110, -0.0591, -0.0152,  0.0355,
        -0.0824, -0.0439, -0.0432, -0.0431, -0.0116, -0.0033, -0.0506, -0.0382,
         0.0547, -0.0551, -0.0380, -0.0248, -0.0430, -0.0692, -0.0255, -0.0141,
        -0.0451,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0742, -0.4769, -0.0383, -0.0382, -0.0079,  0.0022, -0.0068, -0.0107,
        -0.1263, -0.0131,  0.0297, -0.0095,  0.0302, -0.0142,  0.0844, -0.0060,
        -0.0042,  0.0258, -0.0012,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1284, -0.0242,  0.0281, -0.0489, -0.1037,  0.0115, -0.0580, -0.0445,
        -0.0952, -0.0217, -0.0099, -0.0267, -0.0115, -0.0332, -0.0287,  0.0121,
        -0.0176, -0.0582, -0.0110, -0.0139, -0.0353, -0.1242, -0.0185,  0.0112,
         0.0238,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1589, -0.2057, -0.0622, -0.0240, -0.0523, -0.0602, -0.1481,  0.0263,
        -0.0050, -0.0220, -0.0107, -0.0033, -0.0286,  0.0015, -0.0108,  0.1088,
         0.0715,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.0353,  0.0230, -0.0399,  0.0334, -0.0276, -0.0064,  0.0114, -0.0459,
        -0.0290, -0.0587, -0.0079, -0.0376,  0.0061, -0.0073, -0.0182, -0.0124,
        -0.0291, -0.1003, -0.0141, -0.0530,  0.0171, -0.0993, -0.0931, -0.0062,
        -0.0225, -0.0299, -0.0173, -0.0473, -0.0202, -0.0171,  0.0118, -0.0216,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0434,  0.1484,  0.1149,  0.0270, -0.0264,  0.0249,  0.0684,  0.0912,
         0.0155,  0.0012, -0.0118,  0.0388,  0.0012,  0.0108,  0.0608, -0.0217,
         0.0017,  0.0100,  0.0527,  0.0029,  0.0359, -0.0245,  0.0467,  0.0350,
         0.0498,  0.0107,  0.0206, -0.0030,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0549,  0.0559, -0.0045, -0.0107,  0.0313,  0.0466,  0.0929,  0.0014,
         0.0499,  0.0458,  0.0310,  0.0809,  0.0217,  0.0127,  0.0154,  0.0786,
         0.0224,  0.0295,  0.0385, -0.0071,  0.0183,  0.0122,  0.0104,  0.0290,
         0.0448, -0.0286,  0.0026,  0.0625, -0.0601,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0454, -0.0369, -0.0209, -0.0426, -0.1167, -0.0150,  0.0270,  0.0012,
        -0.0017,  0.0283,  0.0047, -0.0181, -0.0315, -0.0912,  0.0177, -0.0359,
        -0.0240, -0.1241, -0.0162, -0.0416,  0.0084,  0.0123, -0.0388, -0.0247,
        -0.0146, -0.0092, -0.0107, -0.0140, -0.0092,  0.0321,  0.0070,  0.0479,
        -0.0303,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0621, -0.2204, -0.0278, -0.0203, -0.0380, -0.0611,  0.0318, -0.0645,
        -0.0066, -0.0488, -0.0356,  0.0012, -0.0085, -0.0156, -0.0139,  0.0003,
        -0.0460, -0.0136, -0.1831,  0.0292,  0.0164, -0.0132,  0.0209, -0.0209,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0612, -0.1897, -0.0914, -0.0137, -0.0073,  0.0806, -0.0379, -0.0692,
         0.0059, -0.0352, -0.0399, -0.0774, -0.0301, -0.0204,  0.0002, -0.0097,
         0.0045, -0.0322, -0.0051, -0.1189, -0.0294, -0.0038, -0.0364,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.5948e-03, -1.8508e-01, -2.3282e-02,  2.6455e-02, -1.3777e-02,
        -4.1847e-02,  6.1116e-03, -1.1348e-03,  1.2641e-02, -3.9784e-02,
         3.5360e-02, -8.9466e-03, -1.6393e-02,  1.3689e-03, -5.3360e-03,
        -5.9191e-02, -5.4752e-02, -4.0912e-02, -2.5773e-02, -1.4775e-02,
        -1.1956e-03, -3.0752e-02, -8.0410e-03,  9.1122e-03, -6.3995e-02,
        -1.9905e-02, -9.3310e-03, -1.4244e-02, -7.9362e-04, -5.6260e-02,
        -3.9313e-02, -1.6787e-02,  2.3946e-03, -5.7219e-03, -2.1487e-02,
        -2.6873e-02, -2.3058e-02, -3.4219e-02,  6.5078e-06,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.8293e-02, -1.1482e-01, -4.2934e-02, -6.1269e-03, -1.6797e-01,
        -2.5457e-03, -2.7975e-04, -2.3493e-04,  5.2764e-03,  2.4385e-02,
        -5.1794e-03, -8.5856e-03,  4.0962e-05,  5.8849e-03,  5.3867e-03,
        -8.6673e-02, -1.5769e-02,  5.4224e-03, -2.9384e-03, -1.0653e-02,
        -2.8529e-02,  3.2609e-03, -2.0463e-03,  3.9537e-03, -9.1324e-02,
        -7.0641e-02, -2.2649e-02, -2.2838e-03, -9.0827e-03, -5.1680e-02,
         1.4954e-03,  1.3643e-03, -7.9783e-03, -8.6844e-03, -1.0110e-02,
        -1.1279e-02, -1.8793e-03, -6.9139e-02,  5.7182e-03,  3.5309e-03,
        -1.8934e-02,  1.9928e-03,  4.3051e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5812e-02, -1.5961e-01,  1.7079e-02, -2.1333e-03, -8.7039e-03,
        -1.7092e-02, -3.5506e-03,  1.5505e-02, -5.3121e-02,  1.9267e-02,
        -7.1255e-03, -5.7205e-03, -9.9248e-03, -3.4695e-03, -1.1770e-02,
        -2.8123e-02, -2.8028e-02, -7.7000e-02, -7.2444e-03, -6.5632e-03,
         4.7468e-03, -6.7248e-03, -5.1981e-03, -3.2737e-03, -3.2779e-03,
        -1.1242e-02, -4.0732e-03,  1.1933e-04, -1.3399e-02, -6.5547e-03,
        -7.4816e-04, -2.6224e-02, -5.0204e-02,  2.2707e-02,  1.7182e-02,
         1.4463e-02,  2.2350e-02, -1.2824e-02, -2.4019e-02, -2.3660e-02,
        -1.7611e-02, -5.0720e-02, -1.1064e-02,  1.3023e-02,  4.7256e-03,
        -3.4803e-03, -4.8401e-03,  3.1905e-03, -3.1865e-03, -1.5438e-02,
         2.5918e-02, -1.9941e-03, -2.2376e-02, -3.2704e-03,  1.6398e-02,
         3.6760e-03, -2.9255e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0249,  0.3577, -0.0005,  0.0803,  0.0839,  0.1031,  0.0666,  0.0809,
         0.0473,  0.0144,  0.1030, -0.0064,  0.0199,  0.0111,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0119,  0.1495,  0.0397,  0.0057,  0.0162,  0.0401,  0.0169,  0.0095,
         0.0020,  0.0273,  0.0365,  0.2342,  0.0501, -0.0012,  0.0480,  0.0419,
         0.0178,  0.0358,  0.0050,  0.0545,  0.0498,  0.0034,  0.0539, -0.0026,
        -0.0049, -0.0013, -0.0121,  0.0146, -0.0071, -0.0066,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2656,  0.2299, -0.0077,  0.1290,  0.1758,  0.1068, -0.0009, -0.0843,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([-1.0993e-01,  1.4844e-01,  5.6576e-02,  3.0833e-02,  3.2951e-02,
         6.6782e-02, -1.0792e-02,  1.3508e-02, -1.3816e-02,  5.4116e-03,
         8.1745e-03,  3.5481e-02,  2.6080e-02,  5.1557e-02, -1.2959e-02,
         4.7227e-03,  3.5926e-06,  5.0646e-03,  4.1216e-04, -1.1671e-03,
        -2.7746e-02, -3.7933e-03,  8.4132e-03,  3.1736e-02,  2.6423e-02,
         3.4958e-03,  2.6091e-02, -6.5135e-03,  4.4016e-02,  8.4589e-02,
         2.6822e-03,  4.7935e-02, -5.1905e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.4594e-02, -1.4770e-01, -6.8404e-02, -2.4659e-02,  5.2724e-03,
        -2.8728e-02, -1.5433e-02, -4.4045e-02,  4.2243e-02,  3.0133e-03,
        -3.0155e-03, -7.1184e-05, -1.2392e-02, -2.7624e-03,  2.6413e-03,
        -1.2480e-03, -1.4622e-03,  5.3038e-03, -1.3426e-02,  8.6330e-03,
         1.1417e-02, -2.7127e-03, -4.7959e-03, -3.1490e-04,  1.3391e-02,
        -1.4395e-02, -2.1956e-02, -2.8524e-02,  4.1462e-03, -6.4548e-03,
        -9.6145e-03,  4.1419e-03, -1.9423e-02, -5.0514e-02,  6.9597e-03,
        -1.8804e-03, -9.4639e-03, -2.1853e-03, -1.9356e-04,  8.5231e-03,
         1.1328e-03, -2.4650e-03, -4.2715e-03, -2.7274e-02,  7.6236e-06,
         5.1362e-05, -8.9053e-03,  1.0743e-02, -1.2453e-02, -1.0910e-02,
        -4.4856e-03, -1.4026e-02, -2.9827e-03, -1.2858e-03, -6.3477e-03,
         1.2274e-02,  9.0653e-03,  2.1752e-03, -7.1890e-03, -1.4526e-02,
        -9.5457e-03, -2.6963e-03, -6.8747e-02,  4.5073e-03, -4.3304e-03,
        -2.8411e-03,  2.7395e-04,  3.4831e-02,  8.9921e-03,  3.2969e-03,
        -3.1316e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0588, -0.0069, -0.0081, -0.0110, -0.0200, -0.0196, -0.1522, -0.0096,
         0.0108, -0.0045, -0.0109,  0.0154,  0.0101,  0.0198,  0.0042, -0.0058,
        -0.0096,  0.0025, -0.0221, -0.0537,  0.0050, -0.0003,  0.0570, -0.0377,
         0.0136, -0.0240, -0.0136, -0.0223, -0.0130, -0.0165,  0.0138,  0.0205,
        -0.0601, -0.0568, -0.0198, -0.0280, -0.0384, -0.0056, -0.0118,  0.0043,
         0.0477, -0.0345,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0922, -0.2142,  0.0036,  0.0067, -0.0196,  0.0039,  0.0242,  0.0218,
        -0.0255, -0.0662, -0.0128, -0.0232, -0.0629, -0.0065,  0.0071,  0.0050,
        -0.0013, -0.0392, -0.0959, -0.0106, -0.0142, -0.0004, -0.0046, -0.0054,
         0.0138, -0.0361, -0.0210, -0.0033, -0.0622, -0.0212, -0.0124, -0.0005,
         0.0118, -0.0030, -0.0050,  0.0199,  0.0226,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0051, -0.4547, -0.1423, -0.1462, -0.0583, -0.0181, -0.0148, -0.0031,
        -0.0125,  0.0032, -0.0196,  0.0300, -0.0685, -0.0012,  0.0224,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1896,  0.0545, -0.0303,  0.0036,  0.0760,  0.0594,  0.0208,  0.0308,
         0.1256,  0.0389, -0.0280,  0.0237,  0.0570,  0.0503, -0.0374, -0.0395,
         0.0609,  0.0519, -0.0053, -0.0163,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1039, -0.0038,  0.0043,  0.0047, -0.1281, -0.0176, -0.0180, -0.0188,
         0.0072, -0.0168, -0.0421, -0.0729, -0.0397, -0.0133, -0.0280, -0.0181,
         0.0134, -0.0157, -0.0048, -0.0471, -0.0036,  0.0016,  0.0071,  0.0018,
         0.0077, -0.0305, -0.0286,  0.0057, -0.0087, -0.0056, -0.0518, -0.0209,
         0.0028, -0.0186, -0.0253, -0.0556, -0.0154, -0.0025, -0.0164, -0.0163,
         0.0219,  0.0171, -0.0158,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1108, -0.0077, -0.0294, -0.0226, -0.0197, -0.0254, -0.0156, -0.0964,
        -0.0272, -0.0099, -0.0237, -0.0064, -0.0085, -0.0562, -0.0998, -0.0206,
         0.0079,  0.0041,  0.0022,  0.0109, -0.0330, -0.0102, -0.0032, -0.0109,
         0.0036, -0.0341, -0.0118, -0.0063, -0.1281, -0.0021, -0.0195, -0.0446,
        -0.0069, -0.0258, -0.0318, -0.0043,  0.0146, -0.0041,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1127, -0.2184, -0.0794, -0.1168, -0.0074, -0.0892,  0.0280,  0.0555,
        -0.0074, -0.0582, -0.0209, -0.0141, -0.0221, -0.0064,  0.0181, -0.0178,
        -0.0215, -0.0384,  0.0182,  0.0201,  0.0078,  0.0217,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0416,  0.0510, -0.0094, -0.0277,  0.0073,  0.0217,  0.0074,  0.0150,
         0.0267,  0.0922, -0.0051, -0.0057,  0.0190, -0.0032,  0.0248,  0.0296,
         0.1415, -0.0042,  0.0102,  0.0405,  0.0120,  0.0013,  0.0328,  0.0160,
         0.0573,  0.0066,  0.0094, -0.0063,  0.0123,  0.0100, -0.0347, -0.0316,
        -0.0116, -0.0163,  0.0100,  0.1073, -0.0081,  0.0133, -0.0079,  0.0114,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0392, -0.0192, -0.0141, -0.0639,  0.0008, -0.0279, -0.0042, -0.0252,
        -0.0859, -0.0262, -0.0096, -0.0116, -0.0193, -0.1362, -0.0668,  0.0027,
        -0.0056, -0.0071, -0.0733, -0.0036, -0.0088,  0.0102,  0.0080, -0.0051,
        -0.0128, -0.0929, -0.0034, -0.0603, -0.0711,  0.0177, -0.0069, -0.0606,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0031,  0.3780,  0.0072, -0.0068,  0.0098, -0.0296, -0.0030,  0.0603,
         0.0921,  0.0200, -0.0159,  0.0209, -0.0076, -0.0108, -0.0065,  0.0077,
         0.0343,  0.0055, -0.0053,  0.0045,  0.0071,  0.0097, -0.0123,  0.0773,
         0.0069, -0.0351,  0.0232,  0.0145, -0.0034,  0.0177,  0.0082,  0.0140,
         0.0223,  0.0119,  0.0073,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.1465, -0.0492, -0.0877, -0.0727, -0.0648, -0.0420, -0.1277, -0.1108,
        -0.0428, -0.0605, -0.0123, -0.0193, -0.0098, -0.0427, -0.0101, -0.0155,
        -0.0134, -0.0032,  0.0080, -0.0291,  0.0317,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0680, -0.3697, -0.0806, -0.0579, -0.0640, -0.0432, -0.0072, -0.0357,
        -0.0935, -0.0079,  0.0038,  0.0014,  0.0058,  0.0179, -0.0427,  0.0342,
        -0.0570,  0.0076, -0.0019,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0308, -0.1603, -0.0832,  0.0298, -0.0416, -0.0193,  0.0103,  0.0022,
        -0.0053, -0.0984, -0.1709,  0.0199, -0.0162, -0.1423, -0.0167, -0.0386,
        -0.0209, -0.0491, -0.0063,  0.0379,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.9570e-02, -2.9426e-01, -6.3022e-03, -7.6634e-02, -1.9143e-02,
         2.9346e-02, -1.0379e-01, -5.9413e-02,  1.0246e-02,  9.8684e-03,
        -1.6065e-02, -1.2398e-02,  1.0973e-04, -2.2643e-02,  5.7361e-03,
        -2.3286e-02, -5.7197e-02,  2.1828e-03, -3.2079e-02, -4.8355e-02,
         8.3387e-03, -1.2155e-02,  7.0883e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.2148e-01,  2.1290e-01,  3.4618e-02,  3.6582e-02, -1.5874e-02,
         8.1084e-03,  6.6076e-02,  1.2594e-02,  2.0527e-02,  1.2650e-01,
         1.4091e-02,  1.3947e-02,  2.4914e-02,  8.3467e-02,  3.0421e-03,
        -1.1751e-04, -2.0681e-02,  8.5640e-03,  1.5011e-02, -1.3164e-02,
         9.3935e-03, -1.2153e-02, -1.2738e-02,  4.3759e-02, -6.9691e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0963,  0.2687,  0.0298,  0.0202, -0.0222, -0.0067, -0.0004, -0.0244,
        -0.0332,  0.0145,  0.0519, -0.0008,  0.0015, -0.0206,  0.0135, -0.0172,
         0.0157,  0.0045, -0.0136,  0.0036,  0.0568,  0.0301,  0.0226,  0.0247,
         0.0159,  0.0215,  0.0154, -0.0124,  0.0344,  0.0212,  0.0192, -0.0219,
         0.0022,  0.0156, -0.0269,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1059,  0.0226, -0.0446, -0.0722, -0.0201,  0.0026, -0.0603, -0.0173,
        -0.1355, -0.0050,  0.0046,  0.0254,  0.0410, -0.0451, -0.0571, -0.1182,
        -0.0225, -0.1361, -0.0105, -0.0314, -0.0080,  0.0137,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0395, -0.0612,  0.0450, -0.0444, -0.0384, -0.0547, -0.0193,  0.0008,
        -0.0251, -0.1734, -0.0355, -0.0365,  0.0130,  0.0031, -0.0834, -0.0922,
        -0.0434, -0.0453,  0.0346,  0.0841,  0.0270,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0611, -0.3086, -0.0404, -0.0403, -0.0141,  0.0116, -0.0349, -0.1049,
        -0.0096,  0.0260,  0.0091, -0.0268,  0.0811, -0.0457,  0.0462, -0.0944,
         0.0288, -0.0030, -0.0134,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0878, -0.0243, -0.0245, -0.0112, -0.0145, -0.0693, -0.0011, -0.0032,
         0.0048, -0.0083,  0.0034, -0.0238, -0.0127, -0.0394, -0.0163,  0.0054,
        -0.0165, -0.0402,  0.0073, -0.0120, -0.0124, -0.0436, -0.0133,  0.0032,
        -0.0023,  0.0093, -0.0119, -0.0467, -0.0215, -0.1135, -0.0374, -0.0404,
        -0.0252, -0.0576,  0.0073,  0.0013,  0.0144,  0.0068,  0.0090, -0.0115,
        -0.0001,  0.0065,  0.0086, -0.0027, -0.0076, -0.0216, -0.0383],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1034, -0.1321, -0.0574, -0.0082,  0.0009, -0.0101, -0.0144,  0.0089,
        -0.0101,  0.0144,  0.0036, -0.0041,  0.0089,  0.0013,  0.0354, -0.0067,
        -0.1035, -0.0136, -0.0192, -0.0084,  0.0106, -0.0108, -0.0526, -0.1325,
        -0.0284,  0.0268, -0.0409, -0.0053, -0.0064, -0.0162,  0.0076, -0.0169,
         0.0372,  0.0432,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0107, -0.3599, -0.0039,  0.0211,  0.0449, -0.0421,  0.0038, -0.0024,
        -0.0031, -0.0143,  0.0395,  0.0140, -0.0748, -0.1092,  0.0046, -0.0197,
        -0.0198, -0.0123, -0.0179, -0.0261,  0.0073,  0.0105, -0.0038, -0.0169,
        -0.0135,  0.0050, -0.0027, -0.0160,  0.0064, -0.0361,  0.0058,  0.0138,
         0.0017,  0.0166,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0526, -0.3157, -0.0444, -0.0671, -0.0623, -0.0602, -0.0039, -0.0352,
        -0.0562, -0.0812, -0.0373, -0.0111, -0.0619, -0.0288, -0.0103, -0.0233,
        -0.0090,  0.0397,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0091, -0.0090, -0.0607, -0.0519, -0.0114,  0.0216, -0.0197,  0.0016,
        -0.1518, -0.2337,  0.0300, -0.0411, -0.0187, -0.0246, -0.0267, -0.0168,
        -0.0161,  0.0005,  0.0181, -0.1468, -0.0009, -0.0181, -0.0087, -0.0097,
        -0.0028,  0.0057, -0.0444,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1810,  0.0139,  0.0547,  0.0136, -0.0075,  0.0129,  0.0295,  0.0023,
        -0.0786,  0.0104,  0.0035, -0.0335, -0.2144, -0.0437, -0.0023,  0.0031,
        -0.0065,  0.0125,  0.0128, -0.2067, -0.0063, -0.0052,  0.0189,  0.0085,
        -0.0066, -0.0110,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0452, -0.3192,  0.0161, -0.0427, -0.1460, -0.0219, -0.0348, -0.0280,
        -0.0193, -0.0380, -0.0699, -0.0143,  0.0114,  0.0220, -0.0361, -0.0672,
        -0.0276,  0.0080, -0.0079, -0.0154, -0.0090,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0466, -0.3249, -0.1419, -0.1923, -0.0285, -0.0103, -0.0912, -0.0095,
         0.0063,  0.0586,  0.0027, -0.0253, -0.0138,  0.0186, -0.0298,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1627, -0.0174,  0.0362, -0.0052,  0.1101, -0.0099, -0.0554, -0.0572,
        -0.0352, -0.0465, -0.0910, -0.0059, -0.1998, -0.0101,  0.0049, -0.0045,
        -0.0724,  0.0753,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.1942e-01, -1.8993e-01, -9.2786e-02, -1.4386e-01, -2.5828e-02,
        -1.8166e-02, -1.0652e-02, -2.0971e-02, -1.8007e-02, -7.5986e-02,
        -7.2793e-02, -3.7149e-02,  4.3034e-02, -6.4510e-02,  1.8950e-04,
        -7.5933e-04, -9.4365e-03,  3.7701e-03,  5.2758e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0606, -0.1438, -0.0487, -0.0398, -0.0563, -0.0070,  0.0259, -0.0061,
        -0.0116, -0.0040,  0.0005,  0.0016, -0.0070,  0.0044, -0.0122, -0.0197,
        -0.0231, -0.0020,  0.0179, -0.0221, -0.0718,  0.0096, -0.0477, -0.0202,
        -0.0500, -0.0174, -0.0212,  0.0132,  0.0124, -0.0242, -0.0016, -0.0165,
        -0.0322, -0.0100, -0.0197, -0.0058, -0.0398, -0.0204, -0.0095, -0.0086,
         0.0185,  0.0053, -0.0037,  0.0065], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0267, -0.1974, -0.0320, -0.0316, -0.0681, -0.0587, -0.0651,  0.0012,
        -0.1010, -0.0231, -0.0222,  0.0044, -0.0173, -0.0073,  0.0178,  0.0102,
         0.0365, -0.0771, -0.0016, -0.0060, -0.0305,  0.0094,  0.0139, -0.0004,
         0.0039, -0.0074,  0.0098, -0.0500,  0.0154, -0.0019, -0.0189, -0.0121,
         0.0092,  0.0043,  0.0075,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0024, -0.3723, -0.0279, -0.0639,  0.0138,  0.0061, -0.0813, -0.0560,
         0.0458, -0.0129,  0.0105, -0.0127, -0.0528,  0.0109, -0.0016, -0.0474,
         0.0067, -0.0042,  0.0110,  0.0194,  0.0095, -0.0079, -0.0373,  0.0857,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0600, -0.2362, -0.0166, -0.0345, -0.0185, -0.0348,  0.0145, -0.0192,
        -0.0687, -0.0158, -0.0101,  0.0003,  0.0114, -0.0297, -0.0623,  0.0141,
         0.0124, -0.0066, -0.0698, -0.0168, -0.0064, -0.0433, -0.0089, -0.0179,
        -0.0151, -0.0392, -0.0170, -0.0171, -0.0060, -0.0025, -0.0336,  0.0004,
        -0.0295,  0.0108,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.5668e-01,  2.5841e-02,  4.4418e-03, -1.3580e-02,  3.0510e-02,
         2.8404e-02, -1.2114e-01, -2.4970e-02,  7.1314e-03, -1.0672e-02,
        -1.4639e-02, -9.7817e-03, -9.9774e-02, -1.1161e-02, -4.4335e-02,
        -1.8701e-01, -2.1101e-05,  6.0217e-03, -7.3184e-03,  1.3599e-02,
         1.8152e-02,  3.0460e-03,  1.9694e-02, -4.6399e-02, -1.5220e-02,
        -3.5084e-03, -1.3772e-02, -3.8237e-02, -1.0343e-02, -1.4595e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0589,  0.0637,  0.0647,  0.2024,  0.0241,  0.0088,  0.0585,  0.0346,
         0.0257,  0.1584, -0.0348, -0.1367,  0.1288,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0152,  0.0265,  0.0540,  0.0749, -0.0887, -0.0392, -0.0021, -0.1024,
        -0.1516, -0.0158, -0.0254, -0.0857, -0.0782, -0.0137, -0.0810, -0.0540,
         0.0297,  0.0619,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0940,  0.2315,  0.2314, -0.0104,  0.1789,  0.0453,  0.1153,  0.0933,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0278, -0.2764,  0.0073,  0.0313, -0.0226,  0.0006,  0.0049,  0.0209,
         0.0106,  0.0097,  0.0157,  0.0118, -0.0136, -0.0048,  0.0054, -0.0005,
         0.0030,  0.0072,  0.0016,  0.0019,  0.0200, -0.0045, -0.0252, -0.0100,
        -0.0038, -0.0198, -0.0062, -0.0114, -0.0595, -0.0310, -0.0029, -0.0445,
        -0.0268, -0.0640, -0.0054, -0.0491, -0.0236,  0.0044,  0.0138,  0.0072,
        -0.0044, -0.0126, -0.0100,  0.0009,  0.0270, -0.0331,  0.0018],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0367,  0.2901,  0.1740,  0.2519,  0.0705,  0.0280, -0.0585,  0.0019,
        -0.0372,  0.0417,  0.0095,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1259,  0.3005,  0.0252,  0.0615,  0.0208,  0.0206,  0.0468,  0.1396,
         0.0163,  0.0638,  0.0339,  0.0044,  0.0648,  0.0273,  0.0045, -0.0008,
         0.0308, -0.0126,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0005,  0.0784,  0.1661, -0.0701, -0.0326,  0.0003,  0.1648,  0.0426,
         0.1263,  0.0331, -0.1266,  0.0043,  0.1190, -0.0083, -0.0270,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0223, -0.3775, -0.0399, -0.0221,  0.0010, -0.1264, -0.1506,  0.0279,
         0.0097, -0.0714, -0.0930, -0.0162,  0.0226, -0.0093, -0.0101,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2087, -0.0092,  0.0028,  0.0299, -0.0087, -0.0221, -0.0012,  0.0352,
        -0.0596, -0.0127,  0.0140, -0.0441, -0.1109,  0.0073, -0.0111, -0.0150,
        -0.0798, -0.0200,  0.0118, -0.0141, -0.1031, -0.0938, -0.0220, -0.0134,
         0.0017, -0.0479,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0581, -0.0907, -0.0557,  0.0172, -0.0551,  0.0197,  0.0096, -0.1200,
        -0.3162, -0.0196, -0.0090, -0.0079, -0.0909,  0.0962, -0.0342,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0004, -0.3559, -0.0533, -0.0324, -0.0778, -0.0138, -0.0610,  0.0325,
        -0.0137, -0.1215,  0.0249,  0.0097, -0.0030, -0.0074,  0.0114, -0.0098,
         0.0006,  0.0271, -0.0672, -0.0093,  0.0019,  0.0130,  0.0040, -0.0009,
        -0.0067, -0.0406,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0841, -0.0262,  0.0350,  0.0051, -0.0229, -0.0274,  0.0107, -0.0189,
        -0.0288, -0.0449, -0.2507, -0.0110, -0.0219, -0.0421, -0.0920, -0.0341,
        -0.0126, -0.0337,  0.0185, -0.0332,  0.0059, -0.0181, -0.0240, -0.0145,
         0.0723, -0.0113,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-2.4655e-02, -3.1640e-01,  4.1499e-02, -8.4180e-03, -4.0771e-03,
         9.4915e-03, -1.8312e-02, -6.2688e-02, -1.6876e-02, -2.0719e-02,
        -2.1062e-02, -3.3506e-02, -1.4952e-02,  2.6561e-04, -4.2109e-02,
         1.6786e-02, -7.7862e-03, -2.9953e-02,  5.9244e-03,  3.5029e-03,
         3.8623e-04, -3.4774e-02, -2.3824e-03, -5.9117e-02, -5.8774e-02,
        -2.1746e-02, -3.0629e-02, -1.4249e-02,  8.3869e-03, -1.1876e-04,
         4.8183e-03, -7.9296e-03, -1.9922e-02,  8.1314e-03,  1.7414e-02,
        -1.2244e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1072, -0.1172, -0.0477, -0.0059, -0.0169, -0.0137, -0.0622, -0.0732,
        -0.0076,  0.0079, -0.0246,  0.0032, -0.0031, -0.0112, -0.0052, -0.0010,
         0.0103, -0.0052,  0.0128, -0.0162,  0.0155, -0.0212,  0.0061,  0.0002,
        -0.0297, -0.0384,  0.0179, -0.0226, -0.0387, -0.0073, -0.0165, -0.0198,
        -0.0344,  0.0069,  0.0512, -0.1212,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0916, -0.2521,  0.0123, -0.0481,  0.0514, -0.0514, -0.0286,  0.0179,
        -0.0158,  0.0639,  0.0173, -0.0390, -0.1470,  0.0193,  0.0236, -0.0230,
        -0.0259, -0.0482,  0.0200, -0.0036,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0965, -0.0484, -0.0110, -0.0206,  0.0029, -0.0147, -0.0409, -0.0466,
        -0.0049, -0.0479,  0.0003,  0.0259,  0.0069,  0.0052, -0.0887, -0.0292,
        -0.0379, -0.0654, -0.0096, -0.0147, -0.0417, -0.0169, -0.0034,  0.0095,
        -0.0328, -0.0035, -0.0176, -0.0075, -0.0427, -0.0024, -0.0361, -0.0016,
         0.0027, -0.0439,  0.0389, -0.0333,  0.0358, -0.0112,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0557, -0.1187, -0.0370, -0.0806, -0.0074,  0.0099, -0.0024,  0.0147,
        -0.0337, -0.0385, -0.0868, -0.0785,  0.0079,  0.0156,  0.0186, -0.0386,
         0.0305, -0.0387, -0.0075,  0.0157,  0.0208,  0.0086,  0.0069,  0.0070,
        -0.0043, -0.0165, -0.0068, -0.0087, -0.0604,  0.0006, -0.0106, -0.0158,
         0.0114, -0.0048, -0.0379,  0.0040,  0.0042, -0.0040,  0.0137, -0.0075,
        -0.0085], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0636, -0.3458,  0.0324,  0.0033, -0.0053, -0.0319, -0.0376, -0.0406,
        -0.0229, -0.0033, -0.0103, -0.0241, -0.0399, -0.0250, -0.0070, -0.0512,
         0.0178, -0.0536, -0.0053,  0.0814, -0.0782,  0.0194,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0374,  0.0050, -0.0026,  0.0070,  0.0103, -0.0147, -0.0092,  0.0030,
        -0.0177,  0.0066, -0.0096, -0.0648, -0.0136, -0.0340,  0.0014, -0.0761,
        -0.0075, -0.0217,  0.0069, -0.0859, -0.1132,  0.0312,  0.0259, -0.0488,
        -0.0193, -0.0061,  0.0037, -0.0386, -0.0763, -0.0108, -0.0109, -0.0512,
        -0.0589,  0.0105, -0.0596,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1857,  0.0598,  0.1371,  0.0710,  0.0634, -0.0041, -0.0087, -0.0081,
        -0.0063,  0.0136,  0.0339,  0.0049,  0.0611, -0.0211, -0.0250,  0.0418,
        -0.0267, -0.0116,  0.0232,  0.0004,  0.0251, -0.0143,  0.0812,  0.0123,
         0.0033,  0.0040,  0.0009,  0.0135,  0.0379,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.2899e-01, -2.7594e-01, -6.2208e-02, -9.0230e-02, -3.0043e-02,
        -7.5949e-02, -3.5602e-02, -2.9016e-02, -6.1943e-03, -1.0481e-02,
        -1.2174e-02, -2.0836e-02, -1.3489e-02,  1.5495e-02,  1.3001e-04,
        -1.1018e-02, -1.8698e-02,  1.1744e-02, -6.0686e-04,  2.0243e-03,
        -5.7223e-03, -1.3783e-02,  1.0438e-02,  5.2317e-02,  6.6868e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0211,  0.0372,  0.0003, -0.0629, -0.2145,  0.0096, -0.0262, -0.1044,
        -0.0884,  0.0151,  0.0086,  0.0457, -0.0282,  0.0270, -0.0680, -0.1570,
         0.0197, -0.0134,  0.0109, -0.0089, -0.0128,  0.0120, -0.0083,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0671, -0.0072,  0.0007,  0.0084,  0.0018,  0.0205, -0.0086, -0.0071,
         0.0011, -0.0130, -0.0003, -0.0119, -0.0546, -0.1107,  0.0336, -0.0044,
        -0.0059, -0.0186, -0.0153,  0.0019, -0.0100, -0.0305,  0.0038, -0.0016,
        -0.0247, -0.0455, -0.0868,  0.0326, -0.0520, -0.0516, -0.0151, -0.0192,
        -0.0361, -0.0303, -0.0011,  0.0156,  0.0030,  0.1096, -0.0381,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1352,  0.0637,  0.0825, -0.1116, -0.2414, -0.0407, -0.0095, -0.0604,
        -0.1436,  0.0159,  0.0599, -0.0357,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.1212, -0.0614, -0.2431,  0.0014, -0.0164,  0.0218, -0.0232,  0.0219,
        -0.0076, -0.0043, -0.0027,  0.0008, -0.0049, -0.0011, -0.0206,  0.0051,
        -0.0106, -0.0428,  0.0193, -0.0164, -0.0094, -0.0138,  0.0221, -0.0069,
         0.0023, -0.0053,  0.0152, -0.0041,  0.0166,  0.0373, -0.0513, -0.0227,
        -0.0174,  0.0082,  0.0025, -0.0097, -0.0085,  0.0053, -0.0033,  0.0079,
         0.0135,  0.0188,  0.0289,  0.0098, -0.0123,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0089, -0.3622,  0.0121, -0.0327,  0.0451,  0.0782,  0.0258, -0.0775,
        -0.1962,  0.0267,  0.0540, -0.0806,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1967,  0.5007,  0.1061,  0.0841, -0.0610, -0.0266,  0.0132,  0.0115,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.8265e-04, -1.6371e-01, -1.7008e-02, -7.0558e-04,  2.3743e-03,
        -2.5433e-03, -2.8043e-03, -7.7601e-04, -5.3590e-03, -7.5359e-02,
        -3.3977e-02,  6.9678e-03, -4.4600e-03,  6.6136e-03,  1.6555e-04,
         2.2154e-03,  6.8382e-03, -1.9691e-02,  4.5850e-03, -5.1272e-03,
        -1.6178e-03,  6.7622e-03, -2.8078e-03, -1.9748e-03,  5.0029e-03,
        -1.3264e-02, -1.4477e-04, -1.5017e-02, -1.8442e-02,  1.6899e-03,
        -1.3517e-02, -5.7744e-03, -3.8135e-03,  1.2014e-03, -7.3914e-03,
        -6.6309e-03,  1.1881e-03,  1.4123e-02,  4.4746e-03,  1.6613e-03,
        -1.2070e-02,  2.5607e-03,  2.3060e-03, -7.4371e-03,  1.3118e-02,
         7.7743e-02, -1.0700e-01, -2.4173e-02, -2.6756e-02, -4.3549e-02,
        -1.6922e-02, -1.6752e-02, -3.1473e-02,  9.9339e-03, -1.6275e-02,
        -3.5483e-03, -1.5818e-02, -1.0619e-02, -1.6508e-02, -8.6798e-04,
         5.6605e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0740, -0.0426,  0.0284, -0.0204, -0.0809,  0.0160, -0.0329,  0.0023,
        -0.0143, -0.0099,  0.0190, -0.0537,  0.0078, -0.0100, -0.1062,  0.0170,
        -0.0025,  0.0272, -0.0348, -0.0433, -0.0573, -0.0603, -0.0444,  0.0138,
        -0.0388, -0.0087, -0.0106, -0.0031,  0.0038, -0.0426, -0.0028,  0.0056,
        -0.0652,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0397, -0.1296, -0.1476, -0.0749, -0.0222, -0.0140, -0.0404,  0.0208,
         0.0022, -0.0126,  0.0091,  0.0034, -0.0113,  0.0004, -0.0160,  0.0494,
         0.0315, -0.0140, -0.1876, -0.0133,  0.0261,  0.0106,  0.0665,  0.0569,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.5509e-02, -8.7703e-02, -1.0954e-01, -2.0633e-02, -2.9966e-02,
        -9.9394e-02, -3.0586e-02, -4.4395e-02, -3.0874e-02, -8.1930e-02,
        -6.8125e-02, -4.8044e-02, -1.6297e-01, -6.3266e-03,  1.3803e-04,
        -3.9163e-03,  7.7344e-03, -2.5782e-03,  4.9091e-02, -4.3982e-03,
        -1.7533e-02,  9.7899e-03,  4.2317e-04, -4.8395e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0619, -0.0503, -0.0684, -0.1069, -0.0328, -0.0046, -0.0533, -0.0546,
        -0.0225, -0.0158,  0.0119, -0.0032, -0.0056,  0.0105, -0.0553, -0.0592,
        -0.0407,  0.0206,  0.0134, -0.0168, -0.0404, -0.0123, -0.0298, -0.0511,
         0.0034,  0.0061, -0.0542, -0.0078, -0.0025,  0.0024, -0.0067, -0.0011,
        -0.0357,  0.0385,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1042, -0.0250,  0.0069,  0.0113, -0.0497,  0.0380,  0.0038,  0.0208,
        -0.0275,  0.0145,  0.0073,  0.0607,  0.0418, -0.1478, -0.0385, -0.0414,
        -0.0453,  0.0278, -0.0097, -0.0321, -0.0816, -0.0987, -0.0522, -0.0011,
        -0.0123,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0625,  0.0152, -0.0591, -0.0210, -0.0459,  0.0202, -0.1170, -0.0358,
        -0.0836, -0.3514,  0.0472, -0.0122, -0.0280,  0.0307,  0.0044, -0.0657,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0474, -0.1825, -0.0077, -0.0543, -0.0372, -0.0146, -0.1944, -0.0090,
         0.0286, -0.0077,  0.0030, -0.0220, -0.0542, -0.0132, -0.0379, -0.0441,
        -0.0267, -0.0309, -0.0323, -0.0083,  0.0090,  0.0591, -0.0756,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1632,  0.0470,  0.0393,  0.1720,  0.3055,  0.0228, -0.0094,  0.0073,
         0.0129, -0.0274, -0.0231,  0.0625,  0.0022,  0.1054,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.2044,  0.0328,  0.0331,  0.0701, -0.0079, -0.0693,  0.0541,  0.0098,
         0.0223,  0.0144,  0.0450,  0.0269,  0.0239,  0.0208, -0.0068,  0.0082,
        -0.0379,  0.0764,  0.0023,  0.1303,  0.0249,  0.0149,  0.0233, -0.0402,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2687, -0.0423, -0.1224, -0.0397, -0.2460, -0.0311, -0.0226, -0.0044,
        -0.0300, -0.1788,  0.0139,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0310, -0.0038, -0.0157, -0.0190,  0.0171, -0.0860, -0.2204, -0.0257,
        -0.1321, -0.0299, -0.0311,  0.0004, -0.0290, -0.0788, -0.0280, -0.0165,
         0.0077, -0.1022, -0.0157,  0.0901,  0.0198,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0773, -0.0352, -0.0222,  0.0146,  0.0061, -0.0314,  0.0012, -0.2559,
        -0.0526, -0.3418,  0.0149,  0.0486,  0.0077, -0.0907,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0881, -0.0295, -0.0055,  0.0587, -0.0008, -0.0240,  0.0109, -0.0363,
        -0.0131, -0.0399, -0.0562,  0.0766,  0.3647,  0.0414,  0.0073,  0.0226,
         0.0227,  0.0791, -0.0008,  0.0200,  0.0018,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.3395e-02, -3.8885e-01,  1.1960e-02, -2.6172e-02,  5.4871e-02,
        -7.1201e-02, -8.2290e-03, -1.7484e-01,  1.1960e-04, -6.2041e-02,
        -6.2545e-02, -1.6372e-03,  5.4134e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1009, -0.0487, -0.0023, -0.0101, -0.0065, -0.0100, -0.0262, -0.0195,
        -0.0476, -0.0173, -0.0047,  0.0189,  0.0070,  0.0019, -0.0035,  0.0065,
         0.0011,  0.0135, -0.0081,  0.0345,  0.0103,  0.0024, -0.0088,  0.0008,
        -0.0940, -0.0040, -0.1145, -0.0627, -0.0693, -0.0423, -0.0663, -0.0245,
        -0.0084,  0.0099, -0.0167, -0.0156, -0.0135, -0.0019, -0.0004, -0.0040,
         0.0037,  0.0062,  0.0193, -0.0093,  0.0024], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0058, -0.0592, -0.0410, -0.0241, -0.0072, -0.0174, -0.0242, -0.0218,
        -0.0025, -0.0034, -0.0299,  0.0029, -0.0081,  0.0157, -0.0530, -0.0325,
        -0.1298, -0.0108,  0.0033,  0.0268, -0.0396, -0.0552,  0.0049, -0.1078,
         0.0222,  0.0103, -0.0198, -0.0436, -0.0781, -0.0140, -0.0196, -0.0054,
         0.0366,  0.0236,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0969,  0.0168, -0.0739, -0.0647, -0.0330,  0.0003, -0.0400, -0.0174,
         0.0277, -0.0276, -0.0131, -0.0169,  0.0270, -0.0060,  0.0173,  0.0251,
         0.0116,  0.0056, -0.0378, -0.1166, -0.2200,  0.0029, -0.0240,  0.0083,
        -0.0443, -0.0250,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0304, -0.2923,  0.0701, -0.1143, -0.0208, -0.0207, -0.0080, -0.0436,
        -0.0026, -0.0451,  0.0033, -0.0132,  0.0045,  0.0151, -0.0499, -0.0419,
        -0.0903,  0.0211, -0.0156, -0.0038, -0.0258, -0.0294,  0.0381,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0102, -0.1190, -0.0425, -0.0038, -0.0014, -0.0007, -0.0206, -0.0102,
        -0.0473, -0.0039,  0.0236,  0.0070, -0.0046, -0.0818, -0.0460,  0.0138,
         0.0127, -0.0242, -0.0743, -0.0129,  0.0098, -0.0194,  0.0121, -0.0253,
         0.0137,  0.0105, -0.0264, -0.0286, -0.0110, -0.0158,  0.0048, -0.0378,
         0.0076, -0.0031,  0.0154, -0.0005, -0.0295, -0.0012, -0.1446, -0.0223,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0293, -0.0507, -0.0188, -0.0409, -0.1126, -0.0102,  0.0055, -0.0790,
        -0.0051, -0.0208, -0.0105, -0.0119, -0.0313, -0.0249, -0.0537, -0.0642,
        -0.0461, -0.0116, -0.0147, -0.0363,  0.0024, -0.0616, -0.0402,  0.0287,
         0.0036,  0.0084, -0.0027, -0.0576,  0.0064,  0.0101, -0.0510,  0.0136,
        -0.0074, -0.0282,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0128, -0.0691,  0.0163, -0.0095, -0.0272,  0.0176, -0.0035,  0.0061,
         0.0049,  0.0359, -0.0003, -0.0053,  0.0160, -0.0050, -0.1682, -0.0109,
         0.0288,  0.0045, -0.0292,  0.0080,  0.0079, -0.0031, -0.0237,  0.0024,
        -0.0444, -0.0458, -0.0329, -0.0204, -0.0067,  0.0033,  0.0203,  0.0015,
         0.0040,  0.0258, -0.0373, -0.0085, -0.0268, -0.0269, -0.0098, -0.0313,
        -0.0258, -0.0168, -0.0212, -0.0070, -0.0027, -0.0067,  0.0060,  0.0030,
        -0.0097, -0.0170,  0.0139,  0.0083], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0756, -0.0411, -0.0413, -0.0460, -0.1186, -0.0099, -0.0299,  0.0058,
        -0.0039, -0.0049,  0.0148, -0.0098, -0.0158, -0.0703, -0.0380, -0.0476,
        -0.0481, -0.0428, -0.0644, -0.0172, -0.0313, -0.0185, -0.0209, -0.0557,
         0.0010, -0.0061, -0.0094,  0.0024,  0.0137, -0.0296,  0.0657,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.4701e-02,  1.0071e-01, -8.5044e-03, -1.1008e-03, -1.6391e-02,
        -2.6286e-02,  2.4323e-02, -1.6652e-02,  7.1841e-03,  3.3127e-03,
         1.5493e-02,  1.0121e-02,  9.7365e-03,  1.6509e-02,  4.8427e-02,
         1.9094e-02, -6.1725e-05,  6.4389e-02,  1.1882e-01,  1.4643e-01,
         6.5963e-02,  1.0594e-02,  7.1908e-02,  6.6913e-02,  1.1638e-01,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0480, -0.0275,  0.0502, -0.0023, -0.0583, -0.1947, -0.2779,  0.0392,
        -0.1285,  0.0136, -0.0127, -0.1473,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.3895e-02, -2.0527e-01, -1.6422e-02, -6.6578e-02, -6.3832e-02,
        -1.2368e-02,  3.0639e-02,  4.5353e-03, -8.6522e-03, -5.9262e-03,
         1.6576e-02,  1.4812e-02, -2.3397e-02, -1.2089e-01, -4.3659e-02,
        -3.7043e-02,  7.8906e-03,  4.4700e-02, -7.9272e-03,  3.5728e-03,
        -1.5528e-03, -8.5309e-03, -3.2137e-06, -7.4778e-02, -3.6309e-02,
        -4.6985e-02,  3.8457e-03,  4.8136e-02, -3.1274e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1893, -0.3453, -0.0159, -0.0249,  0.0117, -0.0772, -0.1041,  0.0450,
        -0.0150,  0.0359,  0.1269,  0.0088,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0986, -0.2812, -0.0949, -0.0931, -0.0433, -0.1329, -0.0889,  0.0446,
        -0.0179, -0.0094, -0.0640,  0.0313,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0025,  0.0820, -0.0960, -0.0219, -0.1059,  0.0070, -0.0141, -0.0123,
        -0.0775, -0.0505, -0.0494, -0.2099, -0.0425, -0.0376, -0.0059, -0.0780,
        -0.0223,  0.0110, -0.0711,  0.0024,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0265, -0.2914, -0.0142, -0.0353, -0.1305, -0.0306, -0.0329,  0.0076,
        -0.0143, -0.0052, -0.0124, -0.0161, -0.0126, -0.0008,  0.0131, -0.0122,
        -0.0094,  0.0062,  0.0084, -0.0035, -0.0093, -0.0120,  0.0071,  0.0182,
        -0.0160, -0.0101, -0.0103, -0.0343, -0.0424, -0.0022,  0.0014, -0.0005,
        -0.0330, -0.0722, -0.0071, -0.0148,  0.0259,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0371,  0.6058,  0.0808,  0.0235,  0.0360, -0.0196,  0.0266,  0.0188,
        -0.1517,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1616,  0.0331, -0.0330, -0.1603, -0.0041,  0.0071, -0.0315, -0.0933,
        -0.0174, -0.0032, -0.0457, -0.0472, -0.0739, -0.0712, -0.0137,  0.0377,
         0.0026,  0.0076, -0.0289, -0.0267,  0.0076,  0.0182, -0.0052, -0.0046,
        -0.0120,  0.0450,  0.0078,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0275,  0.3114,  0.0280,  0.0258,  0.0144,  0.0315,  0.0026, -0.0129,
         0.0082, -0.0056, -0.0059,  0.0131,  0.0016,  0.0250, -0.0059, -0.0238,
         0.0359,  0.0059, -0.0359, -0.0025,  0.0347,  0.0752,  0.0517, -0.0021,
         0.0121,  0.0086, -0.0056,  0.0237,  0.1304,  0.0043, -0.0040,  0.0170,
         0.0017, -0.0054,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0051,  0.0045,  0.0003,  0.0109, -0.0200, -0.0152,  0.0056, -0.0042,
        -0.0116,  0.0292, -0.0136, -0.0840, -0.0024, -0.0216, -0.0430, -0.0538,
        -0.0887, -0.0310, -0.0225, -0.0130,  0.0181, -0.0094, -0.0428, -0.0158,
         0.0037, -0.0159, -0.0096, -0.0074,  0.0002, -0.0065, -0.0085, -0.0526,
        -0.0600, -0.0218, -0.0159, -0.0252, -0.0767, -0.0116, -0.0319, -0.0067,
        -0.0067, -0.0298,  0.0008,  0.0124,  0.0298,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1217,  0.0118,  0.0102, -0.0048,  0.0888,  0.1438,  0.2755,  0.0459,
         0.0250,  0.0806, -0.0175, -0.0281,  0.0526, -0.0192, -0.0283, -0.0463,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1608, -0.0319,  0.0825,  0.0326,  0.0040,  0.0006,  0.0366,  0.0104,
         0.1416,  0.1341,  0.0583,  0.0658, -0.0220,  0.0128,  0.0306, -0.0202,
         0.0186, -0.0087,  0.0127,  0.0127,  0.0066, -0.0167,  0.0089, -0.0047,
         0.0187, -0.0189,  0.0280,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1153, -0.0187, -0.0281,  0.0009, -0.0204,  0.0033, -0.0026,  0.0006,
        -0.0326, -0.0137, -0.0018, -0.0083, -0.0176, -0.0006, -0.0105, -0.0026,
        -0.0127, -0.0060, -0.0326, -0.0307, -0.0134, -0.0227, -0.0041, -0.0868,
         0.0220, -0.0065, -0.0267, -0.0407, -0.0046, -0.0154,  0.0050, -0.0380,
        -0.0225, -0.0376, -0.0329, -0.0549, -0.0240, -0.0380,  0.0011,  0.0119,
        -0.0177, -0.0414, -0.0054, -0.0202,  0.0332,  0.0133], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.4762e-02,  3.5046e-02, -5.5343e-03, -1.2552e-02, -3.5818e-05,
        -1.6865e-02, -1.0528e-02, -1.3973e-01,  3.4659e-02, -1.2449e-02,
        -1.1612e-02,  1.7009e-02, -2.4796e-02, -8.9840e-03,  1.1880e-02,
        -3.8108e-04, -1.7596e-02, -1.1092e-02, -8.4654e-02, -2.1759e-01,
        -3.1824e-02, -2.9446e-02, -1.0826e-02,  5.7052e-03, -2.1829e-02,
        -1.2728e-02, -4.9045e-02, -6.4699e-02, -6.8050e-04, -6.5460e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2067, -0.0121,  0.0367,  0.0045,  0.1730,  0.0152, -0.0063,  0.1329,
        -0.1577, -0.0317,  0.2231,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0579,  0.3459, -0.0006,  0.0204,  0.0847,  0.0219,  0.0797,  0.0852,
         0.0120,  0.0195,  0.0031,  0.0215, -0.0623,  0.0171,  0.0248,  0.0681,
         0.0228,  0.0316,  0.0117,  0.0094,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0846, -0.3708, -0.1497, -0.0449,  0.0198, -0.0921, -0.0962,  0.0208,
        -0.0200, -0.0832,  0.0179,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1013,  0.2541,  0.0948,  0.1285,  0.0189, -0.0036,  0.0146,  0.0515,
         0.0736,  0.0037,  0.0222,  0.0243,  0.0207,  0.0029,  0.0172,  0.0438,
        -0.0162, -0.0133,  0.0104, -0.0035, -0.0808,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0315, -0.2393, -0.0404, -0.0018,  0.1248, -0.3029,  0.0285, -0.0078,
        -0.0187,  0.0049,  0.0110, -0.0126,  0.0173,  0.0015, -0.0349, -0.0181,
         0.0562, -0.0478,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0652, -0.1065, -0.1681, -0.0100, -0.0214, -0.0024, -0.0192,  0.0094,
         0.0024,  0.0017,  0.0101,  0.0078,  0.0216, -0.0347, -0.0362,  0.0147,
        -0.0195, -0.0395, -0.0185, -0.0091,  0.0015, -0.0443, -0.0711, -0.0102,
        -0.0242, -0.0267, -0.0464, -0.0153, -0.0160, -0.0407, -0.0123, -0.0103,
         0.0052,  0.0102, -0.0127,  0.0018, -0.0026,  0.0038,  0.0267,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0651, -0.3176,  0.0063, -0.0704, -0.0109, -0.0176,  0.0079, -0.0276,
         0.0098,  0.0005, -0.0230, -0.0466, -0.0158, -0.0020,  0.0147, -0.0008,
        -0.0504, -0.0132,  0.0122, -0.0095, -0.0042,  0.0115,  0.0149,  0.0074,
         0.0021, -0.0147, -0.0244, -0.0055, -0.0701, -0.0908, -0.0064, -0.0071,
        -0.0124, -0.0066,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.0856, -0.1179, -0.2376, -0.2533,  0.0150, -0.1021,  0.0079, -0.0726,
        -0.1080,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0641, -0.1860, -0.0442, -0.0264,  0.0023, -0.0208,  0.0107, -0.0129,
         0.0170,  0.0061, -0.0352, -0.0004,  0.0152, -0.0259,  0.0091, -0.0534,
        -0.0099, -0.1582, -0.0094, -0.0041,  0.0253,  0.0131, -0.0099,  0.0132,
        -0.0142, -0.0085,  0.0012,  0.0169,  0.0046,  0.0112,  0.0276, -0.0180,
         0.0141,  0.0044,  0.0072, -0.0009,  0.0096, -0.0205, -0.0344,  0.0341,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0068, -0.0489, -0.1052, -0.0039, -0.0290, -0.0022,  0.0034, -0.0224,
        -0.0191, -0.0676,  0.0036,  0.0050,  0.0022, -0.0024,  0.0086,  0.0174,
         0.0002,  0.0139, -0.0311,  0.0123, -0.0004, -0.0014, -0.0011,  0.0138,
         0.0037, -0.0242, -0.0446,  0.0131, -0.0571, -0.0770, -0.0220,  0.0083,
         0.0006, -0.0053,  0.0124,  0.0044, -0.0310, -0.0689, -0.0108, -0.0100,
        -0.0013,  0.0725, -0.0128,  0.0035,  0.0030,  0.0310, -0.0135, -0.0078,
        -0.0070, -0.0043,  0.0189,  0.0088,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0224, -0.2361, -0.0303, -0.0395, -0.0100, -0.0290, -0.0827,  0.0160,
        -0.0339,  0.0020,  0.0098, -0.0649, -0.0965, -0.0432, -0.0161, -0.0868,
        -0.0518,  0.0314, -0.0410,  0.0094,  0.0080, -0.0391,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0137,  0.3106,  0.0537,  0.0441, -0.0318,  0.0164,  0.0720,  0.0219,
        -0.0270,  0.0428,  0.0452,  0.0255,  0.0019,  0.0035, -0.0077,  0.0674,
         0.0243, -0.0129, -0.0012,  0.0786, -0.0979,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.9550e-01, -1.2762e-01, -3.2879e-03, -7.9412e-03, -1.1123e-04,
         3.0136e-03,  2.4945e-02,  1.1525e-02, -1.1699e-02,  3.0564e-02,
         1.2131e-03,  1.5693e-03,  2.3811e-03, -1.0565e-02,  1.1827e-02,
        -3.5070e-03, -6.2670e-03,  1.0425e-02,  7.5701e-04,  4.9022e-03,
         3.1481e-03, -2.9360e-03,  2.0970e-02, -4.7055e-04, -5.3855e-03,
         8.6126e-03,  3.9433e-02, -2.4292e-02,  2.1529e-02, -8.6147e-03,
        -7.8185e-02, -3.9769e-02,  1.1021e-03, -1.3654e-02, -2.3255e-02,
         1.7472e-03,  6.4969e-03,  1.5754e-02,  3.2153e-03, -4.2364e-03,
        -6.0102e-02,  3.5920e-03, -6.1897e-03,  9.1205e-06, -1.2409e-02,
         1.7780e-03, -1.9791e-02, -1.0491e-02, -1.4257e-02,  1.0267e-02,
        -4.6829e-03, -1.6402e-02,  1.6747e-02,  2.1950e-02, -8.9075e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0688, -0.0296,  0.0417, -0.0067, -0.0107, -0.0265, -0.0643, -0.0017,
        -0.0039, -0.0034, -0.0001,  0.0012,  0.0052, -0.0224, -0.0224, -0.0285,
        -0.0032, -0.0411, -0.0398,  0.0012,  0.0101, -0.0209, -0.0512, -0.0061,
         0.0112,  0.0286, -0.0224, -0.1092, -0.0125, -0.0036, -0.0015,  0.0055,
        -0.0088, -0.0683,  0.0129, -0.0097, -0.0037, -0.0223, -0.0407, -0.0179,
        -0.0079, -0.0170,  0.0195,  0.0234,  0.0428,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0882,  0.1911,  0.0928,  0.0851, -0.0137,  0.0068,  0.0446,  0.0089,
         0.0542,  0.0052,  0.0204,  0.0302,  0.0115,  0.0324,  0.0422,  0.0360,
         0.0226,  0.0721,  0.0109,  0.0031,  0.1196, -0.0022,  0.0040,  0.0022,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1272,  0.0207,  0.0114, -0.0145, -0.0121, -0.0069, -0.0194, -0.0402,
         0.0094, -0.0199,  0.0074, -0.0003,  0.0024, -0.0449, -0.0035,  0.0034,
        -0.0547, -0.0479, -0.0591, -0.0368, -0.0665, -0.0107, -0.0347, -0.0371,
         0.0368, -0.0333, -0.0032, -0.0353, -0.0615, -0.0753, -0.0097,  0.0378,
        -0.0159,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0973, -0.2559, -0.0111, -0.0701,  0.0120, -0.0356,  0.0134, -0.0534,
        -0.0682,  0.0166,  0.0192, -0.0071,  0.0511, -0.0868,  0.0300,  0.0123,
         0.0032, -0.0086, -0.1481,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1005,  0.0171, -0.0006, -0.0766, -0.1386,  0.0099, -0.0575, -0.0486,
        -0.1156, -0.0045,  0.0295,  0.0556,  0.0092, -0.0097, -0.0067, -0.0219,
        -0.0108, -0.0675,  0.0034, -0.0112, -0.0106, -0.1098, -0.0006,  0.0562,
        -0.0277,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0143, -0.2150, -0.0095, -0.0427, -0.0509, -0.0691, -0.2317,  0.0263,
        -0.0236, -0.0262, -0.0329, -0.0258, -0.0383, -0.0196, -0.0164,  0.0409,
         0.1169,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0458,  0.0037, -0.0408,  0.0028, -0.0132, -0.0006, -0.0133, -0.0466,
        -0.0301, -0.0469, -0.0101, -0.0351, -0.0133,  0.0111,  0.0194, -0.0199,
        -0.0601, -0.1101,  0.0058, -0.0219, -0.0116, -0.0994, -0.1008, -0.0033,
        -0.0485, -0.0222, -0.0412, -0.0294,  0.0417, -0.0086, -0.0016,  0.0411,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.6629e-02,  1.0090e-01,  2.0301e-01,  2.6591e-02,  3.0634e-02,
         3.1759e-02,  5.3802e-02,  1.1031e-01,  1.2861e-02, -4.8957e-03,
         1.6456e-03,  1.7892e-02,  2.3250e-03,  1.2187e-02,  4.5997e-03,
        -2.7110e-02,  2.2807e-02, -1.9622e-04,  4.4850e-02,  5.2794e-02,
         1.3901e-02, -5.3512e-03,  4.0494e-02,  6.4251e-02,  2.2260e-02,
         2.4804e-02,  3.7117e-02, -4.0196e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0672,  0.1653, -0.0053,  0.0192, -0.0162,  0.0315,  0.0175, -0.0109,
         0.0034,  0.0279,  0.0045,  0.0947,  0.0093,  0.0359,  0.0109,  0.0790,
         0.0330,  0.0964,  0.0250, -0.0104,  0.0199,  0.0217,  0.0054,  0.0314,
         0.0831,  0.0142,  0.0169, -0.0330,  0.0106,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0084, -0.0585, -0.0254, -0.0716, -0.1465, -0.0324,  0.0169, -0.0072,
        -0.0135,  0.0225,  0.0656, -0.0239, -0.0447, -0.1026,  0.0378, -0.0406,
        -0.0215, -0.0686, -0.0123, -0.0115,  0.0257,  0.0096,  0.0008, -0.0071,
        -0.0163, -0.0148, -0.0123, -0.0129, -0.0051,  0.0009, -0.0069,  0.0354,
         0.0202,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0738, -0.3559, -0.0174, -0.0099, -0.0242, -0.0389,  0.0470, -0.0090,
        -0.0056, -0.0451, -0.0248, -0.0069, -0.0367, -0.0262,  0.0010,  0.0186,
        -0.0237, -0.0019, -0.0475, -0.0218, -0.0678, -0.0052,  0.0798, -0.0116,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1064, -0.0961, -0.1163, -0.0126, -0.0225, -0.0324, -0.0332, -0.0912,
        -0.0079, -0.0472, -0.0170, -0.0302, -0.0383, -0.0439, -0.0475, -0.0033,
        -0.0044, -0.0468, -0.0029, -0.1382, -0.0078,  0.0360,  0.0181,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0129, -0.0946, -0.0136, -0.0045, -0.0135, -0.0370, -0.0248,  0.0003,
         0.0071, -0.0048,  0.0239, -0.0347, -0.0124,  0.0049, -0.0088, -0.0217,
        -0.0429, -0.0359, -0.0049,  0.0546, -0.0034, -0.0202,  0.0118, -0.0127,
        -0.0454, -0.0027, -0.0004, -0.0051, -0.0169, -0.0886, -0.1203, -0.0036,
        -0.0476,  0.0188, -0.0308, -0.0526, -0.0167, -0.0335, -0.0110,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.0643e-02, -1.1455e-01, -1.1467e-02, -8.0514e-03, -1.3003e-01,
        -1.2891e-02, -6.9046e-04,  2.7982e-03, -5.0648e-03,  8.4107e-03,
         2.7376e-03,  4.2948e-02,  6.2211e-03, -1.5214e-02,  2.1241e-02,
        -2.6427e-02, -1.0776e-02, -1.2205e-03, -6.1429e-03, -2.0503e-02,
        -3.8971e-02, -1.8585e-02, -1.3947e-02,  1.3757e-02, -4.2925e-02,
        -6.2153e-02, -2.6566e-02, -2.0563e-03,  1.3428e-02, -2.4782e-02,
        -3.4255e-03,  3.7200e-06,  1.3692e-02, -1.0070e-02,  2.8581e-04,
         4.3258e-03, -1.0262e-02, -7.3885e-02, -1.5390e-02, -9.4854e-03,
        -5.4266e-04, -2.4515e-02,  4.8925e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.4545e-03,  2.1328e-01, -1.3748e-02, -4.0243e-03,  8.6698e-04,
        -4.7454e-03,  1.7047e-02, -1.1103e-02,  3.3949e-02, -1.3054e-03,
        -1.7977e-02,  4.7056e-03,  2.3694e-02,  8.8177e-04,  7.9781e-03,
         2.5515e-02,  1.6542e-02,  4.2804e-02,  6.8344e-04,  2.0062e-02,
        -2.0484e-04, -1.4464e-03,  8.6919e-03,  5.3604e-03, -1.6247e-02,
        -1.7974e-03, -1.6390e-02, -2.0022e-03, -1.3342e-02, -1.6303e-03,
        -1.9256e-02,  5.6703e-02,  1.0032e-01,  2.6317e-02, -1.5287e-03,
        -5.1068e-03, -2.9123e-02,  1.2407e-02,  4.5511e-02,  1.2317e-02,
         1.7439e-02,  3.2732e-02, -2.3570e-03,  9.4569e-03, -2.6872e-03,
        -8.3614e-03, -4.0055e-03, -6.6605e-04,  2.2428e-03,  6.1706e-03,
        -2.2811e-02,  4.4412e-03,  4.1299e-03, -6.4233e-03,  1.7492e-02,
         1.4716e-02,  3.7990e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0163,  0.3995,  0.0643,  0.0047,  0.0297,  0.0020,  0.0557,  0.0570,
         0.0486,  0.0422,  0.0400, -0.0876,  0.0045, -0.1477,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0937, -0.1886, -0.0601,  0.0082, -0.0156, -0.0153, -0.0022, -0.0093,
        -0.0085, -0.0073, -0.0193, -0.1191, -0.0268, -0.0227, -0.0232, -0.0180,
        -0.0103, -0.0519,  0.0177, -0.0586, -0.0445, -0.0343, -0.0394, -0.0165,
         0.0298, -0.0123, -0.0059,  0.0074,  0.0140,  0.0195,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1436,  0.1076, -0.0242,  0.3155,  0.0487,  0.0904,  0.1507,  0.1192,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0031, -0.2464, -0.0272, -0.0192, -0.0275, -0.0218, -0.0004, -0.0024,
         0.0145,  0.0053,  0.0079, -0.0233, -0.0736, -0.0981,  0.0072, -0.0073,
         0.0044, -0.0043, -0.0042,  0.0088, -0.0150, -0.0097, -0.0053, -0.0164,
        -0.0835, -0.0222, -0.0335,  0.0037, -0.0405, -0.1322,  0.0083, -0.0023,
        -0.0205,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0956, -0.1397, -0.0517, -0.0422, -0.0139,  0.0043,  0.0069, -0.0026,
         0.0237, -0.0123,  0.0008,  0.0003, -0.0042, -0.0112, -0.0085, -0.0092,
         0.0013,  0.0106, -0.0139,  0.0099, -0.0057, -0.0048, -0.0030, -0.0047,
         0.0019, -0.0122, -0.0190, -0.0168, -0.0088, -0.0044, -0.0150, -0.0141,
        -0.0187, -0.0500, -0.0140, -0.0221, -0.0111,  0.0024, -0.0039,  0.0036,
         0.0084, -0.0077,  0.0017, -0.0459, -0.0292, -0.0031, -0.0113, -0.0117,
        -0.0073, -0.0045,  0.0078, -0.0024, -0.0026, -0.0097,  0.0040,  0.0059,
         0.0023, -0.0054,  0.0031, -0.0179, -0.0037, -0.0083, -0.0281, -0.0026,
        -0.0046, -0.0040, -0.0004,  0.0023,  0.0052, -0.0273, -0.0227],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.1029e-02, -1.1884e-02, -7.8591e-03,  2.5057e-04, -2.5707e-02,
         2.5725e-02, -1.7672e-01, -1.5342e-02, -2.7509e-03,  7.0959e-03,
        -2.1675e-02, -4.6399e-03, -2.0818e-04, -5.8504e-04, -2.3442e-03,
        -1.0173e-02, -1.6772e-03,  1.9366e-02, -3.2989e-03, -5.5737e-02,
        -3.5176e-02,  1.5540e-02,  2.6952e-02, -6.3274e-02,  2.9643e-02,
        -2.7630e-02, -1.2589e-02, -1.0092e-02,  7.5996e-03, -1.5418e-02,
         2.9955e-02,  1.0798e-02, -4.2832e-02, -6.9887e-02, -6.0807e-03,
        -1.6784e-02, -6.0985e-02, -1.2868e-02,  7.0415e-03,  1.9879e-05,
         4.2908e-02,  1.8556e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0128, -0.2557, -0.0538, -0.0255,  0.0068,  0.0122,  0.0039,  0.0074,
         0.0018, -0.0500, -0.0239, -0.0241, -0.0561, -0.0041,  0.0157,  0.0116,
         0.0070, -0.0475, -0.0809, -0.0192, -0.0046, -0.0118, -0.0051,  0.0065,
         0.0129, -0.0339, -0.0380, -0.0271, -0.0660,  0.0040, -0.0013,  0.0087,
        -0.0020, -0.0073,  0.0141,  0.0263, -0.0102,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0139,  0.5227,  0.0966, -0.0017, -0.0399, -0.0316, -0.0222,  0.0319,
        -0.0089,  0.0682,  0.0186,  0.0188,  0.0542,  0.0595, -0.0112,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0268, -0.0345,  0.0202,  0.0004, -0.0622, -0.0832, -0.0878, -0.0565,
        -0.1553, -0.0585,  0.0008, -0.0598, -0.0638,  0.0494,  0.0441, -0.0081,
         0.0048,  0.0565,  0.0451,  0.0824,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0385, -0.0924, -0.0733,  0.1011, -0.0846, -0.0151,  0.0004,  0.0019,
        -0.0105, -0.0087, -0.0322, -0.0364, -0.0131, -0.0028, -0.0134, -0.0234,
         0.0052, -0.0276,  0.0035, -0.0285, -0.0021, -0.0052, -0.0081,  0.0014,
        -0.0135, -0.0299, -0.0551,  0.0034, -0.0056, -0.0431, -0.0485, -0.0188,
        -0.0006, -0.0103, -0.0252, -0.0262, -0.0055, -0.0103, -0.0027,  0.0053,
         0.0259, -0.0107, -0.0299,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0129, -0.0038, -0.0151,  0.0145, -0.0461, -0.0142, -0.0174, -0.1017,
        -0.0116, -0.0122,  0.0214,  0.0217, -0.0219, -0.0622, -0.1045, -0.0382,
         0.0187,  0.0131,  0.0027, -0.0362, -0.0278, -0.0022, -0.0238,  0.0164,
        -0.0184, -0.0101, -0.0098, -0.0170, -0.0692, -0.0063, -0.0044, -0.0462,
        -0.0170, -0.0223, -0.0303, -0.0121, -0.0248,  0.0517,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0814, -0.1340, -0.0282, -0.0542, -0.0022, -0.1029,  0.0097,  0.0041,
        -0.0146,  0.0063, -0.0112, -0.0187, -0.0189,  0.0045, -0.0306, -0.0629,
         0.0111, -0.1088, -0.0117,  0.0060,  0.0828, -0.1953,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0748, -0.0554, -0.0053, -0.0077, -0.0072,  0.0100, -0.0147, -0.0015,
        -0.0474, -0.0836, -0.0133, -0.0142, -0.0093, -0.0044, -0.0122, -0.0288,
        -0.1039, -0.0271, -0.0127, -0.0257,  0.0225,  0.0066, -0.0266, -0.0264,
         0.0038,  0.0033, -0.0383, -0.0012,  0.0047, -0.0316, -0.0452,  0.0416,
         0.0109, -0.0054, -0.0052, -0.0736, -0.0019,  0.0015,  0.0833,  0.0071,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0102,  0.0150, -0.0380,  0.0120, -0.0015, -0.0408,  0.0020,  0.0011,
        -0.0256, -0.0427,  0.0062, -0.0259, -0.0132, -0.1163, -0.0821, -0.0106,
        -0.0089,  0.0141, -0.0861, -0.0013, -0.0038, -0.0037, -0.0134,  0.0045,
        -0.0035, -0.0946, -0.0070, -0.0641, -0.0946, -0.0461, -0.0382, -0.0731,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1231, -0.1921,  0.0049,  0.0120, -0.0122,  0.0186,  0.0081, -0.0341,
        -0.1229, -0.0074,  0.0047, -0.0438, -0.0144,  0.0072,  0.0055,  0.0182,
         0.0205,  0.0116,  0.0051,  0.0091, -0.0052,  0.0190,  0.0200, -0.0721,
        -0.0082,  0.0059, -0.0212, -0.0029, -0.0044, -0.0011, -0.0058, -0.0060,
         0.0004,  0.0135,  0.1389,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0665,  0.0327,  0.1102,  0.0199,  0.0610, -0.0062,  0.1198,  0.0502,
         0.0970,  0.1207,  0.0054,  0.0553,  0.0475,  0.0352,  0.0267,  0.0043,
        -0.0047,  0.0153, -0.0161,  0.0314, -0.0738,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0937, -0.2967, -0.0531, -0.0471, -0.0423, -0.0485, -0.0132, -0.0630,
        -0.1125,  0.0270,  0.0038, -0.0004,  0.0219,  0.0136, -0.0272,  0.0133,
        -0.0220,  0.0902,  0.0105,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0380, -0.1236, -0.1169,  0.0111, -0.0508, -0.0133,  0.0508, -0.0273,
        -0.0226, -0.0987, -0.1580, -0.0188, -0.0335,  0.0466, -0.0047, -0.0212,
         0.0091, -0.0275,  0.0756,  0.0519,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0099, -0.3178, -0.0031, -0.1236, -0.0007,  0.0103,  0.0012, -0.0561,
        -0.0577, -0.0345, -0.0185, -0.0093, -0.0233, -0.0663,  0.0128, -0.0546,
        -0.0543, -0.0166, -0.0360, -0.0444, -0.0126,  0.0097,  0.0265,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0813,  0.1464,  0.0421,  0.0258,  0.0122,  0.0552,  0.0124,  0.0244,
         0.0750,  0.1426,  0.0171,  0.0770,  0.0242,  0.0122, -0.0152, -0.0064,
         0.0049, -0.0012,  0.0210,  0.0221,  0.0506,  0.0038, -0.0011,  0.0763,
        -0.0493,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.0112e-02,  2.6660e-01,  2.4423e-02,  2.8403e-02, -3.3234e-03,
         1.4670e-02,  6.3140e-03,  2.4006e-02,  1.7700e-03,  5.9363e-02,
         6.3698e-02, -1.8664e-03,  1.1250e-02, -1.3744e-02, -1.7573e-02,
        -2.1288e-02,  6.5642e-03, -7.6001e-03,  7.1029e-03,  3.3450e-02,
         8.1770e-02,  6.3274e-02,  5.3138e-02,  2.6122e-02, -2.3186e-02,
         2.4068e-02,  2.2434e-02, -1.2379e-02,  3.7996e-03, -1.5505e-04,
        -1.1101e-02,  4.2594e-03,  1.3213e-02,  5.5025e-03,  2.2479e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0474,  0.0014,  0.0042, -0.1131, -0.0241, -0.0167, -0.0549, -0.0141,
        -0.1047, -0.0342, -0.0569, -0.0278,  0.0048, -0.0594, -0.1041, -0.1307,
        -0.0246, -0.0992,  0.0338, -0.0103,  0.0295, -0.0040,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0250,  0.0442,  0.0589, -0.0013,  0.0452,  0.0506,  0.0478, -0.0018,
        -0.0486,  0.1549,  0.0316,  0.0590,  0.1007, -0.0962, -0.0042,  0.0197,
         0.0216,  0.0139,  0.0528,  0.0414, -0.0806,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.9014e-03,  3.0263e-01,  1.2035e-03, -1.2457e-02,  2.7927e-02,
         1.6919e-02,  6.3663e-02,  1.3870e-01, -3.3974e-02, -1.6918e-02,
         2.1929e-02,  4.6275e-02,  1.8787e-02,  6.2338e-02, -4.4592e-03,
         9.2328e-05, -8.7581e-02, -6.0045e-02, -7.4198e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0930,  0.0136,  0.0009,  0.0085,  0.0211, -0.0470, -0.0009, -0.0091,
         0.0076, -0.0039, -0.0013,  0.0004, -0.0197, -0.0240, -0.0073,  0.0060,
        -0.0026, -0.0539, -0.0099, -0.0303, -0.0176,  0.0161, -0.0616,  0.0073,
        -0.0477, -0.0273,  0.0236, -0.0332, -0.0092, -0.1339, -0.0158, -0.0449,
        -0.0383, -0.0379,  0.0103,  0.0069,  0.0415,  0.0018,  0.0146,  0.0125,
         0.0051,  0.0013,  0.0158,  0.0021,  0.0044,  0.0005, -0.0079],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.2344e-01,  1.8552e-01,  6.0794e-02,  1.0423e-02, -1.5648e-02,
         2.4063e-03, -5.3871e-03, -1.2117e-02,  9.1205e-03,  6.7236e-03,
        -1.8927e-02, -5.2120e-03, -1.2808e-02,  1.0967e-02, -9.9033e-03,
        -1.6885e-02,  3.5959e-02, -1.0871e-02, -2.6367e-02, -5.8113e-03,
         2.4058e-02,  5.9041e-02,  6.7343e-02,  8.8305e-02,  5.5894e-03,
        -4.1133e-02,  1.2971e-02,  5.0551e-03, -1.7045e-04,  9.9372e-03,
         2.1316e-02,  2.0605e-02,  3.4430e-02,  2.4752e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1448e-02, -3.1913e-01,  2.6321e-02,  1.4055e-02, -3.2982e-02,
        -2.5310e-02,  1.3279e-02,  9.4861e-03, -3.0668e-03, -6.9061e-03,
         1.9110e-02, -8.4777e-03, -4.5929e-02, -9.7336e-02,  2.5632e-04,
         1.4349e-02,  7.4094e-05, -1.9783e-02, -4.8595e-02, -9.6365e-02,
         1.4349e-02, -6.8633e-03, -2.2141e-02,  7.8489e-04, -6.4244e-03,
        -3.0620e-03, -7.3312e-03, -2.2285e-02, -1.7404e-02, -2.0976e-02,
         1.3399e-02,  3.2000e-02, -1.4396e-02, -6.3275e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-2.9892e-02, -2.2017e-01, -1.1070e-05, -6.6458e-02, -8.2273e-02,
        -1.3826e-01,  4.3726e-04, -2.6848e-02, -3.0749e-02, -7.4174e-02,
        -7.1987e-02, -3.2968e-02, -2.3193e-02, -1.5136e-01, -7.4854e-03,
        -2.4475e-03,  3.9610e-02,  1.6739e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0626,  0.0023, -0.0003, -0.0788,  0.0091, -0.0002,  0.0361,  0.0270,
         0.0360,  0.2312, -0.0112,  0.0116,  0.0032,  0.0450,  0.0104,  0.0357,
         0.0179,  0.0118,  0.0035,  0.1534, -0.0015,  0.0044,  0.0123, -0.0317,
         0.0078, -0.0776,  0.0775,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1545,  0.0663, -0.0192, -0.0323, -0.0009, -0.0003,  0.0109,  0.0107,
         0.1087,  0.0265, -0.0066,  0.0220,  0.1648,  0.0394, -0.0031, -0.0202,
         0.0336,  0.0246,  0.0047,  0.1367,  0.0054, -0.0033,  0.0262, -0.0311,
         0.0441, -0.0038,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1871, -0.1892,  0.0229, -0.0396, -0.0748,  0.0152, -0.0063,  0.0125,
        -0.0258, -0.0291, -0.0670, -0.0332,  0.0073, -0.0122,  0.0049, -0.0396,
        -0.0177,  0.0047, -0.0031,  0.0128,  0.1951,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1038, -0.2797, -0.1760, -0.0280, -0.0420, -0.0530, -0.0822, -0.0234,
        -0.0223,  0.0422, -0.0210, -0.0349, -0.0043, -0.0140, -0.0732,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1733,  0.0288, -0.0425, -0.0398, -0.0560,  0.0353, -0.0198,  0.0800,
        -0.0037,  0.1199, -0.0145,  0.0368,  0.1850,  0.0608, -0.0118,  0.0167,
         0.0627, -0.0128,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0279, -0.2649, -0.0836, -0.0978,  0.0082,  0.0201, -0.0274, -0.0055,
        -0.0277, -0.0756, -0.0674, -0.0498,  0.0505, -0.0719,  0.0354,  0.0219,
        -0.0092, -0.0038,  0.0516,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0325, -0.1456, -0.0672, -0.0016, -0.0555, -0.0157,  0.0176, -0.0024,
         0.0010,  0.0112, -0.0028, -0.0014, -0.0127, -0.0020,  0.0030, -0.0100,
        -0.0312, -0.0096,  0.0133,  0.0202, -0.0598, -0.0063, -0.0440, -0.0135,
        -0.0736, -0.0185, -0.0341, -0.0049,  0.0147, -0.0372, -0.0086,  0.0029,
        -0.0515,  0.0029, -0.0230, -0.0088, -0.0367,  0.0270, -0.0061,  0.0102,
        -0.0136, -0.0004,  0.0109, -0.0342], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.3425e-02, -2.5698e-01, -1.4704e-02, -3.5299e-02, -3.6706e-02,
        -4.2354e-03, -1.0490e-02,  8.6076e-03, -5.8511e-02, -1.1366e-03,
        -1.1730e-02,  1.1900e-02,  9.9411e-03,  3.5385e-03,  3.1157e-02,
         6.3997e-03,  5.0669e-02, -9.2571e-02,  1.4240e-02, -4.8759e-03,
        -2.8050e-02,  1.5034e-02,  6.5437e-03,  1.3756e-03, -2.5097e-02,
         7.2752e-04,  1.5307e-04, -8.3514e-02,  1.1544e-02, -1.4141e-02,
        -1.1679e-02, -1.6102e-02, -4.1628e-03, -6.9007e-02,  5.7556e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0199, -0.3992, -0.0690, -0.0288, -0.0312, -0.0320, -0.1052, -0.0608,
         0.0115,  0.0022, -0.0060, -0.0449, -0.0482,  0.0085,  0.0100, -0.0436,
        -0.0108, -0.0125,  0.0016,  0.0015,  0.0028, -0.0011,  0.0312, -0.0177,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0141, -0.1696,  0.0047, -0.0213, -0.0322, -0.0102, -0.0180, -0.0414,
        -0.1016, -0.0170,  0.0098,  0.0004,  0.0114, -0.0254, -0.0894, -0.0068,
         0.0151, -0.0122, -0.0479,  0.0005,  0.0002, -0.0184,  0.0033, -0.0303,
        -0.0146, -0.0279, -0.0104, -0.0254, -0.0008, -0.0165, -0.0619, -0.0075,
         0.0755, -0.0584,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1909,  0.0338, -0.0051,  0.0047,  0.0191, -0.0365, -0.0648, -0.0064,
        -0.0036, -0.0013, -0.0156, -0.0201, -0.0490, -0.0017, -0.0394, -0.1610,
        -0.0008, -0.0025, -0.0215,  0.0913,  0.0087, -0.0097, -0.0202, -0.0444,
        -0.0202, -0.0112, -0.0215, -0.0282,  0.0219, -0.0449,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 6.0484e-02,  8.3265e-02, -9.7447e-02, -1.4667e-01, -5.7792e-02,
        -1.2178e-02, -1.5013e-02, -5.8531e-02, -5.5376e-02, -2.8940e-01,
         1.6668e-04,  7.6238e-02, -4.7438e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5665e-01,  3.2136e-02,  3.6786e-02,  2.8027e-02,  3.5390e-02,
        -3.6132e-02, -5.7422e-02, -9.0940e-02, -1.9128e-01,  1.9857e-02,
        -1.2773e-04, -2.8704e-03, -1.5104e-02, -2.5309e-02, -1.0824e-01,
         1.0730e-02,  1.2663e-01, -2.6363e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0513, -0.2554, -0.2327, -0.0310, -0.1960, -0.1169, -0.1135, -0.0032,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0371, -0.2215, -0.0107, -0.0119, -0.0232, -0.0212,  0.0006,  0.0003,
        -0.0043,  0.0055, -0.0022,  0.0268, -0.0006, -0.0024,  0.0016, -0.0075,
        -0.0002,  0.0044,  0.0013, -0.0060,  0.0087, -0.0135, -0.0274, -0.0219,
        -0.0135, -0.0242, -0.0165, -0.0279, -0.0681, -0.0544, -0.0060, -0.0423,
        -0.0168, -0.0657,  0.0335, -0.0429, -0.0306, -0.0016,  0.0197,  0.0075,
        -0.0026, -0.0119, -0.0068,  0.0068,  0.0109,  0.0074, -0.0217],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2366, -0.1573, -0.0038, -0.3558, -0.0953,  0.0148, -0.0211,  0.0456,
        -0.0291, -0.0011, -0.0392,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0978,  0.1937,  0.0448, -0.0270, -0.0381,  0.0575,  0.0042,  0.1629,
         0.0168,  0.0793, -0.0065, -0.0241,  0.0676,  0.0252,  0.0070,  0.0502,
        -0.0399, -0.0574,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0658, -0.1471, -0.2145, -0.0428,  0.0687, -0.0043, -0.1051, -0.0544,
        -0.0505, -0.0292, -0.0112, -0.0026, -0.0561, -0.0953, -0.0525,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1399,  0.2469,  0.0354,  0.0171,  0.0169,  0.0495,  0.1012, -0.0284,
        -0.0261,  0.0581,  0.1304,  0.0048, -0.0045, -0.0586,  0.0822,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0576,  0.0162, -0.0511,  0.0072,  0.0276, -0.0120, -0.0060,  0.0088,
        -0.0864, -0.0640, -0.0482, -0.0578, -0.1602, -0.0096, -0.0150, -0.0245,
        -0.0611,  0.0582, -0.0118,  0.0219, -0.0722, -0.0489,  0.0135,  0.0400,
        -0.0126, -0.0074,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0465, -0.0528, -0.0701, -0.0450, -0.0711, -0.0232, -0.0341, -0.1778,
        -0.2932, -0.0411, -0.0102, -0.0644, -0.0331,  0.0048,  0.0326,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1095, -0.1434, -0.0086, -0.0227, -0.0725, -0.0327, -0.0888,  0.0014,
        -0.0130, -0.1973, -0.0018,  0.0216, -0.0081, -0.0182,  0.0467,  0.0031,
        -0.0076, -0.0087, -0.0590,  0.0151, -0.0039,  0.0011,  0.0128,  0.0029,
         0.0465,  0.0529,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0337,  0.0042,  0.0287,  0.0489, -0.0175, -0.0192,  0.0249, -0.0514,
        -0.0384, -0.0928, -0.1341, -0.0297, -0.0342, -0.0732, -0.0820, -0.0237,
         0.0050, -0.0179,  0.0220, -0.0361,  0.0076, -0.0277, -0.0199, -0.0303,
         0.0398,  0.0568,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0613, -0.1711, -0.0172, -0.0054, -0.0110, -0.0063, -0.0321, -0.0554,
        -0.0174, -0.0232, -0.0206, -0.0086, -0.0743,  0.0190, -0.0222, -0.0196,
        -0.0096, -0.0016, -0.0173, -0.0071, -0.0433, -0.0336, -0.0098, -0.1001,
        -0.0210, -0.0350, -0.0424, -0.0048,  0.0045,  0.0132,  0.0032, -0.0265,
        -0.0065,  0.0015,  0.0387, -0.0155,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0373, -0.0591, -0.0376, -0.0384, -0.0615, -0.0048, -0.0635, -0.0129,
        -0.0014,  0.0027, -0.0138, -0.0039, -0.0019,  0.0030,  0.0032, -0.0043,
        -0.0036, -0.0028,  0.0176,  0.0078,  0.0424, -0.0185, -0.0127,  0.0206,
        -0.0346, -0.0975,  0.0034, -0.0319, -0.1108,  0.0007, -0.0056, -0.0662,
        -0.1024,  0.0141,  0.0148,  0.0426,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1256, -0.1247, -0.0683, -0.0516, -0.0620, -0.1487, -0.0273, -0.0675,
         0.0108, -0.0059, -0.0908, -0.0283, -0.0285, -0.0075,  0.0246,  0.0174,
        -0.0341, -0.0205, -0.0163,  0.0396,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1286, -0.0227, -0.0068, -0.0334,  0.0094,  0.0043, -0.0696, -0.0240,
         0.0016, -0.0293, -0.0062,  0.0024, -0.0094,  0.0032, -0.1122, -0.0285,
        -0.0134, -0.0520, -0.0183, -0.0284, -0.0443, -0.0069, -0.0014, -0.0103,
        -0.0782,  0.0066, -0.0150, -0.0166, -0.0132, -0.0336, -0.0229,  0.0065,
        -0.0131, -0.0842,  0.0110,  0.0280,  0.0003,  0.0041,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.7064e-05, -1.4573e-01, -3.7110e-02, -7.2929e-02, -9.8982e-03,
        -1.6049e-03, -1.6417e-04,  1.0073e-02, -1.0925e-02, -1.9144e-02,
        -5.3091e-02, -4.7728e-02, -6.6786e-03,  3.0878e-02, -5.1107e-03,
        -3.2540e-02,  2.2155e-02, -8.5251e-02, -1.2023e-02, -7.1820e-04,
         1.2550e-02,  2.7276e-02,  4.6758e-02, -8.1828e-03, -8.0468e-03,
         2.0281e-04, -7.2489e-03, -2.7317e-02, -8.3143e-02,  3.6021e-02,
        -2.4404e-02,  6.0171e-03,  2.2022e-02, -5.9685e-03, -2.0632e-02,
         4.4852e-04,  6.2184e-03,  2.2478e-02,  1.2288e-03, -2.1779e-02,
         8.2681e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0075,  0.3144, -0.0446, -0.0012,  0.0580,  0.0879,  0.0086,  0.0330,
         0.0266,  0.0170,  0.0196,  0.0718,  0.0297,  0.0105, -0.0209,  0.0484,
         0.0069,  0.0611,  0.0099,  0.0574,  0.0426, -0.0229,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0221,  0.0029, -0.0107, -0.0081,  0.0050, -0.0161,  0.0063, -0.0062,
        -0.0043,  0.0069, -0.0094, -0.0557, -0.0033, -0.0536, -0.0161, -0.0890,
         0.0033, -0.0644, -0.0354, -0.0831, -0.1395, -0.0167,  0.0485, -0.0602,
        -0.0037, -0.0026, -0.0207, -0.0331, -0.0412,  0.0447, -0.0058, -0.0523,
        -0.0194,  0.0012, -0.0086,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0180, -0.0536, -0.1487, -0.0054, -0.0898, -0.0048,  0.0141,  0.0385,
         0.0145, -0.0274, -0.0327, -0.0167,  0.0145, -0.0105, -0.0123,  0.0278,
         0.0326,  0.0158, -0.0834, -0.0526, -0.0704, -0.0172,  0.0137, -0.0199,
        -0.0052, -0.0234, -0.0288,  0.0723,  0.0354,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.2461e-02, -2.5352e-01, -8.1284e-02, -3.6507e-02, -2.9101e-02,
        -6.9783e-02,  8.3962e-04,  1.7166e-02, -2.9497e-02, -1.8722e-02,
        -3.7927e-02, -2.9985e-02, -2.9765e-02, -2.3036e-02, -2.6861e-02,
        -5.5564e-02, -2.6176e-02,  1.5515e-02, -1.1918e-02,  1.5198e-02,
        -1.5747e-04, -3.8987e-03,  1.2633e-02,  4.4132e-02,  6.8350e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1071,  0.0225, -0.0171,  0.0006, -0.1643, -0.0134, -0.0289, -0.1173,
        -0.1247, -0.0308,  0.0042,  0.0196, -0.0027,  0.0150, -0.0632, -0.1342,
         0.0263, -0.0113, -0.0172, -0.0108, -0.0233,  0.0217,  0.0238,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0774,  0.0164,  0.0070,  0.0040,  0.0138,  0.0340,  0.0168, -0.0079,
        -0.0053, -0.0080, -0.0196, -0.0287, -0.0352, -0.0503,  0.0055, -0.0014,
        -0.0104, -0.0132, -0.0107,  0.0027, -0.0049, -0.0044, -0.0016,  0.0133,
        -0.0413, -0.0362, -0.1027, -0.0027, -0.1081, -0.0423, -0.0646, -0.0256,
        -0.0407, -0.0511, -0.0288,  0.0148, -0.0251,  0.0224, -0.0010,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1017,  0.0143, -0.0704, -0.1866, -0.2387, -0.0271,  0.0229, -0.0798,
        -0.1044, -0.0038, -0.1261,  0.0242,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.1267, -0.0704, -0.1809,  0.0133, -0.0111, -0.0217, -0.0224,  0.0158,
         0.0054, -0.0180, -0.0046,  0.0025, -0.0174,  0.0168,  0.0005,  0.0066,
        -0.0088, -0.0536,  0.0014, -0.0291, -0.0221, -0.0244,  0.0170, -0.0018,
         0.0111,  0.0028,  0.0230,  0.0063,  0.0137,  0.0100,  0.0188,  0.0270,
        -0.0241, -0.0034,  0.0018, -0.0133, -0.0375, -0.0041,  0.0008,  0.0004,
         0.0163,  0.0242,  0.0186,  0.0078,  0.0427,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0965, -0.3750, -0.0467, -0.1171, -0.0296,  0.1051,  0.0050, -0.0437,
        -0.1025,  0.0024, -0.0120,  0.0645,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1637,  0.4493,  0.0043,  0.1435,  0.0649,  0.0948, -0.0305,  0.0490,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0520, -0.0874, -0.0148,  0.0123, -0.0343, -0.0145,  0.0058, -0.0019,
        -0.0250, -0.0647, -0.0017,  0.0052, -0.0040, -0.0095,  0.0004,  0.0144,
         0.0026, -0.0064,  0.0043, -0.0010, -0.0045,  0.0040, -0.0010, -0.0120,
        -0.0070, -0.0179, -0.0046, -0.0024, -0.0351, -0.0100, -0.0084, -0.0021,
        -0.0082, -0.0059,  0.0005, -0.0043, -0.0146, -0.0023, -0.0124,  0.0006,
         0.0043, -0.0070, -0.0067, -0.0046,  0.0249,  0.0265, -0.0417, -0.0120,
        -0.0454, -0.0492, -0.0230, -0.0057, -0.0162, -0.0011, -0.0235,  0.0231,
         0.0014,  0.0278, -0.0353,  0.0948, -0.0059], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0189, -0.0114,  0.0056, -0.0186, -0.0843, -0.0034, -0.0098, -0.0184,
        -0.0102,  0.0285,  0.0234, -0.0587, -0.0126, -0.0330, -0.0861,  0.0119,
         0.0150, -0.0068, -0.0298, -0.0397, -0.0609, -0.0564, -0.0028, -0.0058,
        -0.0732, -0.0183, -0.0207, -0.0242, -0.0472, -0.0278, -0.0069,  0.0347,
        -0.0948,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0724, -0.1246, -0.2127, -0.0556, -0.0162, -0.0157, -0.0662, -0.0052,
         0.0053, -0.0239,  0.0077,  0.0034, -0.0151, -0.0051, -0.0019,  0.0411,
         0.0057, -0.0141, -0.2141, -0.0013,  0.0133, -0.0205,  0.0324,  0.0263,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0161, -0.1404, -0.0405, -0.0043, -0.0361, -0.0899, -0.0249, -0.0513,
         0.0087, -0.1233, -0.1023, -0.0709, -0.0951, -0.0125,  0.0142, -0.0250,
        -0.0040, -0.0047,  0.0303,  0.0116, -0.0134,  0.0023,  0.0287, -0.0496,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0191, -0.0086, -0.0792, -0.1078,  0.0007, -0.0203, -0.0288, -0.0494,
        -0.0182,  0.0023, -0.0006,  0.0050, -0.0254,  0.0068, -0.0517, -0.0837,
        -0.0642,  0.0172, -0.0055, -0.0285, -0.0615,  0.0206, -0.0204, -0.0695,
        -0.0190, -0.0186, -0.0276, -0.0176, -0.0103,  0.0120, -0.0285,  0.0141,
         0.0374,  0.0199,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1155,  0.0373, -0.0196,  0.0529, -0.0926,  0.0695, -0.0081,  0.0002,
         0.0151,  0.0041, -0.0356, -0.0090, -0.0007, -0.0580, -0.0022, -0.0675,
        -0.1339, -0.0230, -0.0752, -0.0072, -0.0463, -0.0761, -0.0077,  0.0004,
        -0.0422,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0686, -0.0187,  0.0318, -0.0071,  0.0453,  0.0306,  0.1309,  0.0645,
         0.0975,  0.2447, -0.0139,  0.0295,  0.0389, -0.0330, -0.0551, -0.0898,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0037, -0.2215, -0.0209, -0.1113, -0.0658, -0.0222, -0.1529, -0.0275,
         0.0071,  0.0047, -0.0348,  0.0016, -0.0560, -0.0159,  0.0036, -0.0025,
        -0.0292, -0.0452, -0.0357,  0.0131,  0.0088, -0.0707,  0.0450,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0125,  0.0428,  0.0419,  0.1999,  0.3206, -0.0354, -0.0861,  0.0098,
         0.0182, -0.0849, -0.0265,  0.0886,  0.0155, -0.0171,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.1142,  0.0805,  0.0915,  0.1388,  0.0259, -0.0726,  0.0319, -0.0209,
         0.0341,  0.0218,  0.0464,  0.0367,  0.0249,  0.0138,  0.0016, -0.0003,
        -0.0215,  0.0742,  0.0023,  0.0449,  0.0110, -0.0049, -0.0300, -0.0553,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1055, -0.1112, -0.1124, -0.0146, -0.2284, -0.0560, -0.1898,  0.0153,
        -0.0451, -0.1053, -0.0165,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0552,  0.0125,  0.0390, -0.0272, -0.0384, -0.1223, -0.1520,  0.0056,
        -0.1465, -0.0631, -0.0336,  0.0023,  0.0030,  0.0141, -0.0542, -0.0264,
        -0.0027, -0.1391,  0.0299, -0.0057,  0.0272,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0858,  0.0085,  0.0013,  0.0249, -0.0763, -0.0278, -0.0553, -0.1736,
        -0.1166, -0.2590, -0.0314,  0.0048, -0.0264, -0.1083,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0207,  0.0591,  0.0160, -0.0087,  0.0145, -0.0235, -0.0051, -0.0291,
        -0.0223, -0.0060,  0.0610, -0.0250,  0.2921,  0.0475,  0.0202,  0.0227,
         0.0507,  0.0930,  0.0458, -0.0212,  0.1156,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0341, -0.3393,  0.0045, -0.0612,  0.1240, -0.0410, -0.0016, -0.1371,
        -0.0179, -0.0474, -0.0469,  0.0827,  0.0622,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0460, -0.0616, -0.0192,  0.0012, -0.0203, -0.0035, -0.0143, -0.0183,
        -0.0153, -0.0094,  0.0132, -0.0139,  0.0170, -0.0021, -0.0040, -0.0058,
         0.0109,  0.0094,  0.0098, -0.0145,  0.0407,  0.0396,  0.0128, -0.0044,
        -0.0805,  0.0036, -0.0813, -0.1063, -0.0647, -0.0411, -0.0543, -0.0193,
         0.0146,  0.0004,  0.0176, -0.0118, -0.0442,  0.0004,  0.0081, -0.0047,
        -0.0177,  0.0027, -0.0075,  0.0033,  0.0089], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0432, -0.0227, -0.0523, -0.0121,  0.0083, -0.0199, -0.0194, -0.0138,
         0.0037, -0.0211, -0.0650, -0.0070,  0.0088, -0.0053, -0.0257,  0.0182,
        -0.1260, -0.0228, -0.0163, -0.0034, -0.0222, -0.0915, -0.0110, -0.0852,
        -0.0091,  0.0318, -0.0008, -0.0282, -0.0936, -0.0225, -0.0424, -0.0020,
         0.0230, -0.0214,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0033, -0.0034,  0.0155,  0.0358,  0.0030, -0.0049,  0.0029,  0.0080,
        -0.0742,  0.0183,  0.0217, -0.0013,  0.0382, -0.0033, -0.0152, -0.0284,
         0.0057, -0.0351,  0.0569,  0.1778,  0.2265,  0.0159,  0.0422, -0.0565,
         0.0630,  0.0430,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0096, -0.3707, -0.0351, -0.1034, -0.0243, -0.0305, -0.0213, -0.0934,
        -0.0330, -0.0101, -0.0040,  0.0024, -0.0019, -0.0285, -0.0317, -0.0225,
        -0.0609,  0.0309, -0.0225, -0.0155, -0.0162,  0.0116,  0.0200,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.1424e-02, -8.8174e-02,  1.1920e-02, -6.8975e-03,  3.0642e-03,
        -2.8443e-04, -6.3019e-04, -1.9090e-02, -3.1137e-02, -1.5958e-03,
         8.8533e-03,  1.7780e-03,  1.5631e-02,  1.3243e-02, -1.6914e-01,
         3.9795e-02, -3.5849e-02, -4.5947e-02, -5.6516e-02, -3.7577e-02,
        -3.0696e-03, -4.8207e-03,  3.5574e-02,  1.7625e-02,  9.5155e-03,
        -4.4788e-03, -4.4873e-02, -8.2450e-02,  9.8082e-03, -9.3418e-06,
         1.8708e-02,  9.5329e-03,  1.0737e-02,  2.0063e-02, -1.6213e-03,
         1.3884e-02, -1.3942e-02,  3.4676e-02, -1.7962e-02, -4.8104e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0060, -0.0490, -0.0272, -0.0497, -0.1053, -0.0123,  0.0555, -0.0579,
         0.0111,  0.0237, -0.0030, -0.0190, -0.0178, -0.0224, -0.0382, -0.0656,
        -0.0345, -0.0060, -0.0016, -0.0274,  0.0173, -0.0566, -0.0123,  0.0217,
         0.0009,  0.0020,  0.0062, -0.0383, -0.0162,  0.0262, -0.0557, -0.0058,
        -0.0252, -0.0824,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0154, -0.0319, -0.0224, -0.0089, -0.0914,  0.0060,  0.0155, -0.0005,
         0.0229,  0.0051, -0.0095, -0.0209,  0.0373, -0.0047, -0.1033, -0.0065,
        -0.0216, -0.0021, -0.0473,  0.0018,  0.0024, -0.0010, -0.0268,  0.0177,
        -0.0211, -0.0520,  0.0033, -0.0201, -0.0045,  0.0050,  0.0056,  0.0105,
         0.0191,  0.0279, -0.0370,  0.0078, -0.0213, -0.0443, -0.0278, -0.0325,
        -0.0226, -0.0151, -0.0242, -0.0159, -0.0018, -0.0012, -0.0002,  0.0005,
        -0.0222, -0.0087, -0.0092, -0.0160], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0686,  0.0087, -0.0098, -0.0546, -0.1521, -0.0375, -0.0006,  0.0370,
        -0.0013, -0.0186, -0.0026, -0.0040, -0.0097, -0.1144,  0.0081, -0.0526,
        -0.0921, -0.0485, -0.0223,  0.0050, -0.0394, -0.0207, -0.0441, -0.0603,
         0.0071,  0.0101, -0.0163,  0.0058,  0.0159, -0.0165,  0.0155,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0532, -0.2229, -0.0105, -0.0237,  0.0132,  0.0214,  0.0127,  0.0145,
        -0.0179,  0.0087, -0.0156, -0.0536,  0.0071, -0.0202, -0.0465, -0.0715,
        -0.0231, -0.0488, -0.0674, -0.0797, -0.0386, -0.0184, -0.0365, -0.0308,
        -0.0433,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0108, -0.1251,  0.0180,  0.0404,  0.0880,  0.2335,  0.4055,  0.0295,
         0.0256, -0.0112, -0.0068, -0.0057,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0448, -0.2940, -0.0082, -0.0670, -0.0481, -0.0055, -0.0354,  0.0047,
        -0.0018,  0.0162,  0.0009, -0.0321, -0.0301, -0.0660,  0.0157, -0.0467,
        -0.0128, -0.0219,  0.0035,  0.0260, -0.0264, -0.0205,  0.0086,  0.0006,
         0.0417, -0.0764, -0.0125, -0.0119,  0.0197,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1068, -0.2344,  0.1255,  0.0545, -0.0364, -0.0725, -0.2029,  0.0647,
        -0.0519, -0.0216, -0.0076, -0.0212,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0497, -0.3679, -0.0071, -0.1006,  0.0352, -0.1173, -0.1762,  0.0301,
        -0.0529,  0.0132, -0.0333, -0.0166,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0451,  0.0557, -0.0953,  0.0145, -0.2017,  0.0539,  0.0084, -0.0351,
        -0.1331, -0.0384, -0.0014, -0.0935, -0.0297,  0.0167,  0.0103, -0.0698,
         0.0165,  0.0505,  0.0126, -0.0178,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0018, -0.2199,  0.0070, -0.0224, -0.0916, -0.0326, -0.0112,  0.0036,
        -0.0366,  0.0038,  0.0060, -0.0057, -0.0036, -0.0176,  0.0701,  0.0161,
        -0.0340,  0.0102, -0.0039, -0.0155, -0.0229,  0.0287, -0.0137, -0.0075,
        -0.0184,  0.0165, -0.0081, -0.0413, -0.0358,  0.0427,  0.0237,  0.0043,
        -0.0240, -0.0399, -0.0129, -0.0070,  0.0395,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.0216e-01,  5.0402e-01,  3.4512e-02, -8.5155e-02,  2.7051e-02,
         1.8132e-04,  1.6269e-02,  9.5012e-02,  3.5642e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0474,  0.0403,  0.0127, -0.2102, -0.0395,  0.0033, -0.0818, -0.1121,
        -0.0046, -0.0103, -0.0332, -0.0139, -0.0215, -0.0937, -0.0061,  0.0008,
         0.0090,  0.0194, -0.0278, -0.0411, -0.0131,  0.0178,  0.0091,  0.0022,
         0.0118,  0.0893, -0.0278,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0596,  0.2534, -0.0154,  0.0684,  0.0035,  0.0276, -0.0063,  0.0068,
         0.0085, -0.0268, -0.0236,  0.0554,  0.0054,  0.0014, -0.0212, -0.0104,
         0.0355, -0.0102, -0.0212, -0.0028, -0.0069,  0.0338,  0.0252, -0.0202,
        -0.0065, -0.0183,  0.0180,  0.0333,  0.1182,  0.0125, -0.0141, -0.0020,
         0.0124, -0.0155,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0376,  0.0045,  0.0027, -0.0041, -0.0183, -0.0407,  0.0011, -0.0024,
         0.0357,  0.0352, -0.0274, -0.0623,  0.0031, -0.0271,  0.0171, -0.0303,
        -0.1005, -0.0009, -0.0175, -0.0063, -0.0158, -0.0118, -0.0141, -0.0042,
         0.0028, -0.0043,  0.0100,  0.0108,  0.0149, -0.0262,  0.0170, -0.0715,
        -0.0777, -0.0225, -0.0294, -0.0231, -0.0406, -0.0035,  0.0065,  0.0004,
        -0.0117, -0.0555,  0.0117,  0.0236,  0.0154,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0339, -0.0816, -0.0475, -0.0128, -0.0623, -0.1388, -0.2128, -0.0041,
         0.0064, -0.0477,  0.0833, -0.0202, -0.1069, -0.0376, -0.0086, -0.0954,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0561, -0.0053,  0.0171,  0.0146,  0.0288, -0.0042, -0.0760,  0.0403,
         0.1799,  0.1723,  0.0359,  0.1087,  0.0261,  0.0372,  0.0214, -0.0029,
         0.0092, -0.0136, -0.0136, -0.0080, -0.0044, -0.0043, -0.0136, -0.0162,
         0.0148,  0.0188,  0.0566,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0233, -0.0230, -0.0573,  0.0026, -0.0200,  0.0069, -0.0210,  0.0004,
        -0.0298, -0.0103,  0.0100, -0.0023, -0.0129, -0.0011, -0.0049,  0.0020,
        -0.0163,  0.0051, -0.0219, -0.0203,  0.0222, -0.0552,  0.0106, -0.0735,
         0.0031, -0.0090, -0.0445, -0.0791,  0.0212, -0.0584, -0.0086, -0.0595,
        -0.0321, -0.0120, -0.0542, -0.0619, -0.0189, -0.0230,  0.0032, -0.0145,
        -0.0094, -0.0246, -0.0013, -0.0014,  0.0007, -0.0071], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0065,  0.0481,  0.0148, -0.0257, -0.0019, -0.0039, -0.0040, -0.0941,
         0.0283, -0.0489, -0.0044, -0.0341,  0.0219, -0.0019, -0.0143, -0.0008,
        -0.0058, -0.0308, -0.0925, -0.2059, -0.0351, -0.0507, -0.0124, -0.0381,
         0.0191, -0.0232, -0.0695, -0.0118,  0.0475,  0.0039,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1680,  0.0695,  0.0898, -0.0544,  0.0011,  0.0950, -0.0314,  0.1971,
         0.1166,  0.1048,  0.0723,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0991, -0.3291, -0.0177,  0.0073, -0.0849, -0.0420, -0.0618, -0.0586,
         0.0035, -0.0147,  0.0165,  0.0158, -0.0828,  0.0033, -0.0345, -0.0394,
        -0.0297, -0.0346,  0.0118,  0.0128,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0091, -0.2819, -0.2153, -0.0667,  0.0528, -0.1223, -0.0941,  0.0101,
        -0.0555, -0.0755,  0.0168,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0534, -0.2061, -0.0657, -0.2055,  0.0096,  0.0019, -0.0198, -0.0233,
        -0.1070, -0.0122, -0.0070, -0.0245, -0.0203, -0.0037, -0.0088, -0.0768,
        -0.0319,  0.0033, -0.0062,  0.0219,  0.0911,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0226, -0.3102, -0.1540, -0.0164,  0.0367, -0.1818,  0.1042, -0.0242,
        -0.0297,  0.0074,  0.0097,  0.0004,  0.0279,  0.0011, -0.0107, -0.0442,
        -0.0029,  0.0158,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0814, -0.1333, -0.1537, -0.0008, -0.0167,  0.0017, -0.0552,  0.0075,
        -0.0051,  0.0002,  0.0130,  0.0116, -0.0365, -0.0346, -0.0450,  0.0061,
        -0.0099, -0.0299, -0.0168,  0.0090, -0.0037, -0.0201, -0.0381,  0.0072,
        -0.0329,  0.0122, -0.0392, -0.0150, -0.0186, -0.0678,  0.0076, -0.0033,
        -0.0026,  0.0144, -0.0134,  0.0007,  0.0097,  0.0205, -0.0049,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.1136e-02, -3.2946e-01,  7.0114e-03, -5.2146e-02, -1.3026e-02,
        -7.1120e-03,  1.6816e-02,  2.8070e-02, -3.0527e-02, -2.5822e-02,
        -1.5542e-02, -4.7284e-02, -1.8085e-02,  4.3370e-03, -5.2682e-02,
        -2.2390e-02, -4.8880e-02, -2.2854e-02,  1.5669e-02, -2.4975e-02,
        -3.0829e-03,  1.9534e-03,  4.3766e-05,  1.3803e-03, -8.3323e-03,
        -9.6093e-03, -1.2590e-02, -6.3552e-03, -4.3332e-02, -4.9893e-02,
        -7.8440e-03, -1.8221e-02, -2.1645e-02, -2.1892e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.1841, -0.1811, -0.1693, -0.2126, -0.0569, -0.1017, -0.0097,  0.0279,
        -0.0567,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.7806e-02, -1.7237e-01, -1.8589e-02, -3.7307e-02, -8.5999e-03,
        -7.4327e-02, -1.4492e-03,  1.8887e-03, -7.7375e-03, -4.5847e-02,
        -2.0598e-02, -3.8470e-03, -9.1458e-03, -7.5121e-04,  3.1085e-03,
        -5.0240e-02,  1.3140e-02, -1.6196e-01,  6.7718e-03, -2.7751e-03,
        -2.9619e-05,  1.8083e-02,  2.8004e-02,  1.6176e-02, -5.1786e-02,
        -4.5716e-03, -3.2990e-03, -3.5419e-03, -8.9937e-04, -3.4388e-03,
         1.6106e-04, -5.4414e-02,  7.4572e-03, -3.7642e-02,  4.2069e-02,
        -1.4788e-03,  1.5112e-02, -5.9496e-03,  9.4075e-03, -1.8218e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0603, -0.0765, -0.1168, -0.0070, -0.0378, -0.0023,  0.0025, -0.0053,
        -0.0296, -0.0002, -0.0027, -0.0189, -0.0030, -0.0142,  0.0120, -0.0110,
         0.0187,  0.0836,  0.0003, -0.0088,  0.0017,  0.0055, -0.0215,  0.0370,
        -0.0114, -0.0167, -0.0351, -0.0059, -0.0420, -0.0522, -0.0066, -0.0039,
         0.0019, -0.0011,  0.0102,  0.0104, -0.0150, -0.0104,  0.0072,  0.0121,
         0.0099,  0.0330,  0.0025,  0.0025, -0.0052,  0.0233, -0.0056,  0.0115,
         0.0046,  0.0157,  0.0369,  0.0301,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0366,  0.2331,  0.0309,  0.0187,  0.0276,  0.0442,  0.0949,  0.0373,
         0.0123,  0.0074, -0.0413,  0.0257,  0.0507,  0.0434,  0.0167,  0.0487,
         0.0144, -0.0236,  0.0524,  0.0662,  0.0072,  0.0667,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.3161e-05,  5.3146e-01,  2.8540e-02, -1.3873e-02,  4.7938e-02,
        -1.2382e-02, -5.5818e-03,  4.7934e-02,  3.8208e-02, -2.0616e-02,
        -8.9741e-03, -8.7312e-03, -6.2521e-03, -1.8056e-02, -1.5365e-02,
         8.2249e-03, -1.9358e-02, -2.0047e-02,  2.1641e-04, -7.3613e-02,
        -7.4580e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0096, -0.0736, -0.0401, -0.0236, -0.0135, -0.0088, -0.0133,  0.0121,
        -0.0175, -0.0054, -0.0030,  0.0041, -0.0091, -0.0036,  0.0036,  0.0029,
         0.0058,  0.0106, -0.0056,  0.0034,  0.0123, -0.0021,  0.0018,  0.0107,
        -0.0089, -0.0016, -0.0128, -0.0313, -0.0151, -0.0353, -0.0803, -0.0401,
        -0.0079, -0.0238, -0.0329, -0.0151,  0.0050, -0.0121, -0.0085, -0.0033,
        -0.1648,  0.0064, -0.0017, -0.0186, -0.0439,  0.0053, -0.0177, -0.0102,
        -0.0022,  0.0132, -0.0152, -0.0101,  0.0143, -0.0127, -0.0333],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2021, -0.0067,  0.0049, -0.0172, -0.0019, -0.0133, -0.0639,  0.0052,
         0.0373,  0.0025,  0.0152, -0.0057, -0.0013, -0.0196, -0.0408, -0.0048,
         0.0078,  0.0016, -0.0067, -0.0118,  0.0045, -0.0068, -0.0185,  0.0116,
        -0.0112,  0.0103, -0.0034, -0.0221,  0.0023,  0.0025, -0.0337, -0.0118,
        -0.0158, -0.0853,  0.0108, -0.0342, -0.0147, -0.0285, -0.0361, -0.0297,
        -0.0107, -0.0521,  0.0201,  0.0160,  0.0371,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0327, -0.2655, -0.0963, -0.0713,  0.0019, -0.0093, -0.0246, -0.0095,
        -0.0302, -0.0083,  0.0009, -0.0223,  0.0218, -0.0442, -0.0611,  0.0232,
         0.0144, -0.0342, -0.0129, -0.0215, -0.0589,  0.0322, -0.0385,  0.0646,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0071, -0.0095,  0.0157,  0.0108, -0.0213, -0.0178, -0.0580, -0.0421,
         0.0096,  0.0354, -0.0087, -0.0129,  0.0087, -0.0500, -0.0026, -0.0080,
        -0.0731, -0.0291, -0.0291, -0.0741, -0.0210, -0.0326, -0.0567, -0.0804,
         0.0021, -0.0628, -0.0769, -0.0221, -0.0184, -0.0823, -0.0028,  0.0157,
        -0.0028,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0642, -0.2601, -0.0805, -0.0957,  0.0071, -0.0419,  0.0139, -0.0181,
        -0.1618,  0.0073, -0.0141, -0.0148, -0.0225, -0.0349,  0.0217, -0.0132,
        -0.0465,  0.0250, -0.0568,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1979, -0.0286,  0.0189, -0.0632, -0.0959,  0.0027, -0.0607, -0.0502,
        -0.1607, -0.0114,  0.0081,  0.0019,  0.0058,  0.0510, -0.0131, -0.0096,
        -0.0262, -0.0796, -0.0071,  0.0037, -0.0365, -0.0143, -0.0037,  0.0366,
         0.0127,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.1647e-02, -1.0006e-01, -4.7101e-02, -3.6979e-02, -6.6297e-02,
        -1.1895e-01, -2.9648e-01,  5.4962e-02, -1.6215e-02, -2.0456e-02,
        -1.0776e-02,  3.5312e-02, -2.4202e-02, -1.9228e-02,  9.8361e-04,
         6.0258e-02, -9.7706e-05,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.0552, -0.0109, -0.0299,  0.0074, -0.0022, -0.0087,  0.0131, -0.0623,
        -0.0219, -0.0323, -0.0195, -0.0048,  0.0153,  0.0205,  0.0126, -0.0190,
        -0.0379, -0.1446,  0.0090, -0.0855,  0.0183, -0.0853, -0.1334, -0.0015,
        -0.0250, -0.0012,  0.0138,  0.0153,  0.0051,  0.0171, -0.0053,  0.0660,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0848, -0.1199, -0.1079, -0.0363, -0.0102, -0.0163, -0.0441, -0.1081,
        -0.0009,  0.0090, -0.0042,  0.0026, -0.0608, -0.0134, -0.0285,  0.0083,
        -0.0198, -0.0012, -0.0790, -0.0081, -0.0100,  0.0113, -0.0362, -0.0892,
         0.0072, -0.0187,  0.0439,  0.0201,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0077, -0.1128,  0.0266, -0.0243, -0.0275, -0.0539, -0.1239, -0.0167,
        -0.0140, -0.0095, -0.0114, -0.0751, -0.0196,  0.0007, -0.0240, -0.0520,
        -0.0367, -0.0552, -0.0451,  0.0474, -0.0177, -0.0138, -0.0133,  0.0002,
        -0.0370, -0.0151, -0.0272,  0.0082,  0.0832,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0072,  0.0046, -0.0105, -0.0440, -0.1460, -0.0033,  0.0036, -0.0010,
        -0.0174,  0.0283,  0.0573, -0.0155, -0.0351, -0.1132,  0.0580, -0.0602,
         0.0094, -0.1244, -0.0058, -0.0467, -0.0292,  0.0094, -0.0390,  0.0103,
        -0.0068, -0.0058, -0.0134, -0.0071, -0.0114,  0.0018,  0.0106,  0.0625,
        -0.0013,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.8047e-02, -2.1066e-01, -1.3134e-02, -2.8210e-02, -2.1777e-02,
        -4.4901e-02,  2.0908e-02, -2.3151e-02, -1.6497e-02, -6.6222e-02,
        -4.4945e-02,  8.1501e-03,  6.6314e-04, -1.9941e-02, -2.1348e-02,
         6.4966e-03, -2.9753e-02, -1.6459e-02, -1.2648e-01, -4.0646e-05,
        -8.1133e-02, -1.3551e-02, -1.7488e-02,  7.0045e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.9653e-02, -1.9513e-01, -1.3750e-01, -2.3943e-02, -3.8074e-02,
        -2.3762e-02, -4.4478e-02, -6.5213e-02,  3.5905e-02, -4.7366e-02,
        -2.0014e-02, -3.3064e-02, -3.9594e-02, -1.1745e-02, -2.3073e-02,
        -3.0465e-03,  2.3551e-05, -2.7889e-02,  5.5780e-03, -1.1935e-01,
        -2.1300e-02,  9.9454e-03, -4.3567e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0684, -0.1493, -0.0100,  0.0111, -0.0178, -0.0402, -0.0018, -0.0162,
        -0.0114, -0.0202,  0.0007, -0.0206, -0.0187,  0.0145, -0.0042, -0.0764,
        -0.0781, -0.0289, -0.0359, -0.0094, -0.0112, -0.0125, -0.0154, -0.0137,
        -0.0360, -0.0137, -0.0042,  0.0007, -0.0023, -0.0612, -0.0249, -0.0272,
        -0.0103,  0.0074, -0.0234, -0.0179, -0.0091,  0.0310, -0.0444,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0408, -0.0299, -0.0054, -0.0281, -0.0961, -0.0109,  0.0205,  0.0074,
        -0.0119,  0.0277,  0.0104, -0.0024,  0.0084, -0.0025,  0.0155, -0.0295,
        -0.0129,  0.0109,  0.0041,  0.0152, -0.0342, -0.0002, -0.0222, -0.0261,
        -0.0916, -0.1217, -0.0373, -0.0253, -0.0107, -0.0324, -0.0147, -0.0233,
        -0.0338, -0.0129, -0.0176,  0.0040, -0.0023, -0.0441,  0.0028, -0.0015,
        -0.0053, -0.0087,  0.0366,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0086, -0.3154, -0.0247,  0.0028,  0.0252,  0.0191,  0.0136,  0.0026,
        -0.0699,  0.0053,  0.0215, -0.0005, -0.0194, -0.0005,  0.0012,  0.0052,
        -0.0096, -0.0515,  0.0106, -0.0020,  0.0007, -0.0036, -0.0022,  0.0004,
         0.0027,  0.0034,  0.0056,  0.0070, -0.0154,  0.0066, -0.0145, -0.0459,
        -0.0814, -0.0004, -0.0120,  0.0156,  0.0099, -0.0227, -0.0207, -0.0019,
        -0.0140, -0.0353, -0.0018, -0.0010,  0.0012, -0.0035, -0.0055,  0.0008,
         0.0123, -0.0025,  0.0115, -0.0021, -0.0033,  0.0047, -0.0028, -0.0008,
        -0.0151], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.2994e-02,  3.6564e-01,  5.4570e-02,  3.4436e-04,  5.2123e-02,
        -1.1122e-03,  7.1813e-02,  1.4376e-01, -9.4705e-03,  8.5917e-02,
         8.8381e-02,  1.7024e-02, -3.4696e-03, -2.3377e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0284, -0.1378,  0.0263, -0.0022, -0.0132, -0.0139, -0.0019, -0.0046,
        -0.0205, -0.0047, -0.0254, -0.1734, -0.0216, -0.0024, -0.0245, -0.0335,
        -0.0134, -0.0357,  0.0348, -0.0690, -0.1086, -0.0383, -0.0252, -0.0187,
         0.0146, -0.0083,  0.0076, -0.0017,  0.0062, -0.0838,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0258, -0.2185, -0.0479, -0.2419, -0.3012, -0.0977,  0.0118, -0.0554,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0023, -0.1085, -0.0118, -0.0060, -0.0558, -0.0614, -0.0185, -0.0100,
        -0.0059, -0.0197,  0.0167,  0.0166, -0.0417, -0.0494,  0.0024, -0.0098,
         0.0029,  0.0055,  0.0100,  0.0054, -0.0089,  0.0025,  0.0057, -0.0586,
        -0.0431, -0.0321, -0.0475,  0.0105, -0.0703, -0.0753, -0.0275, -0.0544,
        -0.1032,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0123, -0.1748, -0.1003, -0.0109,  0.0134,  0.0073, -0.0148, -0.0191,
         0.0084, -0.0108,  0.0051,  0.0095, -0.0002,  0.0052,  0.0215,  0.0050,
        -0.0032,  0.0171,  0.0031,  0.0026,  0.0027,  0.0054, -0.0049,  0.0077,
         0.0044, -0.0121, -0.0145, -0.0662, -0.0017, -0.0108, -0.0100, -0.0172,
        -0.0279, -0.0416, -0.0020,  0.0055, -0.0196,  0.0098,  0.0023,  0.0012,
         0.0089,  0.0025,  0.0025, -0.0240,  0.0036,  0.0036, -0.0037,  0.0009,
        -0.0047, -0.0010,  0.0145, -0.0183, -0.0040, -0.0106, -0.0126,  0.0073,
        -0.0009,  0.0172, -0.0076, -0.0245, -0.0046, -0.0083, -0.0466, -0.0050,
         0.0002,  0.0101,  0.0023,  0.0168,  0.0107, -0.0016,  0.0088],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0252, -0.0195, -0.0195, -0.0145, -0.0373, -0.0147, -0.0726,  0.0047,
        -0.0013,  0.0108, -0.0145,  0.0146, -0.0080,  0.0005, -0.0060,  0.0109,
        -0.0028,  0.0284, -0.0378, -0.1311, -0.0472, -0.0249,  0.0336, -0.0502,
         0.0298, -0.0132, -0.0272,  0.0019,  0.0037, -0.0098, -0.0076, -0.0060,
        -0.0521, -0.0840, -0.0225, -0.0109, -0.0398, -0.0097,  0.0013, -0.0089,
         0.0135, -0.0276,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0216, -0.2340,  0.0252, -0.0225, -0.0076,  0.0386, -0.0114,  0.0344,
        -0.0104, -0.0414, -0.0449, -0.0307, -0.0447,  0.0141,  0.0017,  0.0065,
        -0.0232, -0.0576, -0.0573, -0.0139, -0.0109, -0.0091,  0.0049,  0.0125,
        -0.0014, -0.0568, -0.0216, -0.0113, -0.0459,  0.0045,  0.0067,  0.0003,
        -0.0011, -0.0044, -0.0067, -0.0331, -0.0270,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0527, -0.3474, -0.0409, -0.0817,  0.0144,  0.2063, -0.0440, -0.0239,
        -0.0193, -0.0294, -0.0056, -0.0064, -0.0240, -0.0873,  0.0166,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0332, -0.0467,  0.0432,  0.0236, -0.0361, -0.1381, -0.0249, -0.0262,
        -0.2010, -0.0042,  0.0636,  0.0397, -0.0933,  0.0128,  0.0286,  0.0128,
         0.0046,  0.0514,  0.1000, -0.0157,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1453e-02, -6.2881e-03, -4.2932e-03, -1.4636e-02, -1.2224e-01,
        -3.9859e-02, -1.2669e-02, -1.6459e-02,  1.0397e-03, -2.5971e-02,
        -4.1307e-02, -8.9168e-02, -4.9125e-02, -2.2403e-02, -3.1984e-02,
        -2.4969e-02, -5.3759e-03, -3.2629e-02,  2.3214e-02, -4.5600e-02,
        -8.7623e-03, -8.0396e-03, -1.7906e-02, -3.0988e-03, -1.7985e-05,
        -4.6351e-02, -2.4754e-02, -1.2188e-03, -1.2785e-02, -3.2888e-02,
        -7.7821e-02, -7.2825e-03,  1.1810e-02, -9.3021e-04, -2.9138e-02,
        -3.4230e-02, -2.3645e-02, -8.7878e-03,  3.8947e-03, -2.0007e-04,
        -4.8947e-03,  3.6969e-03, -1.7168e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0432, -0.0035,  0.0150, -0.0556, -0.0409, -0.0317, -0.0342, -0.0442,
        -0.0295, -0.0080, -0.0006, -0.0061, -0.0249, -0.0705, -0.1132, -0.0175,
        -0.0030, -0.0026,  0.0111, -0.0238, -0.0508, -0.0148,  0.0120,  0.0031,
        -0.0331, -0.0171, -0.0091, -0.0181, -0.0699,  0.0005, -0.0105, -0.0443,
        -0.0134, -0.0349, -0.0368,  0.0050,  0.0136,  0.0342,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1162, -0.2697, -0.0620, -0.0951,  0.0289, -0.0975,  0.0220, -0.0034,
        -0.0016, -0.0807, -0.0063, -0.0299, -0.0709, -0.0046,  0.0201,  0.0217,
        -0.0105, -0.0318,  0.0008, -0.0090,  0.0072,  0.0101,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0818, -0.1124,  0.0176,  0.0075,  0.0103, -0.0010, -0.0014, -0.0035,
         0.0025, -0.0996, -0.0046,  0.0101, -0.0198,  0.0245, -0.0302, -0.0249,
        -0.1071, -0.0104, -0.0080, -0.0341,  0.0104,  0.0095, -0.0222, -0.0196,
        -0.0157, -0.0019, -0.0101,  0.0019, -0.0076, -0.0574,  0.0127,  0.0059,
         0.0091,  0.0172, -0.0166, -0.0908,  0.0155,  0.0141, -0.0178,  0.0326,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0502, -0.0172, -0.0074, -0.0017, -0.0081, -0.0226,  0.0026,  0.0024,
        -0.0456, -0.0196, -0.0034, -0.0099, -0.0097, -0.0878, -0.0894, -0.0227,
         0.0227, -0.0040, -0.0915, -0.0219, -0.0196, -0.0134,  0.0239, -0.0102,
        -0.0076, -0.0909, -0.0037, -0.0678, -0.0737, -0.0334, -0.0214, -0.0942,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0540, -0.3062,  0.0025, -0.0056, -0.0283, -0.0081,  0.0071, -0.0466,
        -0.1364,  0.0023,  0.0012, -0.0185, -0.0434, -0.0019,  0.0124,  0.0034,
        -0.0325, -0.0156,  0.0026, -0.0017, -0.0122,  0.0069, -0.0034, -0.0164,
         0.0119,  0.0261, -0.0177, -0.0054,  0.0269, -0.0096, -0.0017,  0.0100,
         0.0044,  0.0362, -0.0807,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0705, -0.0153, -0.0689, -0.1228,  0.0148, -0.0389, -0.1485, -0.1068,
         0.0266, -0.1387, -0.0071, -0.0517,  0.0023, -0.0546, -0.0317,  0.0129,
        -0.0124, -0.0390, -0.0044,  0.0128, -0.0192,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.8588e-02, -2.7357e-01,  8.9346e-02, -7.2041e-02, -9.7877e-02,
        -3.0110e-02, -7.7926e-03, -5.9947e-02, -7.0395e-02,  6.4286e-02,
         1.6758e-02,  9.7064e-03, -2.1348e-04,  2.1546e-02, -3.1835e-02,
        -1.2535e-03,  7.8208e-03, -4.4600e-02,  1.2313e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0602, -0.1252, -0.0937,  0.0082, -0.0315,  0.0013, -0.0007, -0.0127,
        -0.0178, -0.1138, -0.1155, -0.0628, -0.0341, -0.1991,  0.0033, -0.0318,
        -0.0046, -0.0364,  0.0317, -0.0156,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0350, -0.3379,  0.0333, -0.0693,  0.0184, -0.0255, -0.0632, -0.0456,
        -0.0130, -0.0129, -0.0177, -0.0076,  0.0071, -0.0459, -0.0390, -0.0376,
        -0.0781, -0.0075, -0.0303, -0.0450,  0.0059, -0.0140,  0.0103,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0486,  0.1385,  0.0907,  0.0084, -0.0116,  0.0218, -0.0119,  0.0402,
         0.0298,  0.1045,  0.0077,  0.0881,  0.0489,  0.0772, -0.0391, -0.0159,
         0.0006,  0.0166,  0.0437,  0.0133,  0.0201,  0.0113, -0.0349,  0.0597,
        -0.0169,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0297,  0.2958,  0.0098,  0.0152,  0.0077,  0.0071,  0.0452,  0.0306,
        -0.0279, -0.0638,  0.0411, -0.0199,  0.0060, -0.0014, -0.0025,  0.0194,
         0.0160,  0.0042,  0.0273, -0.0122,  0.0395,  0.0264,  0.0320,  0.0172,
        -0.0018,  0.0012,  0.0066,  0.0073,  0.0099, -0.0029,  0.0144,  0.0084,
         0.0238,  0.0420, -0.0836,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0336, -0.0177,  0.0024, -0.1176, -0.0199, -0.0377, -0.0952, -0.0428,
        -0.0825, -0.0124, -0.0005,  0.0097,  0.0220, -0.0439, -0.0496, -0.1692,
        -0.0220, -0.0684,  0.0313,  0.0119,  0.0662,  0.0437,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1106,  0.1106, -0.0022,  0.0157,  0.0818,  0.0717,  0.0318,  0.0295,
         0.0061,  0.1320,  0.0135,  0.0707,  0.0027, -0.0681,  0.0800,  0.0797,
         0.0204,  0.0016,  0.0092,  0.0220, -0.0400,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0300, -0.3583, -0.0192,  0.0233,  0.0096,  0.0178, -0.0636, -0.1061,
         0.0054, -0.0592, -0.0422, -0.0384, -0.0061, -0.0376,  0.0141, -0.0556,
         0.0384,  0.0417, -0.0336,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0266,  0.0030, -0.0125,  0.0203,  0.0262, -0.0415,  0.0089,  0.0014,
        -0.0038, -0.0084, -0.0218, -0.0196, -0.0117, -0.0396,  0.0001, -0.0049,
         0.0004, -0.0318, -0.0122, -0.0455, -0.0337,  0.0408, -0.0205, -0.0165,
        -0.0508, -0.0343, -0.0146, -0.0511,  0.0023, -0.1005, -0.0050, -0.0550,
        -0.0434, -0.0589,  0.0153, -0.0086, -0.0021, -0.0004,  0.0265, -0.0115,
         0.0095,  0.0136,  0.0110,  0.0008, -0.0036,  0.0265, -0.0031],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0439,  0.2077,  0.0811,  0.0171,  0.0047,  0.0167, -0.0018,  0.0066,
         0.0236,  0.0103, -0.0128,  0.0312,  0.0022, -0.0327, -0.0296, -0.0032,
         0.0492, -0.0002, -0.0144,  0.0233, -0.0057,  0.0224,  0.0450,  0.0921,
        -0.0165, -0.0282,  0.0660,  0.0060, -0.0107, -0.0280,  0.0070, -0.0004,
        -0.0178,  0.0420,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.1167e-02, -3.4923e-01, -1.8932e-02,  1.9198e-02, -2.3888e-02,
        -4.5741e-02,  1.3260e-02,  2.5333e-02, -8.0870e-03, -1.5739e-02,
         1.4106e-02, -9.0246e-04, -3.6900e-02, -6.8294e-02, -3.0360e-05,
        -1.9966e-02, -2.5210e-02,  4.1912e-03,  2.5151e-02, -2.8343e-02,
         9.2234e-05,  1.1852e-02, -1.2241e-02,  4.7720e-03, -7.2745e-03,
         3.3072e-03, -1.2716e-03, -8.8490e-03, -1.0094e-02, -2.6983e-02,
        -1.7584e-02, -1.2332e-02, -4.1000e-02,  4.8676e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0399,  0.3736,  0.0123,  0.0091,  0.0739,  0.1326, -0.0070,  0.0252,
        -0.0307,  0.0698,  0.0243,  0.0062,  0.0482,  0.1128,  0.0009,  0.0062,
        -0.0062, -0.0212,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0476,  0.0461,  0.0370,  0.0578,  0.0135, -0.0101,  0.0157,  0.0386,
         0.0862,  0.2116, -0.0221,  0.0319,  0.0281,  0.0335,  0.0091,  0.0313,
         0.0227,  0.0023,  0.0022,  0.1560, -0.0101,  0.0049,  0.0446, -0.0031,
        -0.0047, -0.0108,  0.0184,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1347,  0.0151,  0.0230, -0.0209,  0.0120, -0.0052,  0.0240,  0.0056,
        -0.1030, -0.0185,  0.0103, -0.0584, -0.1799, -0.0383, -0.0087,  0.0028,
        -0.0238, -0.0120, -0.0085, -0.1860, -0.0294, -0.0080, -0.0273, -0.0138,
        -0.0131,  0.0177,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0164,  0.2099, -0.0002,  0.0627,  0.0780,  0.0491,  0.0312,  0.0019,
         0.0772,  0.0694,  0.0754,  0.0462,  0.0363,  0.0059,  0.0108,  0.0775,
         0.0278,  0.0238,  0.0080,  0.0088, -0.0834,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0338, -0.1508, -0.2642, -0.1198, -0.0740, -0.0516, -0.1156, -0.0182,
         0.0110, -0.0056, -0.0696, -0.0539, -0.0279,  0.0034, -0.0006,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1736,  0.0179,  0.0332,  0.0339, -0.0388,  0.0647,  0.0474,  0.0219,
         0.0548,  0.0522,  0.0058, -0.0045,  0.2119,  0.1390,  0.0129,  0.0420,
        -0.0301, -0.0154,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0913, -0.2023, -0.0685, -0.0906, -0.0178,  0.0637,  0.0274, -0.0100,
         0.0015, -0.0473, -0.0378, -0.0166,  0.0225, -0.1148, -0.0097,  0.0119,
        -0.0098,  0.0155,  0.1409,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0006, -0.1146, -0.0579, -0.0380, -0.0401, -0.0112,  0.0204,  0.0017,
         0.0059, -0.0038,  0.0039,  0.0212, -0.0181,  0.0091,  0.0048,  0.0069,
        -0.0583,  0.0067,  0.0056,  0.0325, -0.0861, -0.0014, -0.0476, -0.0022,
        -0.0606, -0.0106, -0.0364,  0.0165, -0.0215, -0.0090,  0.0079, -0.0137,
        -0.0731,  0.0010, -0.0093, -0.0024,  0.0372,  0.0202, -0.0078,  0.0242,
         0.0117,  0.0080, -0.0059,  0.0243], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.1573e-01, -1.9761e-01, -1.9131e-02, -4.4591e-02, -3.9460e-02,
        -1.2000e-03, -4.3241e-02,  1.1001e-02, -5.1890e-02,  1.2520e-02,
        -1.8404e-02,  5.2648e-03,  1.1688e-02, -4.4317e-05,  7.0952e-03,
         5.9652e-03,  3.9951e-03, -5.9284e-02, -4.7076e-03, -1.4438e-02,
        -2.6219e-02,  8.8801e-03,  1.5663e-02, -3.3612e-03,  2.8114e-02,
        -1.0740e-03, -1.7812e-03, -7.4142e-02,  1.2318e-02, -1.8470e-02,
        -2.1353e-02, -2.3240e-02, -7.2129e-03, -9.0772e-02,  1.4109e-04,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0198,  0.3749,  0.0547, -0.0152,  0.0390,  0.0257,  0.0760,  0.0480,
        -0.0056, -0.0040,  0.0074,  0.0305,  0.0343,  0.0059,  0.0039,  0.0207,
         0.0054,  0.0076, -0.0128, -0.0022, -0.0140,  0.0056, -0.1162,  0.0706,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0843,  0.2572,  0.0139,  0.0205,  0.0360,  0.0023,  0.0258,  0.0613,
         0.0890,  0.0124,  0.0112, -0.0099,  0.0034,  0.0137,  0.0487, -0.0184,
         0.0009,  0.0203,  0.0650, -0.0186,  0.0197,  0.0297,  0.0015,  0.0222,
        -0.0153,  0.0286,  0.0078,  0.0178,  0.0042,  0.0120,  0.0077, -0.0083,
         0.0018, -0.0106,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0536,  0.0210,  0.0235, -0.0565,  0.0252,  0.0002, -0.0884, -0.0010,
        -0.0022, -0.0048, -0.0077, -0.0066, -0.0995, -0.0058, -0.0808, -0.1091,
        -0.0165,  0.0179, -0.0089, -0.0433,  0.0095, -0.0239,  0.0018, -0.0754,
        -0.0427, -0.0238, -0.0800,  0.0143, -0.0387,  0.0174,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0166, -0.0804, -0.0582, -0.2815, -0.0597, -0.0097, -0.0409, -0.0415,
        -0.0399, -0.1516,  0.0598, -0.0740, -0.0863,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0207,  0.0157,  0.1306,  0.0467, -0.0240, -0.0409, -0.0158, -0.1240,
        -0.1680, -0.0110,  0.0325, -0.1408, -0.0020, -0.0215, -0.1119,  0.0036,
         0.0244,  0.0659,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0502, -0.2345, -0.3322, -0.0208, -0.1360,  0.0315,  0.0576, -0.1373,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.8085e-02, -1.0769e-01, -4.8160e-03,  7.0193e-04,  6.8255e-03,
        -1.1103e-02,  3.6058e-03, -2.8585e-03,  1.3917e-02, -8.7118e-04,
        -2.5927e-03,  3.4294e-03, -3.3588e-03,  4.0327e-03,  2.9048e-03,
        -5.0970e-03,  6.6346e-03,  3.0802e-03, -2.7466e-05, -3.1899e-03,
        -3.4962e-03,  3.9018e-03, -1.3956e-02, -1.4685e-02,  1.0725e-02,
        -1.4683e-02, -9.3587e-03, -1.6757e-02, -5.3804e-02, -2.7166e-02,
        -5.5097e-03, -9.6656e-02, -6.1852e-03, -1.6201e-01, -2.2957e-02,
        -7.0644e-02, -5.4999e-02, -1.3438e-02,  1.3810e-02,  6.3153e-03,
        -8.8081e-03,  2.0334e-02, -1.4417e-02,  1.9283e-02,  2.1755e-02,
         8.0051e-03, -1.1523e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1141,  0.2793,  0.2185,  0.1284,  0.0056,  0.0347, -0.0682,  0.0058,
         0.0557,  0.0550, -0.0347,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0477, -0.2683, -0.0382, -0.0825,  0.0022, -0.0061, -0.0279, -0.1496,
        -0.0077, -0.0780, -0.0078, -0.0316, -0.0977, -0.0038, -0.0036, -0.0406,
         0.0069, -0.0996,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0778e-02,  2.0108e-01,  3.2344e-01, -1.0789e-02, -3.3912e-02,
         3.2367e-02,  9.5425e-02, -2.2469e-02,  5.6598e-02,  2.6496e-02,
         1.9042e-02,  1.2911e-04,  2.2997e-03,  1.2001e-01,  4.5161e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0345, -0.3118, -0.0637, -0.0176, -0.0028, -0.0879, -0.1080, -0.0247,
        -0.0278, -0.0959, -0.0885,  0.0625, -0.0072,  0.0555, -0.0114,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1809,  0.0115, -0.0028,  0.0154, -0.0314,  0.0042, -0.0190, -0.0085,
        -0.0723, -0.0084, -0.0089, -0.0357, -0.1024,  0.0024, -0.0172,  0.0006,
        -0.0461,  0.0633,  0.0005,  0.0346, -0.1138, -0.1392,  0.0060,  0.0056,
        -0.0371, -0.0319,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1348, -0.1276, -0.0579, -0.0905, -0.0538,  0.0335,  0.0148, -0.1198,
        -0.2078, -0.0508,  0.0252,  0.0003,  0.0230, -0.0361,  0.0242,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0464, -0.2761, -0.0208, -0.0209, -0.0656, -0.0574, -0.0653,  0.0056,
        -0.0042, -0.1328,  0.0165, -0.0012, -0.0169, -0.0160,  0.0051, -0.0089,
        -0.0079, -0.0204, -0.0168,  0.0118, -0.0155,  0.0023, -0.0142,  0.0067,
         0.0446,  0.0999,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.7558e-02, -2.9553e-02, -6.6523e-03,  9.6914e-03, -4.8197e-02,
        -5.5709e-02,  3.1234e-03,  7.6140e-03, -3.6088e-02, -7.7987e-02,
        -2.4748e-01,  8.9436e-03, -4.3321e-03,  1.7343e-02, -1.3433e-01,
        -5.3116e-02,  2.9431e-02, -3.7557e-02,  1.6695e-04, -4.9338e-02,
        -1.8170e-02, -3.4544e-04, -1.6160e-02, -2.4881e-02,  8.6564e-03,
        -2.7580e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0538, -0.1212,  0.0208, -0.0194, -0.0123,  0.0262, -0.0500, -0.0502,
        -0.0131, -0.0284, -0.0239, -0.0228, -0.0567,  0.0175, -0.0251, -0.0220,
        -0.0117,  0.0018, -0.0136,  0.0149,  0.0150, -0.0279, -0.0289, -0.0671,
        -0.1219, -0.0143, -0.0203, -0.0137,  0.0105, -0.0066,  0.0060, -0.0250,
         0.0022,  0.0110,  0.0061, -0.0176,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0368, -0.0893, -0.0410, -0.0411, -0.0334, -0.0095, -0.0459, -0.0755,
         0.0068,  0.0212, -0.0256, -0.0092, -0.0089,  0.0257, -0.0092, -0.0116,
        -0.0048, -0.0061,  0.0096,  0.0221, -0.0134, -0.0446, -0.0199, -0.0206,
        -0.0373, -0.0679,  0.0097, -0.0212, -0.0677, -0.0086, -0.0166, -0.0429,
        -0.0441,  0.0093, -0.0107, -0.0323,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1544,  0.0916,  0.0117,  0.0367,  0.0011,  0.2142,  0.0278,  0.0729,
         0.0149, -0.0032,  0.0527,  0.0391,  0.0784, -0.0349, -0.0261, -0.0032,
         0.0106, -0.0160,  0.0375, -0.0731,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.4649e-01, -5.8631e-02, -1.3671e-02,  1.8090e-02,  1.7745e-03,
        -3.7565e-03, -8.4928e-03, -2.0298e-02,  5.8573e-03, -1.1318e-02,
         1.9855e-03,  1.5453e-03, -4.1906e-03, -4.1468e-03, -8.4580e-02,
        -1.8863e-02, -1.5369e-02, -2.2572e-02, -4.1457e-03, -1.5507e-03,
         3.5221e-03, -2.5166e-03,  7.7382e-03,  4.4106e-03, -4.3387e-02,
        -4.3532e-03, -4.1577e-03, -1.4020e-02, -1.1927e-02, -4.8181e-02,
        -4.5323e-02,  1.0370e-04, -7.4244e-03, -7.2439e-02,  9.9332e-03,
         1.9527e-02, -2.5618e-02, -1.2809e-01,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0420, -0.1607, -0.0536, -0.0695, -0.0115,  0.0002, -0.0208,  0.0005,
        -0.0357, -0.0324, -0.0646, -0.0667, -0.0065,  0.0069,  0.0161, -0.0572,
        -0.0029, -0.0426, -0.0078,  0.0069, -0.0048,  0.0028,  0.0327,  0.0084,
        -0.0117,  0.0072, -0.0125, -0.0103, -0.0666,  0.0211, -0.0148, -0.0140,
         0.0106, -0.0026, -0.0177,  0.0133,  0.0055, -0.0073, -0.0079,  0.0105,
        -0.0129], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.8323e-01, -2.3878e-01, -2.9908e-02, -6.6748e-02,  1.4344e-02,
        -3.8826e-02, -5.1553e-03, -3.4722e-02, -1.5992e-02, -4.6700e-03,
         3.2943e-03, -4.5656e-02, -2.8529e-02, -1.8576e-02,  1.4399e-02,
        -1.0759e-01, -1.0478e-02, -4.8799e-02,  2.3433e-04, -1.7410e-02,
         3.2118e-02,  4.0541e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4276e-01,  5.3391e-03,  4.2836e-03,  2.3070e-03,  9.4631e-03,
        -3.8054e-02, -2.5854e-02, -7.0771e-03, -1.7390e-02,  2.0706e-03,
         1.2025e-05, -3.9593e-02, -1.0280e-02, -1.1292e-02, -2.1512e-02,
        -6.1842e-02, -1.0832e-02, -4.5716e-02,  7.3320e-03, -7.5936e-02,
        -8.9865e-02, -2.6745e-03,  7.6382e-03, -7.4094e-02, -1.0657e-02,
         2.6028e-02, -2.2253e-02, -8.7404e-03, -4.8472e-02, -1.7919e-02,
        -1.6963e-02, -2.6625e-02, -1.4577e-02,  5.2843e-03, -8.9268e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0612, -0.0397, -0.1633,  0.0060, -0.0915, -0.0306, -0.0317, -0.0015,
        -0.0457, -0.0064, -0.0383, -0.0071, -0.0154, -0.0099, -0.0045, -0.0539,
         0.0534,  0.0096, -0.0435, -0.0095, -0.0549, -0.0286,  0.0023, -0.0434,
        -0.0178, -0.0378, -0.0225,  0.0131,  0.0569,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1051, -0.3013, -0.0787, -0.0478, -0.0151, -0.0503, -0.0256, -0.0012,
        -0.0184, -0.0074, -0.0080, -0.0107, -0.0172,  0.0089,  0.0063, -0.0510,
        -0.0411, -0.0130, -0.0190,  0.0145, -0.0413, -0.0356, -0.0027, -0.0142,
         0.0658,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0302,  0.0315, -0.0282, -0.0310, -0.1725, -0.0656, -0.0628, -0.0918,
        -0.1224,  0.0059,  0.0316,  0.0593,  0.0103,  0.0276, -0.0511, -0.0698,
        -0.0386, -0.0149,  0.0089, -0.0105,  0.0081,  0.0165, -0.0109,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0145, -0.0199,  0.0034,  0.0055,  0.0441,  0.0131,  0.0012,  0.0025,
        -0.0062,  0.0090, -0.0019, -0.0085, -0.0635, -0.1240,  0.0164, -0.0271,
        -0.0211, -0.0493, -0.0211,  0.0016,  0.0176,  0.0356,  0.0001,  0.0087,
        -0.0068, -0.0261, -0.1002, -0.0271, -0.0404, -0.0385, -0.0790, -0.0268,
        -0.0459, -0.0221, -0.0166, -0.0239,  0.0035, -0.0157, -0.0112,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0785,  0.0494, -0.0586, -0.1692, -0.2205, -0.0291, -0.0598, -0.1182,
        -0.1666,  0.0149,  0.0339, -0.0012,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.0187, -0.0883, -0.1123, -0.0024, -0.0124,  0.0066, -0.0140,  0.0043,
        -0.0043, -0.0204,  0.0026,  0.0088,  0.0325,  0.0163, -0.0249,  0.0030,
        -0.0387, -0.0808,  0.0045, -0.0223, -0.0173, -0.0246,  0.0117, -0.0117,
        -0.0134,  0.0033, -0.0110, -0.0347, -0.0260, -0.0224,  0.0337,  0.0325,
        -0.0534,  0.0068,  0.0029, -0.0249, -0.0558,  0.0050,  0.0007, -0.0065,
         0.0095,  0.0082,  0.0138, -0.0402,  0.0119,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0400, -0.4075, -0.0615, -0.1204, -0.0264, -0.0184,  0.0275, -0.0703,
        -0.1244, -0.0117,  0.0653, -0.0266,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1570, -0.3534, -0.2360, -0.0128,  0.1011,  0.0503, -0.0628,  0.0265,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1137, -0.1570, -0.0246, -0.0138, -0.0161,  0.0083, -0.0128, -0.0131,
        -0.0114, -0.0729, -0.0142,  0.0082, -0.0058, -0.0013, -0.0046,  0.0019,
        -0.0061,  0.0103,  0.0078, -0.0041,  0.0117,  0.0053, -0.0056, -0.0078,
        -0.0032,  0.0007,  0.0051, -0.0167, -0.0319,  0.0164, -0.0034, -0.0086,
         0.0005, -0.0036,  0.0019, -0.0029,  0.0024,  0.0107,  0.0050,  0.0017,
         0.0019,  0.0020, -0.0337,  0.0034,  0.0107,  0.0114, -0.0469, -0.0311,
        -0.0307, -0.0609, -0.0062, -0.0106, -0.0122, -0.0030, -0.0081,  0.0055,
        -0.0081,  0.0047,  0.0017,  0.0323,  0.0218], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0549, -0.0179,  0.0073, -0.0252, -0.1075, -0.0116, -0.0440,  0.0056,
         0.0054,  0.0004,  0.0084, -0.0257, -0.0044, -0.0280, -0.0808, -0.0086,
        -0.0181, -0.0097, -0.0351, -0.0094, -0.0990, -0.0994, -0.0229,  0.0190,
        -0.0271, -0.0081,  0.0092, -0.0034, -0.0104, -0.0383, -0.0324,  0.0392,
        -0.0834,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0115, -0.0805, -0.1675, -0.0427, -0.0137, -0.0136, -0.0533, -0.0044,
         0.0269,  0.0027,  0.0087, -0.0026, -0.0127,  0.0207,  0.0082,  0.0884,
        -0.0083, -0.0266, -0.2084,  0.0270,  0.0416, -0.0540,  0.0506,  0.0253,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0258, -0.0961, -0.0731, -0.0348, -0.0437, -0.0498,  0.0121, -0.0006,
         0.0189, -0.1330, -0.0793, -0.0708, -0.0796, -0.0193,  0.0019, -0.0063,
         0.0447, -0.0248,  0.0268, -0.0086,  0.0003,  0.0223, -0.0661, -0.0614,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0528, -0.0065, -0.0219, -0.0535,  0.0021, -0.0074, -0.0265, -0.0808,
        -0.0119, -0.0162,  0.0213, -0.0185, -0.0087, -0.0071, -0.0594, -0.1180,
        -0.0708,  0.0113, -0.0178, -0.0168, -0.0495,  0.0058, -0.0148, -0.0533,
        -0.0252, -0.0122, -0.0453, -0.0093, -0.0023,  0.0007, -0.0550, -0.0128,
         0.0357, -0.0489,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1226, -0.0095, -0.0283, -0.0005, -0.1018, -0.0245, -0.0046, -0.0100,
        -0.0026,  0.0020,  0.0136,  0.0300,  0.0112, -0.2009, -0.0053, -0.0154,
        -0.0951,  0.0004, -0.0757, -0.0097, -0.0320, -0.1518, -0.0033, -0.0136,
         0.0355,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0925,  0.0415,  0.0080,  0.0127, -0.0162,  0.0111, -0.1272, -0.1196,
        -0.0526, -0.3682, -0.0300, -0.0247, -0.0209, -0.0293,  0.0268,  0.0187,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0484, -0.1779,  0.0154, -0.0720, -0.0601, -0.0648, -0.1056, -0.0285,
         0.0458, -0.0019, -0.0106, -0.0214, -0.0389, -0.0200, -0.0284, -0.0253,
        -0.0338, -0.0703, -0.0551, -0.0079, -0.0056, -0.0120,  0.0502,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1671,  0.0648,  0.1024,  0.0998,  0.2159,  0.0147,  0.0297,  0.0769,
         0.0229,  0.0256,  0.0043, -0.0106,  0.0852,  0.0803,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0754, -0.0571, -0.0481, -0.1588, -0.0216,  0.0056, -0.0294,  0.0312,
        -0.0213, -0.0185, -0.0591, -0.0383, -0.0311, -0.0026, -0.0183,  0.0250,
         0.0313, -0.1126, -0.0133, -0.1083, -0.0398,  0.0146,  0.0166,  0.0219,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0653,  0.0123, -0.1370, -0.0377, -0.1931, -0.0787, -0.3223, -0.0056,
        -0.0442,  0.0092, -0.0947,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0236,  0.0230, -0.0348, -0.0139, -0.0231, -0.0894, -0.1595,  0.0107,
        -0.1493, -0.0516, -0.0090,  0.0020, -0.0304,  0.0422, -0.0835, -0.0146,
        -0.0234, -0.0659, -0.0265, -0.0755, -0.0482,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0073,  0.0318, -0.0484,  0.0328, -0.0702, -0.0040, -0.0146, -0.3667,
        -0.0453, -0.1605, -0.0742, -0.0404, -0.0705, -0.0334,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0501,  0.0508,  0.0202,  0.0173,  0.0222,  0.0100, -0.0065, -0.0188,
        -0.0049,  0.0379, -0.0405, -0.0559,  0.3309, -0.0100,  0.0444,  0.0385,
         0.0291,  0.0428, -0.0208, -0.0505,  0.0978,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0403, -0.2023, -0.0216, -0.0991,  0.1030, -0.1315, -0.0385, -0.1726,
        -0.0652, -0.0481, -0.0222, -0.0389, -0.0167,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0361, -0.0357, -0.0100, -0.0039, -0.0060, -0.0114, -0.0192, -0.0550,
        -0.0368,  0.0016, -0.0055,  0.0187,  0.0173, -0.0029, -0.0042, -0.0038,
         0.0033,  0.0099, -0.0074,  0.0103,  0.0165,  0.0120, -0.0079, -0.0062,
        -0.1130, -0.0259, -0.2182, -0.0760, -0.0835, -0.0161, -0.0226, -0.0049,
         0.0126, -0.0081, -0.0111,  0.0103, -0.0162,  0.0019,  0.0050, -0.0042,
        -0.0095,  0.0023,  0.0104,  0.0010, -0.0057], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0948, -0.0603, -0.0329, -0.0128,  0.0012, -0.0100, -0.0213, -0.0136,
         0.0109,  0.0152, -0.1023, -0.0039, -0.0139,  0.0026,  0.0113, -0.0405,
        -0.0571,  0.0066,  0.0048,  0.0293, -0.0551, -0.0839,  0.0029, -0.0551,
         0.0206,  0.0268,  0.0099,  0.0064, -0.0650, -0.0094, -0.0353, -0.0120,
        -0.0483, -0.0242,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0898,  0.0325, -0.0523, -0.0359, -0.0206,  0.0066, -0.0075, -0.0047,
         0.0340, -0.0258, -0.0065, -0.0070, -0.0052, -0.0195, -0.0114,  0.0271,
        -0.0079,  0.0064, -0.0389, -0.1620, -0.2794,  0.0137,  0.0090,  0.0406,
        -0.0459, -0.0099,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0372, -0.3066, -0.0254, -0.0593,  0.0608, -0.0198,  0.0168, -0.0348,
        -0.0218,  0.0279, -0.0066,  0.0067,  0.0070, -0.0106,  0.0006, -0.0296,
        -0.1721,  0.0322, -0.0322,  0.0053, -0.0022,  0.0068,  0.0775,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0143, -0.1456, -0.0576, -0.0249, -0.0005,  0.0119, -0.0217, -0.0243,
        -0.0482, -0.0069,  0.0149, -0.0129, -0.0054, -0.0162, -0.1270,  0.0266,
         0.0023, -0.0286, -0.0419,  0.0100, -0.0046,  0.0095,  0.0071,  0.0005,
         0.0156, -0.0020, -0.0707, -0.0853,  0.0038, -0.0190, -0.0076, -0.0141,
         0.0221,  0.0020,  0.0043,  0.0080, -0.0238,  0.0086, -0.0342, -0.0157,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1070, -0.0630, -0.0251, -0.0227, -0.0748, -0.0212, -0.0481, -0.0556,
        -0.0106, -0.0144, -0.0051, -0.0155, -0.0352, -0.0113, -0.0421, -0.0447,
        -0.0425, -0.0022, -0.0076, -0.0440, -0.0051, -0.0633, -0.0070,  0.0146,
         0.0132,  0.0099, -0.0059, -0.0573, -0.0052, -0.0436, -0.0693,  0.0021,
         0.0025,  0.0086,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0180, -0.0297, -0.0124, -0.0163, -0.0646, -0.0006,  0.0043, -0.0087,
         0.0047,  0.0058, -0.0132, -0.0120,  0.0227, -0.0184, -0.0724,  0.0081,
        -0.0092, -0.0168, -0.0570, -0.0063, -0.0093, -0.0144, -0.0120,  0.0146,
        -0.0102, -0.1110, -0.0344, -0.0283, -0.0031,  0.0180, -0.0156, -0.0044,
         0.0060, -0.0109, -0.0436, -0.0012, -0.0246, -0.0609, -0.0086, -0.0347,
        -0.0170, -0.0183, -0.0187, -0.0107, -0.0008, -0.0015,  0.0024,  0.0032,
        -0.0095, -0.0107,  0.0121, -0.0283], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.0293e-02, -1.1725e-02, -6.5096e-04, -2.1968e-02, -9.6673e-02,
        -1.1090e-02,  1.5460e-04,  4.7853e-02,  2.6910e-03, -2.5563e-02,
         1.2449e-03,  4.2268e-03,  1.1097e-02, -2.1989e-01, -4.5830e-02,
        -4.3210e-02, -1.1491e-01, -5.3395e-02, -5.1450e-02, -1.8325e-04,
        -3.4425e-02, -1.3622e-02, -2.0868e-02, -4.9154e-02,  1.5994e-02,
        -6.9945e-03,  2.2504e-02,  2.4132e-02,  3.8324e-03,  2.3500e-02,
         1.0876e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0016, -0.1307,  0.0492, -0.0007, -0.0076, -0.0037,  0.0113, -0.0056,
        -0.0365,  0.0021, -0.0255, -0.1102,  0.0418,  0.0252, -0.0452, -0.0459,
         0.0407, -0.0789, -0.1169, -0.1053, -0.0307, -0.0262, -0.0222,  0.0163,
         0.0204,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0742,  0.0952,  0.0114,  0.0121, -0.0431, -0.2038, -0.2989, -0.0624,
         0.0397, -0.0720, -0.0295,  0.0576,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.4791e-02, -3.6505e-01,  5.4459e-02, -4.6479e-02, -9.3583e-02,
        -2.9620e-02, -8.5121e-04,  8.1454e-03, -4.5193e-03, -1.5750e-02,
         7.2377e-03,  9.6292e-03, -5.6284e-02, -7.9911e-02,  1.6648e-03,
        -2.7974e-02, -9.8660e-03, -1.0362e-02,  1.0243e-02,  2.4462e-03,
        -2.9990e-02, -7.1877e-03, -5.9398e-04, -2.1597e-04,  2.1464e-02,
        -4.2600e-02, -7.7250e-03,  2.4135e-03, -1.8946e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0181, -0.3374, -0.0053, -0.0399,  0.0204, -0.1268, -0.2205,  0.0458,
         0.0325,  0.0467, -0.0513, -0.0553,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0457, -0.2443, -0.0994, -0.1147, -0.0334, -0.0550, -0.1405, -0.0678,
        -0.0946, -0.0272,  0.0089, -0.0686,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1058,  0.0889, -0.0813, -0.0031, -0.0874,  0.0600, -0.0091, -0.0198,
        -0.1098, -0.0455, -0.0169, -0.1186, -0.0341, -0.0035, -0.0134, -0.0643,
        -0.0265,  0.0364, -0.0350, -0.0404,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0356, -0.2669, -0.0089, -0.0348, -0.1562, -0.0205, -0.0085,  0.0081,
        -0.0156, -0.0013, -0.0106, -0.0166, -0.0082, -0.0415, -0.0070,  0.0030,
        -0.0045,  0.0116, -0.0006, -0.0123, -0.0303,  0.0145,  0.0178, -0.0025,
        -0.0119,  0.0181, -0.0083, -0.0419, -0.0204,  0.0063,  0.0265, -0.0065,
        -0.0408, -0.0539, -0.0161,  0.0078,  0.0041,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0341,  0.5345,  0.0039, -0.2676,  0.0412,  0.0189,  0.0733,  0.0059,
        -0.0206,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0076,  0.0323, -0.0157, -0.1265, -0.0405, -0.0753, -0.0699, -0.0849,
         0.0001,  0.0075, -0.0227, -0.0351, -0.0124, -0.1036, -0.0322, -0.0018,
         0.0133, -0.0102, -0.0864, -0.0598,  0.0008,  0.0102, -0.0014,  0.0091,
        -0.0212,  0.0933,  0.0259,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1370, -0.2269,  0.0100, -0.0317,  0.0028,  0.0020,  0.0086, -0.0019,
        -0.0085,  0.0023,  0.0181, -0.0613, -0.0186,  0.0003, -0.0020,  0.0031,
        -0.0378, -0.0074,  0.0038, -0.0220,  0.0084, -0.0448, -0.0334, -0.0040,
        -0.0113,  0.0056, -0.0061, -0.0028, -0.1346, -0.0116,  0.0049, -0.0351,
         0.0557,  0.0356,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0293,  0.0153, -0.0102, -0.0219, -0.0146, -0.0025,  0.0124,  0.0102,
         0.0232,  0.0223, -0.0184, -0.0988, -0.0309,  0.0005,  0.0141, -0.0461,
        -0.0638, -0.0537, -0.0231, -0.0017, -0.0035, -0.0031, -0.0537, -0.0127,
        -0.0024, -0.0122, -0.0011, -0.0010, -0.0167, -0.0160,  0.0022, -0.0433,
        -0.1035, -0.0319, -0.0242, -0.0197, -0.0461, -0.0229,  0.0066, -0.0029,
        -0.0050, -0.0317, -0.0053,  0.0182, -0.0014,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0421, -0.0272, -0.0114, -0.0053, -0.0817, -0.1222, -0.2023, -0.0453,
        -0.0722, -0.1475, -0.0236, -0.0042, -0.0409,  0.0431, -0.0324, -0.0985,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0281, -0.0123,  0.0435,  0.0491,  0.0587,  0.0230, -0.0315,  0.0048,
         0.1900,  0.1461,  0.0515,  0.0902, -0.0139,  0.0368,  0.0240, -0.0242,
         0.0115,  0.0052, -0.0020,  0.0237, -0.0053, -0.0089,  0.0016, -0.0302,
        -0.0243,  0.0399, -0.0198,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.9399e-02, -4.2256e-02, -4.8285e-02,  3.9541e-03, -8.3883e-03,
         5.3434e-03, -1.3743e-02,  2.3088e-03, -1.6972e-02,  3.0914e-03,
         1.0109e-03,  5.2909e-03, -2.1437e-02, -4.2736e-03, -1.3373e-02,
        -5.1415e-03, -1.5563e-02,  2.8477e-03, -2.7775e-05, -1.7234e-02,
         2.6288e-02, -2.5039e-02,  7.8184e-04, -1.0408e-01, -1.6167e-04,
        -1.7603e-02, -5.2754e-02, -7.2051e-02, -6.7404e-03, -5.0428e-02,
         1.2207e-04, -6.0478e-02, -2.3600e-02, -1.8481e-02, -2.7616e-02,
        -5.8687e-02, -1.0287e-02, -1.3445e-02,  3.0731e-03,  5.3678e-03,
        -1.3780e-02, -2.4731e-02, -3.4152e-03,  1.4528e-02, -2.9767e-02,
         2.6758e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0753,  0.0164, -0.0036,  0.0148, -0.0065, -0.0130, -0.0064, -0.0764,
        -0.0111, -0.0465, -0.0100,  0.0130,  0.0190, -0.0151, -0.0373,  0.0027,
        -0.0114, -0.0429, -0.0975, -0.2366, -0.0524, -0.0678, -0.0080,  0.0011,
         0.0126,  0.0139, -0.0505,  0.0165, -0.0194, -0.0025,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0071, -0.0600, -0.0944, -0.0763, -0.0347, -0.0264,  0.1638,  0.3174,
         0.0336,  0.0206,  0.1657,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1300, -0.2564, -0.0508, -0.0386, -0.0596, -0.0183, -0.0593, -0.0635,
         0.0074,  0.0188,  0.0231, -0.0100, -0.0323, -0.0033, -0.0231, -0.0602,
        -0.0202, -0.0462, -0.0214, -0.0573,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0209, -0.2502, -0.1799, -0.0863,  0.0161, -0.1233, -0.0938,  0.0042,
        -0.0565, -0.1321, -0.0366,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0049, -0.2290, -0.1098, -0.1154,  0.0029, -0.0881, -0.0472, -0.0385,
        -0.0491, -0.0136, -0.0138, -0.0152, -0.0332, -0.0210,  0.0098, -0.0397,
        -0.0162, -0.0030,  0.0013, -0.0100,  0.1380,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0363,  0.3383,  0.1548, -0.0178,  0.0351,  0.2104, -0.0339,  0.0037,
        -0.0005,  0.0258, -0.0135,  0.0045,  0.0136,  0.0132,  0.0030, -0.0081,
        -0.0650, -0.0224,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0272, -0.0698, -0.1238, -0.0147, -0.0096,  0.0123,  0.0231,  0.0112,
         0.0009, -0.0063, -0.0017,  0.0054, -0.0087, -0.0468, -0.0451,  0.0089,
        -0.0297, -0.0861, -0.0169, -0.0172, -0.0219, -0.0489, -0.1020, -0.0067,
        -0.0135,  0.0006, -0.0233, -0.0137, -0.0149, -0.0958, -0.0067, -0.0167,
        -0.0118, -0.0016, -0.0245,  0.0004, -0.0009,  0.0287,  0.0023,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.3461e-02,  3.8651e-01,  4.3148e-02,  4.2838e-02,  5.9130e-03,
        -4.0113e-03,  1.6320e-02,  1.3531e-04,  2.5045e-02,  1.1052e-02,
        -3.9172e-04,  5.5599e-02,  1.8876e-02,  4.3630e-03,  1.0111e-02,
         8.5948e-04,  3.5425e-02,  9.7951e-03,  8.2304e-03,  1.0624e-02,
         8.4381e-03,  6.1124e-03,  6.3568e-04, -8.1305e-03, -7.7705e-03,
         1.4810e-02,  1.8820e-02, -5.8052e-03,  4.3965e-02,  6.4867e-02,
         6.7669e-03,  2.0159e-02,  1.5707e-02,  5.5300e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0835, -0.2934, -0.1776, -0.1802, -0.0540, -0.0574, -0.0640, -0.0448,
         0.0451,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0383, -0.1234, -0.0876, -0.0196, -0.0057, -0.0497,  0.0020, -0.0056,
         0.0028, -0.0059, -0.0215,  0.0021, -0.0062, -0.0169,  0.0092, -0.0500,
        -0.0306, -0.1907, -0.0032, -0.0118, -0.0169,  0.0103,  0.0456, -0.0010,
        -0.0476,  0.0002, -0.0119, -0.0037,  0.0071, -0.0055, -0.0231, -0.0349,
         0.0086, -0.0163,  0.0070, -0.0005,  0.0059, -0.0022,  0.0370,  0.0320,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.0300e-02, -4.5237e-02, -1.2393e-01,  2.9729e-02, -3.7252e-02,
        -9.0460e-03,  7.5208e-03,  7.0084e-03, -2.2165e-02, -5.9286e-02,
         1.0019e-04,  5.0929e-03,  5.9944e-03,  5.2615e-03, -1.6181e-02,
        -2.8976e-03, -1.5749e-02,  4.8020e-02, -2.8408e-02, -3.7842e-03,
        -6.3740e-03, -4.5231e-04, -2.3513e-02, -8.0120e-04, -5.4945e-03,
        -2.5192e-02, -4.1557e-02, -4.1340e-03, -7.1209e-02, -7.7827e-02,
        -4.3208e-03, -1.5520e-02,  9.7582e-03,  1.0511e-02,  3.3791e-03,
         7.8536e-03, -3.9847e-02, -6.0988e-02,  1.2742e-03,  8.0793e-03,
         1.8251e-02,  9.3628e-03, -7.2850e-04,  7.5082e-04, -3.8730e-03,
         7.6857e-03, -6.3283e-03,  2.3040e-03,  9.2069e-03, -1.3294e-03,
         4.6343e-03,  2.4492e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0355, -0.1592, -0.0567, -0.0359, -0.0429, -0.0401, -0.0455, -0.0433,
        -0.0429,  0.0083, -0.0069, -0.0665, -0.0707, -0.0260, -0.0129, -0.0644,
        -0.0467, -0.0197, -0.0747, -0.0070, -0.0392, -0.0552,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0194, -0.3074,  0.0435, -0.0466, -0.0073,  0.0045, -0.0059, -0.0720,
         0.0072,  0.0004, -0.0444, -0.0514,  0.0007, -0.0596,  0.0022, -0.1133,
        -0.0869,  0.0254,  0.0136,  0.0279,  0.0606,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.1806e-02, -8.2309e-02, -2.3427e-02, -2.3754e-02, -1.5427e-02,
         8.0315e-05, -1.1506e-02,  3.6398e-02, -1.8527e-02, -2.7297e-02,
        -1.8044e-03,  2.1789e-05,  2.3126e-03,  5.0365e-03,  2.2173e-03,
        -4.5339e-04,  2.5163e-03,  2.1174e-03, -1.1743e-02,  1.3124e-02,
         9.4078e-03,  2.4755e-02, -7.0705e-04,  2.0753e-03,  5.9867e-03,
         8.8707e-03, -2.9765e-02, -4.1618e-02, -6.1556e-03, -9.0238e-03,
        -1.0718e-01, -6.7677e-02, -1.4447e-02, -2.1444e-02, -1.2678e-02,
         2.4198e-03,  6.2974e-03,  8.8723e-03, -8.9071e-03, -2.9504e-03,
        -6.3498e-02,  1.9479e-02,  1.2901e-02, -5.5552e-03, -3.5666e-02,
         5.0630e-03, -1.9357e-02, -2.6616e-03,  2.8028e-02,  1.0373e-02,
        -1.1354e-02,  4.1563e-04, -8.8116e-03, -2.6257e-03, -6.1097e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0411, -0.0061,  0.0331, -0.0340, -0.0105, -0.0128, -0.1040, -0.0510,
         0.0034, -0.0041,  0.0245, -0.0003, -0.0025, -0.0196, -0.0418, -0.0128,
         0.0019, -0.0217, -0.0263, -0.0386, -0.0138, -0.0070, -0.0185,  0.0123,
         0.0224,  0.0117,  0.0113, -0.0560, -0.0024, -0.0187, -0.0255, -0.0158,
        -0.0122, -0.1033, -0.0101, -0.0257, -0.0042, -0.0282, -0.0254, -0.0184,
        -0.0089, -0.0016,  0.0012,  0.0102, -0.0453,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0067, -0.2216, -0.0589, -0.0760,  0.0510, -0.0129, -0.0294, -0.0159,
        -0.0526, -0.0333,  0.0370, -0.0216,  0.0020, -0.0357, -0.0769,  0.0467,
        -0.0094, -0.0435,  0.0200, -0.0107, -0.0627,  0.0388, -0.0266,  0.0101,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0473,  0.0228,  0.0010, -0.0089, -0.0272, -0.0108, -0.0395, -0.0357,
         0.0033,  0.0129,  0.0069, -0.0111,  0.0031, -0.0149,  0.0034,  0.0321,
        -0.0337, -0.0435, -0.0096, -0.0500,  0.0047, -0.0046, -0.0272, -0.0964,
        -0.0121, -0.0537, -0.0599, -0.0280, -0.0486, -0.0958, -0.0273,  0.0295,
        -0.0945,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0050, -0.3911, -0.0919, -0.0610,  0.0142, -0.0175,  0.0194, -0.0224,
        -0.0958,  0.0109,  0.0214,  0.0065,  0.0232, -0.0730,  0.0378, -0.0216,
        -0.0251, -0.0431, -0.0191,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1590, -0.0201,  0.0494, -0.0266, -0.0690, -0.0024, -0.0491, -0.0723,
        -0.0807,  0.0013,  0.0140, -0.0290,  0.0084, -0.0299,  0.0122, -0.0182,
        -0.0178, -0.0480, -0.0145,  0.0048, -0.0586, -0.1037, -0.0370, -0.0520,
         0.0218,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0728, -0.1888, -0.0973, -0.0432, -0.0356, -0.0599, -0.2841,  0.0287,
         0.0164, -0.0227,  0.0005, -0.0147, -0.0556, -0.0136, -0.0097,  0.0521,
         0.0043,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.0287,  0.0029, -0.0327, -0.0066, -0.0133, -0.0134, -0.0045, -0.0117,
        -0.0047, -0.1125, -0.0135, -0.0197,  0.0143,  0.0019,  0.0057, -0.0213,
        -0.0592, -0.1261,  0.0022, -0.0534,  0.0047, -0.1547, -0.1429, -0.0021,
        -0.0259, -0.0086,  0.0107, -0.0082, -0.0207,  0.0251, -0.0063, -0.0418,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0244, -0.1260, -0.1348, -0.0053, -0.0194, -0.0082, -0.0569, -0.1160,
        -0.0103, -0.0038,  0.0090, -0.0043,  0.0692,  0.0005, -0.0009,  0.0205,
        -0.0128,  0.0075, -0.0392,  0.0163, -0.0147,  0.0015, -0.0729, -0.1093,
         0.0116, -0.0225, -0.0679, -0.0144,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0729, -0.0861,  0.0099, -0.0216,  0.0382, -0.0648, -0.1690, -0.0455,
        -0.0211, -0.0490,  0.0055, -0.0612,  0.0031,  0.0050, -0.0183, -0.0580,
        -0.0526, -0.0528,  0.0052,  0.0031, -0.0083, -0.0121, -0.0050, -0.0059,
        -0.0390,  0.0491, -0.0146,  0.0147,  0.0086,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0283, -0.0875, -0.0314, -0.0140, -0.0734, -0.0380, -0.0317, -0.0127,
        -0.0120, -0.0031,  0.0311, -0.0161, -0.0395, -0.1449,  0.0209, -0.0528,
         0.0137, -0.0628, -0.0299, -0.0483, -0.0076,  0.0016, -0.0137, -0.0072,
        -0.0388, -0.0158, -0.0254, -0.0111, -0.0048,  0.0041, -0.0090,  0.0438,
        -0.0251,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0823, -0.2818, -0.0228,  0.0038, -0.0114, -0.0387,  0.0020, -0.0252,
         0.0148, -0.0899, -0.0193, -0.0379,  0.0138, -0.0347, -0.0182,  0.0055,
        -0.0361, -0.0239, -0.1086, -0.0011, -0.0282,  0.0239,  0.0256, -0.0504,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0829, -0.1226, -0.1544, -0.0052, -0.0061,  0.0147, -0.0137, -0.0626,
        -0.0213, -0.0612, -0.0369, -0.0726, -0.0473, -0.0202,  0.0194, -0.0133,
         0.0045, -0.0390, -0.0154, -0.0870, -0.0393, -0.0300, -0.0303,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0205, -0.0449,  0.0011, -0.0133, -0.0036, -0.0687, -0.0231,  0.0077,
         0.0185, -0.0119,  0.0214, -0.0133, -0.0048,  0.0134, -0.0077,  0.0010,
        -0.0437, -0.0311, -0.0193, -0.0143, -0.0106, -0.0085, -0.0046, -0.0152,
        -0.0672, -0.0196, -0.0071, -0.0067,  0.0146, -0.1439, -0.1672,  0.0042,
        -0.0432,  0.0007, -0.0214, -0.0377, -0.0021, -0.0312,  0.0109,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.0406e-02,  6.3330e-02,  7.5257e-03,  2.2947e-02,  1.1311e-01,
         7.3929e-03, -5.6974e-03, -9.2332e-03, -5.1785e-03, -1.9013e-02,
         5.0437e-03, -4.5722e-03, -2.8724e-03, -5.6196e-03, -6.6376e-03,
         2.4699e-02,  2.8044e-02, -1.1819e-02,  3.7798e-05, -1.8764e-03,
         4.1618e-02,  6.8098e-03, -3.9250e-04,  1.4384e-02,  9.4203e-02,
         1.0537e-01,  3.5160e-02,  2.0320e-02, -1.1175e-02,  4.1683e-02,
        -3.6543e-03,  1.8188e-02, -2.8937e-02,  9.8375e-03,  1.3724e-02,
        -4.7969e-03,  3.8978e-03,  6.4835e-02,  3.4049e-03,  4.8987e-03,
         6.1370e-03,  1.8651e-02, -1.2869e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.2908e-02, -3.2247e-01, -1.4379e-02, -5.8317e-03,  7.3463e-03,
        -2.8490e-03, -2.8151e-02, -1.0605e-02, -7.3063e-02,  6.2189e-03,
         1.0725e-02, -4.8644e-04, -2.5700e-02, -4.9141e-03, -3.8031e-03,
        -1.5662e-02, -1.2129e-02, -3.8068e-02, -7.9754e-03, -3.8095e-03,
         2.5762e-03,  2.7123e-04, -4.1596e-03, -4.2643e-03, -5.7492e-03,
        -7.1657e-03,  2.2477e-02, -1.0158e-03, -6.7001e-03,  1.6463e-02,
         1.7898e-02, -1.8919e-02, -5.4088e-02, -9.6736e-04, -1.9436e-02,
         4.5039e-04, -5.3814e-04, -1.2827e-02, -3.7716e-02, -3.8431e-03,
        -1.3728e-02, -2.8534e-02, -4.3367e-03,  1.6712e-02, -5.8762e-03,
        -5.3518e-03, -6.0587e-03,  5.0900e-03,  2.4216e-03, -1.3305e-03,
         8.2207e-03, -3.6014e-03,  3.5018e-03, -1.5556e-03,  1.1153e-02,
         2.2556e-02, -5.3540e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0375, -0.4493,  0.0328, -0.0151, -0.0675,  0.0243, -0.0355, -0.0889,
         0.0306, -0.0627, -0.0533,  0.0521,  0.0306, -0.0199,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1813, -0.1554, -0.0452, -0.0027, -0.0077, -0.0331, -0.0241, -0.0058,
        -0.0291, -0.0140, -0.0398, -0.0506, -0.0048, -0.0079, -0.0368,  0.0099,
        -0.0045, -0.0273,  0.0393, -0.0420, -0.1142, -0.0222, -0.0241, -0.0178,
         0.0204,  0.0074,  0.0070, -0.0012, -0.0142, -0.0104,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1000,  0.2186, -0.0902,  0.3411,  0.0818,  0.0345, -0.0439,  0.0899,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0245, -0.1509, -0.0230, -0.0472, -0.0158, -0.1013,  0.0014,  0.0165,
        -0.0200, -0.0249, -0.0127,  0.0100, -0.0389, -0.1057, -0.0028,  0.0016,
         0.0032, -0.0121, -0.0109,  0.0096,  0.0077,  0.0025,  0.0031,  0.0192,
        -0.0664, -0.0157, -0.0389, -0.0091, -0.0489, -0.0990,  0.0241,  0.0197,
        -0.0128,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0060, -0.1203, -0.0601, -0.0208,  0.0057, -0.0191, -0.0206, -0.0241,
         0.0102, -0.0282, -0.0002, -0.0023, -0.0200,  0.0022,  0.0147,  0.0009,
         0.0027,  0.0119, -0.0107,  0.0051, -0.0032,  0.0107,  0.0023, -0.0032,
        -0.0042, -0.0139, -0.0095, -0.0409,  0.0032,  0.0015, -0.0109, -0.0041,
        -0.0165, -0.0771, -0.0020, -0.0128, -0.0306, -0.0179, -0.0017,  0.0048,
         0.0066, -0.0049, -0.0071, -0.0262, -0.0078, -0.0016,  0.0035, -0.0025,
         0.0073,  0.0079,  0.0008, -0.0197,  0.0010, -0.0045, -0.0060,  0.0072,
        -0.0057, -0.0015, -0.0085, -0.0344, -0.0162, -0.0066, -0.0965, -0.0077,
         0.0039, -0.0014, -0.0008,  0.0158,  0.0019, -0.0195, -0.0081],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.0742e-01, -5.1687e-03, -4.1065e-03,  6.2883e-03, -2.1555e-02,
         5.0943e-03, -6.5037e-02, -1.9283e-02, -2.8036e-03, -8.0600e-03,
         4.8417e-03, -3.9575e-03, -1.1924e-03,  2.3185e-02,  6.2233e-05,
         4.6784e-03,  5.5947e-03,  2.1282e-03, -1.2084e-02, -1.0085e-01,
         2.7827e-04, -1.6491e-02,  2.3488e-02, -4.4997e-02,  5.4826e-04,
        -2.7578e-02, -3.1954e-02, -1.1216e-02,  2.4464e-03, -1.4219e-02,
        -1.9519e-02, -1.8662e-02, -4.4222e-02, -7.6960e-02,  5.7423e-03,
        -3.8966e-02, -3.0413e-02,  1.7664e-03,  4.9639e-04, -8.5099e-03,
         3.1753e-02, -4.6385e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0129, -0.2600,  0.0265, -0.0038,  0.0039,  0.0011,  0.0242,  0.0054,
        -0.0215, -0.0328, -0.0127, -0.0261, -0.0757, -0.0054,  0.0316, -0.0042,
         0.0122, -0.0443, -0.0811, -0.0095, -0.0118, -0.0189,  0.0067, -0.0056,
        -0.0128, -0.0553, -0.0279, -0.0217, -0.1018, -0.0025, -0.0059, -0.0017,
         0.0013, -0.0114, -0.0085,  0.0097,  0.0013,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0930, -0.3163, -0.0511, -0.0819, -0.0528,  0.0738, -0.0055, -0.0433,
        -0.0234, -0.0490, -0.0269,  0.0121, -0.0744, -0.0314,  0.0650,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0346, -0.0265,  0.0060, -0.0405, -0.1104, -0.1540, -0.0313, -0.0372,
        -0.1738, -0.0139,  0.0188, -0.0108, -0.1245,  0.0247,  0.0385,  0.0028,
        -0.0097,  0.0664,  0.0366, -0.0390,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0120,  0.0298,  0.0106, -0.0071, -0.1739, -0.0119,  0.0068,  0.0107,
         0.0138, -0.0089, -0.0202, -0.0390, -0.0208, -0.0109, -0.0349, -0.0379,
        -0.0053, -0.0074,  0.0066, -0.0545,  0.0005, -0.0064, -0.0006, -0.0149,
         0.0007, -0.0411, -0.0695, -0.0043, -0.0185, -0.0336, -0.0831, -0.0308,
         0.0025, -0.0171, -0.0295, -0.0241,  0.0013, -0.0094, -0.0115, -0.0041,
         0.0355, -0.0081,  0.0297,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0289,  0.0040,  0.0360,  0.0280,  0.0255,  0.0265,  0.0269,  0.1295,
         0.0112,  0.0046, -0.0099,  0.0179,  0.0390,  0.0788,  0.0827, -0.0075,
        -0.0141, -0.0063, -0.0055,  0.0281,  0.0593,  0.0224,  0.0125,  0.0046,
        -0.0084,  0.0211,  0.0289, -0.0108,  0.0617,  0.0046,  0.0048,  0.0343,
         0.0044,  0.0332,  0.0440,  0.0034,  0.0058, -0.0249,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0367, -0.3112, -0.1208, -0.0694,  0.0227, -0.0953,  0.0105,  0.0239,
         0.0112, -0.0574, -0.0145, -0.0096, -0.0477, -0.0088,  0.0207, -0.0096,
         0.0126, -0.0519,  0.0289,  0.0054,  0.0109, -0.0204,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0646, -0.1439, -0.0480,  0.0134,  0.0198, -0.0402, -0.0014,  0.0113,
        -0.0145, -0.0817, -0.0049,  0.0213, -0.0046,  0.0051, -0.0141, -0.0153,
        -0.0741, -0.0082, -0.0121, -0.0657, -0.0304, -0.0304, -0.0272, -0.0116,
        -0.0242,  0.0013, -0.0074, -0.0037, -0.0134, -0.0438, -0.0190, -0.0048,
        -0.0126,  0.0056, -0.0081, -0.0678,  0.0146, -0.0020, -0.0018, -0.0058,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0217,  0.0227, -0.0009,  0.0049, -0.0051,  0.0015, -0.0337, -0.0159,
        -0.1087,  0.0033, -0.0182, -0.0184, -0.0257, -0.0786, -0.0563,  0.0019,
         0.0048,  0.0051, -0.0843,  0.0158,  0.0030,  0.0144,  0.0160, -0.0009,
        -0.0096, -0.0881, -0.0055, -0.0859, -0.1147, -0.0373, -0.0138, -0.0835,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0473, -0.3425, -0.0532, -0.0105, -0.0259, -0.0038, -0.0389, -0.0370,
        -0.0726, -0.0127,  0.0102, -0.0221, -0.0294, -0.0044, -0.0084, -0.0133,
        -0.0184, -0.0301,  0.0054,  0.0030,  0.0032, -0.0069, -0.0015, -0.0521,
         0.0007,  0.0035, -0.0218, -0.0093,  0.0065, -0.0225, -0.0007, -0.0057,
        -0.0056,  0.0207, -0.0501,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0406,  0.0771,  0.0573,  0.1144,  0.0519,  0.0094,  0.1079,  0.1374,
         0.0085,  0.1302,  0.0308,  0.0261,  0.0064,  0.0622,  0.0290,  0.0068,
         0.0040,  0.0260, -0.0057, -0.0093,  0.0590,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0373, -0.4133,  0.0145, -0.0669, -0.0648, -0.0516, -0.0105, -0.0326,
         0.0109,  0.0819,  0.0163,  0.0124,  0.0092,  0.0109, -0.0394,  0.0230,
         0.0141, -0.0035,  0.0868,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0303, -0.0827, -0.2059, -0.0759, -0.0310, -0.0145,  0.0239, -0.0373,
         0.0059, -0.0432, -0.1269,  0.0010, -0.0527, -0.0361,  0.0230, -0.0334,
        -0.0106, -0.0264,  0.1246, -0.0146,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0588, -0.2051, -0.0033, -0.1158, -0.0117,  0.0182, -0.0146, -0.0730,
        -0.0082,  0.0169, -0.0322, -0.0241, -0.0076, -0.0969, -0.0121, -0.0355,
        -0.1158,  0.0019, -0.0128, -0.0449,  0.0050, -0.0196,  0.0660,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1906, -0.0877, -0.0360,  0.0176, -0.0614, -0.0367, -0.0229, -0.0007,
        -0.0621, -0.0908, -0.0105, -0.0533,  0.0075, -0.1152,  0.0047,  0.0154,
         0.0225, -0.0138, -0.0139,  0.0191, -0.0571,  0.0098, -0.0127, -0.0215,
         0.0167,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0765,  0.2747,  0.0096,  0.0251,  0.0023,  0.0156,  0.0070,  0.0201,
        -0.0341, -0.0122,  0.1061,  0.0253,  0.0095, -0.0149, -0.0015,  0.0275,
        -0.0196,  0.0122, -0.0171, -0.0017,  0.0355,  0.0390,  0.0191,  0.0405,
         0.0253,  0.0146,  0.0044,  0.0045, -0.0086, -0.0073,  0.0191,  0.0085,
         0.0186, -0.0409, -0.0018,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0081,  0.0049,  0.0095, -0.1182, -0.0178, -0.0492, -0.1122, -0.0329,
        -0.0756, -0.0168, -0.0046, -0.0183,  0.0105, -0.0380, -0.0830, -0.1273,
        -0.0297, -0.1699,  0.0061,  0.0108, -0.0309, -0.0257,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0187,  0.0309, -0.0356, -0.0239,  0.0394,  0.0481,  0.0054,  0.0102,
         0.0584,  0.1947,  0.0974,  0.0250,  0.0178,  0.0263,  0.0738,  0.0916,
         0.0510,  0.0266,  0.0060,  0.0710,  0.0484,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0063, -0.3369, -0.0648,  0.0079, -0.0203, -0.0161, -0.0693, -0.0981,
         0.0992,  0.0454, -0.0135,  0.0004,  0.0180, -0.0244, -0.0189, -0.0452,
         0.0308, -0.0635, -0.0210,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0142,  0.0046,  0.0146, -0.0040,  0.0015,  0.0474, -0.0164,  0.0002,
        -0.0145, -0.0048, -0.0169,  0.0088,  0.0119,  0.0168, -0.0079, -0.0171,
         0.0125,  0.0552,  0.0100,  0.0215,  0.0193,  0.0194, -0.0028,  0.0060,
         0.0531,  0.0083,  0.0201,  0.0829,  0.0395,  0.1398,  0.0034,  0.0159,
         0.0421,  0.0579, -0.0012,  0.0048, -0.0288, -0.0094, -0.0269, -0.0003,
        -0.0151, -0.0158, -0.0156, -0.0066,  0.0186, -0.0160,  0.0294],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0501, -0.1750, -0.0784, -0.0190,  0.0140, -0.0006, -0.0179, -0.0053,
        -0.0295,  0.0094,  0.0095, -0.0075,  0.0095, -0.0524,  0.0107, -0.0240,
        -0.0443, -0.0162,  0.0258, -0.0143, -0.0086, -0.0530, -0.0791, -0.0854,
         0.0014,  0.0030, -0.0293,  0.0010,  0.0153, -0.0092, -0.0206,  0.0116,
        -0.0469, -0.0221,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0136, -0.2984,  0.0101,  0.0095, -0.0204, -0.0399, -0.0082,  0.0046,
        -0.0070, -0.0310, -0.0311, -0.0009, -0.0719, -0.0947, -0.0065, -0.0015,
         0.0094,  0.0425,  0.0258, -0.0427, -0.0119,  0.0024,  0.0043,  0.0070,
        -0.0057,  0.0079, -0.0068, -0.0097, -0.0140, -0.0280,  0.0189, -0.0128,
        -0.0372,  0.0636,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-9.5815e-03, -3.2021e-01,  2.9467e-02,  3.8336e-02, -5.6024e-02,
        -7.6332e-02,  9.8879e-03, -2.6356e-03, -1.0112e-04, -9.2780e-02,
        -3.4447e-02, -2.2312e-02, -4.6123e-02, -1.4020e-01, -3.3491e-02,
        -2.6987e-02,  2.1455e-02,  3.9637e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0154,  0.0241, -0.0092,  0.0934,  0.0051, -0.0204, -0.0064,  0.0262,
         0.0871,  0.2645, -0.0136, -0.0150,  0.0193,  0.0370,  0.0036, -0.0162,
         0.0233,  0.0545,  0.0171,  0.1513, -0.0037, -0.0163,  0.0011, -0.0157,
        -0.0066,  0.0354,  0.0184,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.8215e-02,  6.8026e-02, -1.4233e-02,  1.2026e-04, -6.9285e-03,
         2.6433e-02,  5.5942e-02,  1.7136e-02,  1.1294e-01, -3.5438e-02,
         3.8916e-03,  2.3613e-02,  2.0461e-01,  3.4795e-02, -1.6883e-02,
         5.6061e-03,  1.2521e-02, -9.2658e-03, -1.0094e-02,  1.8439e-01,
        -5.2057e-03,  2.3517e-02,  5.5689e-02, -1.0009e-02, -3.0822e-03,
         2.1421e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1332, -0.2589, -0.0215, -0.0818, -0.0814, -0.0314, -0.0292,  0.0120,
        -0.0177, -0.0277, -0.0510, -0.0124, -0.0181, -0.0236,  0.0046, -0.0502,
        -0.0034, -0.0211, -0.0075,  0.0741, -0.0393,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0208, -0.2168, -0.2023, -0.1408, -0.0309, -0.0263, -0.1961,  0.0117,
        -0.0348, -0.0404, -0.0008, -0.0424, -0.0070,  0.0154, -0.0135,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0974,  0.0498, -0.0391,  0.0103, -0.0320, -0.0328,  0.0050,  0.0283,
         0.0138, -0.0135,  0.0659, -0.0448,  0.2965,  0.1018, -0.0804,  0.0677,
        -0.0187, -0.0021,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0898, -0.2852, -0.1131, -0.1226, -0.0149, -0.0244,  0.0031, -0.0181,
        -0.0153, -0.0496, -0.0716, -0.0467,  0.0216, -0.0553, -0.0046,  0.0130,
         0.0039, -0.0085,  0.0388,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.0530e-02, -1.1050e-01, -8.4108e-02, -2.6225e-02, -1.8663e-02,
        -8.5443e-03,  3.6261e-02,  7.7682e-03, -4.3991e-03,  2.5276e-03,
         1.2196e-02,  3.7374e-03, -1.1123e-02,  1.6282e-03, -7.7853e-05,
        -1.4629e-02, -6.7350e-02, -4.1979e-03,  7.9390e-03,  1.3933e-02,
        -6.0147e-02, -3.5338e-03, -4.8131e-02, -2.8031e-02, -7.0229e-02,
        -4.3692e-03, -4.2717e-02,  1.1663e-02, -1.8250e-02, -2.1198e-02,
        -1.2895e-03, -9.4481e-03, -2.4458e-02, -9.7816e-03, -4.5645e-02,
         7.6644e-03, -3.7442e-02,  2.4865e-02, -7.0910e-03,  8.6773e-03,
         2.3265e-02,  1.6344e-02,  1.3174e-02, -1.6249e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0976, -0.1715, -0.0013, -0.0640, -0.0640, -0.0483, -0.0213,  0.0089,
        -0.1129, -0.0057, -0.0261,  0.0220,  0.0200,  0.0040,  0.0203, -0.0088,
        -0.0163, -0.0801, -0.0024, -0.0314, -0.0193,  0.0013,  0.0077, -0.0039,
         0.0095,  0.0020,  0.0067, -0.0673,  0.0007, -0.0014, -0.0063, -0.0008,
        -0.0083,  0.0262,  0.0116,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0316,  0.4171,  0.0470,  0.0369,  0.0063,  0.0273,  0.0853,  0.0726,
         0.0011,  0.0073, -0.0075,  0.0053,  0.0370, -0.0092, -0.0071,  0.0365,
         0.0035,  0.0083, -0.0045,  0.0018, -0.0038,  0.0024,  0.0306, -0.1103,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0456, -0.2192, -0.0198, -0.0169, -0.0237, -0.0087, -0.0116, -0.0491,
        -0.0695, -0.0005, -0.0070,  0.0112, -0.0100, -0.0487, -0.0603,  0.0043,
         0.0004,  0.0057, -0.0789, -0.0071, -0.0037, -0.0421,  0.0053, -0.0263,
         0.0163, -0.0495, -0.0172, -0.0399, -0.0105, -0.0226, -0.0289,  0.0047,
        -0.0138, -0.0212,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0788,  0.0557, -0.0018, -0.0167,  0.0326, -0.0007, -0.0858,  0.0030,
        -0.0015, -0.0099,  0.0136, -0.0153, -0.0987, -0.0106, -0.0462, -0.1673,
        -0.0064,  0.0098, -0.0026,  0.0139,  0.0118, -0.0189,  0.0363, -0.0755,
        -0.0385, -0.0111, -0.0453, -0.0224,  0.0551,  0.0140,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0317,  0.0097, -0.1040, -0.2923, -0.0713,  0.0104,  0.0150, -0.0564,
        -0.0111, -0.2444,  0.0224,  0.1008,  0.0304,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0316, -0.0149,  0.0589,  0.0529, -0.0142, -0.0484, -0.0456, -0.0746,
        -0.1651, -0.0076,  0.0307, -0.1985,  0.0103, -0.0719, -0.1111,  0.0046,
        -0.0300, -0.0290,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1596, -0.3224, -0.1721, -0.0282, -0.1730, -0.0180, -0.0907, -0.0360,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0202, -0.1476, -0.0101, -0.0058, -0.0013, -0.0118, -0.0130,  0.0006,
        -0.0023,  0.0128, -0.0026,  0.0059, -0.0002,  0.0171, -0.0025, -0.0056,
         0.0107,  0.0104,  0.0048,  0.0045,  0.0126, -0.0111, -0.0328, -0.0201,
        -0.0159, -0.0191, -0.0069, -0.0090, -0.0455, -0.0400, -0.0124, -0.0617,
        -0.0122, -0.1043,  0.0174, -0.0883, -0.0595, -0.0120,  0.0018,  0.0040,
        -0.0136, -0.0242, -0.0085,  0.0092,  0.0388, -0.0272,  0.0023],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0957, -0.0061, -0.1836,  0.2581, -0.0024, -0.0376, -0.0984,  0.0129,
        -0.1130,  0.1310,  0.0613,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0934,  0.2482,  0.0528,  0.1025,  0.0006, -0.0079,  0.0039,  0.1352,
         0.0680,  0.0709,  0.0061,  0.0195,  0.0861,  0.0049, -0.0387,  0.0486,
         0.0068,  0.0062,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0403, -0.0684, -0.2721,  0.0361,  0.0387,  0.0478, -0.0967,  0.0545,
        -0.1230, -0.0734,  0.0075,  0.0433,  0.0076,  0.0078, -0.0828,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0340, -0.3348, -0.0717, -0.0058, -0.0223, -0.1072, -0.1403, -0.0079,
         0.0430, -0.0514, -0.0744,  0.0126,  0.0331,  0.0106,  0.0507,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0773, -0.0023,  0.0538,  0.0249,  0.0120,  0.0235,  0.0082, -0.0158,
        -0.0544, -0.0335, -0.0265, -0.0542, -0.1337,  0.0075,  0.0130, -0.0426,
        -0.0239, -0.0059,  0.0339, -0.0335, -0.0907, -0.1203,  0.0189, -0.0213,
        -0.0319, -0.0366,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0545,  0.0231,  0.0407, -0.0534, -0.0548, -0.0133, -0.0050, -0.1831,
        -0.2633, -0.0744,  0.0138, -0.0244, -0.0584, -0.1101, -0.0276,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.4569e-04,  3.0438e-01, -4.6921e-03,  4.9753e-02,  7.7356e-02,
         2.9799e-02,  1.0260e-01, -1.4377e-02,  3.3423e-03,  1.4983e-01,
        -3.3535e-02, -9.5071e-03,  5.6569e-02,  8.4081e-03,  9.0239e-03,
        -5.4622e-03, -3.6001e-03, -1.3049e-04,  3.1064e-02, -1.7566e-02,
         2.3570e-02, -7.2189e-03, -6.6125e-03, -1.0180e-02, -9.4945e-03,
         3.1077e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0012, -0.0063,  0.0204,  0.0186, -0.0441, -0.0217,  0.0107, -0.0171,
        -0.0103, -0.0722, -0.1599,  0.0031, -0.0226,  0.0216, -0.1380, -0.0550,
         0.0458, -0.0770,  0.0055, -0.0619,  0.0032, -0.0245, -0.0041, -0.0838,
        -0.0413, -0.0302,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-9.2215e-02, -2.1126e-01,  1.0431e-02,  3.8724e-04, -5.7476e-04,
         3.0597e-02, -2.1284e-02, -4.7857e-02, -1.1234e-02, -3.1690e-02,
        -1.5591e-02, -2.4919e-02, -7.8423e-02, -1.0115e-02, -2.9728e-02,
        -1.5009e-02, -6.8294e-03, -1.2503e-02, -2.1969e-02, -3.6055e-03,
         9.9561e-03, -3.1398e-02,  9.4014e-04, -4.5846e-02, -5.3567e-02,
        -3.2497e-02, -3.6415e-02, -4.2077e-03,  3.0356e-02, -5.1025e-04,
        -7.8031e-05, -1.8535e-02,  6.7280e-03, -7.4529e-03, -5.1472e-03,
        -4.0144e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0155, -0.0963, -0.0305, -0.0146, -0.0411, -0.0276, -0.1006, -0.0769,
        -0.0055,  0.0239, -0.0206, -0.0030, -0.0167, -0.0029, -0.0016,  0.0075,
         0.0103, -0.0053,  0.0041,  0.0116, -0.0227, -0.0426, -0.0272, -0.0065,
        -0.0463, -0.0812,  0.0141, -0.0249, -0.0501, -0.0196,  0.0093, -0.0243,
        -0.0535,  0.0114, -0.0495,  0.0010,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1553,  0.1522,  0.0138, -0.0195, -0.0059,  0.1740,  0.0339,  0.0545,
         0.0200, -0.0118,  0.0601,  0.0085,  0.0719, -0.0134, -0.0358, -0.0085,
         0.0079,  0.0292,  0.0690, -0.0546,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0320, -0.1422, -0.0412, -0.0300, -0.0271, -0.0119, -0.0064, -0.0625,
        -0.0004, -0.0353, -0.0156, -0.0040, -0.0137,  0.0009, -0.1007, -0.0351,
        -0.0447, -0.0500, -0.0055, -0.0098,  0.0243, -0.0060,  0.0132, -0.0012,
        -0.0017,  0.0077, -0.0093,  0.0142, -0.0024, -0.0097, -0.0340, -0.0189,
         0.0022, -0.0772,  0.0382,  0.0277, -0.0147,  0.0283,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1008e-02, -1.9313e-01, -3.4150e-02, -5.3953e-02, -1.8920e-02,
        -1.7816e-02, -1.8539e-02,  1.5750e-02,  7.7216e-03,  1.7811e-02,
        -3.3435e-02, -5.6986e-02, -7.4147e-03, -4.8849e-03,  7.1038e-05,
        -7.1817e-02,  4.9972e-03, -6.5059e-02, -9.7670e-03,  4.2114e-03,
        -1.4944e-02,  1.9642e-02,  2.3288e-02, -1.3874e-02,  1.7242e-02,
        -2.5273e-02, -2.2852e-02, -1.9652e-02, -7.3287e-02,  2.3204e-03,
        -5.0821e-03, -1.7296e-02,  6.0551e-03, -2.1045e-03, -1.8053e-02,
         3.0999e-03, -7.1982e-03,  4.0280e-03,  1.3553e-03, -4.1384e-02,
        -1.4528e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0374,  0.4102, -0.0233,  0.0514, -0.0192,  0.0563,  0.0426,  0.0145,
         0.0082, -0.0018,  0.0265,  0.0372,  0.0444,  0.0012,  0.0106,  0.0541,
         0.0169,  0.0432,  0.0116, -0.0109, -0.0127, -0.0657,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0548, -0.0072, -0.0110, -0.0169,  0.0057,  0.0099, -0.0026,  0.0148,
        -0.0099,  0.0014, -0.0136, -0.0527,  0.0187, -0.0183,  0.0188, -0.0691,
         0.0139, -0.0622, -0.0048, -0.0191, -0.1104, -0.0130,  0.0180, -0.0477,
        -0.0049,  0.0160,  0.0003, -0.0307, -0.1323, -0.0139, -0.0173, -0.0484,
        -0.0517,  0.0529,  0.0169,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.4421e-01,  5.3098e-02,  2.0256e-01,  2.4030e-02,  6.3568e-02,
         8.4481e-03,  9.4526e-03, -2.8578e-02,  3.0964e-02,  1.7200e-02,
         2.0365e-02,  7.3022e-04,  1.0498e-02,  8.2812e-03, -2.9982e-03,
         1.8052e-02, -3.7136e-02,  3.2041e-02,  3.4764e-02,  1.4764e-02,
         4.6927e-02, -1.0503e-04,  1.1961e-02,  3.7434e-02,  1.3072e-03,
        -2.5032e-02,  2.7238e-02,  5.7018e-02, -3.1241e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0898,  0.2657,  0.0910,  0.0483,  0.0268,  0.0757, -0.0131,  0.0223,
         0.0146,  0.0175,  0.0310,  0.0456, -0.0013,  0.0090, -0.0089,  0.0352,
         0.0192,  0.0286,  0.0010, -0.0073,  0.0344,  0.0017,  0.0129,  0.0516,
        -0.0473,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0591, -0.0165, -0.0069,  0.0148, -0.1704, -0.0249, -0.0204, -0.1600,
        -0.1318,  0.0451,  0.0418,  0.0142, -0.0386, -0.0128, -0.0423, -0.1116,
         0.0006, -0.0086,  0.0097, -0.0164, -0.0310, -0.0019, -0.0204,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0550,  0.0021,  0.0065,  0.0150, -0.0026,  0.0095,  0.0049,  0.0032,
         0.0001,  0.0031, -0.0005, -0.0027, -0.0416, -0.1287, -0.0048, -0.0370,
        -0.0180, -0.0513, -0.0117, -0.0082, -0.0131, -0.0025, -0.0248, -0.0053,
        -0.0603, -0.0268, -0.0997, -0.0027, -0.0726, -0.0212, -0.0552, -0.0160,
        -0.0403, -0.0023, -0.0125, -0.0260,  0.0178, -0.0293,  0.0652,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0285,  0.0289, -0.0250, -0.1093, -0.2676,  0.0091, -0.0416, -0.1356,
        -0.1722,  0.0332, -0.0726,  0.0765,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.0226, -0.0709, -0.2248, -0.0121, -0.0140, -0.0367, -0.0765,  0.0035,
        -0.0307, -0.0269,  0.0029, -0.0303,  0.0004, -0.0071, -0.0263,  0.0021,
        -0.0191, -0.0472,  0.0015, -0.0169, -0.0180, -0.0277,  0.0126,  0.0005,
        -0.0039,  0.0134,  0.0037, -0.0300, -0.0498, -0.0129,  0.0274,  0.0039,
        -0.0261,  0.0012,  0.0008,  0.0010, -0.0473, -0.0031, -0.0072, -0.0067,
         0.0039, -0.0021,  0.0124, -0.0047, -0.0077,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0673,  0.3428,  0.0748,  0.0956,  0.0474, -0.0064, -0.0798,  0.0588,
         0.0945,  0.0598,  0.0612,  0.0116,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0155, -0.6318, -0.0540, -0.1482,  0.0134,  0.0115, -0.0590,  0.0665,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0558,  0.0955,  0.0420,  0.0316,  0.0488,  0.0027, -0.0058,  0.0142,
         0.0187,  0.0635,  0.0129,  0.0011,  0.0031, -0.0054, -0.0101, -0.0063,
         0.0136,  0.0107,  0.0048,  0.0070, -0.0017,  0.0085,  0.0079, -0.0018,
         0.0023, -0.0016, -0.0090,  0.0144,  0.0254,  0.0069,  0.0076, -0.0028,
        -0.0031,  0.0103,  0.0068, -0.0058, -0.0160, -0.0074, -0.0047,  0.0015,
         0.0077,  0.0058,  0.0103,  0.0073,  0.0023, -0.0865,  0.0533, -0.0143,
         0.0493,  0.0625,  0.0276,  0.0006,  0.0077, -0.0035,  0.0122, -0.0050,
         0.0096, -0.0032, -0.0070, -0.0159,  0.0095], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0583, -0.0209,  0.0161, -0.0110, -0.0662,  0.0060, -0.0294,  0.0011,
        -0.0032,  0.0165,  0.0170, -0.0255, -0.0109, -0.0482, -0.0695, -0.0121,
         0.0021, -0.0067, -0.0535, -0.0284, -0.1279, -0.0611, -0.0694, -0.0057,
        -0.0698, -0.0128, -0.0288, -0.0199, -0.0192, -0.0352, -0.0029, -0.0045,
        -0.0401,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0261, -0.1280, -0.1954, -0.0528, -0.0778, -0.0107, -0.0788,  0.0104,
         0.0101,  0.0057, -0.0093, -0.0109, -0.0190,  0.0087,  0.0028, -0.0102,
         0.0024, -0.0801, -0.1922,  0.0162, -0.0085, -0.0341,  0.0077,  0.0020,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1165, -0.0865, -0.0951, -0.0252, -0.0283, -0.0830, -0.0121, -0.0273,
        -0.0186, -0.0843, -0.0693, -0.0736, -0.1027, -0.0037, -0.0241,  0.0057,
         0.0083,  0.0193,  0.0313,  0.0060, -0.0342, -0.0081, -0.0082,  0.0286,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1086,  0.0085, -0.0317, -0.0765,  0.0040, -0.0013, -0.0482, -0.0581,
        -0.0086, -0.0166, -0.0057, -0.0029,  0.0022, -0.0013, -0.0578, -0.0761,
        -0.0654,  0.0005, -0.0121, -0.0455, -0.0693,  0.0032, -0.0227, -0.0977,
        -0.0135, -0.0233, -0.0755, -0.0158,  0.0009,  0.0079, -0.0086, -0.0203,
         0.0026, -0.0072,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.9024e-02, -2.0527e-03, -5.4096e-03, -5.2478e-03,  1.0792e-01,
        -3.5779e-02, -1.1375e-03,  1.6550e-02, -1.5516e-02, -1.6903e-02,
        -1.3845e-04,  3.0532e-02,  4.4775e-02,  1.5928e-01,  1.5972e-02,
        -5.1930e-02,  1.3381e-01,  1.0165e-02,  5.2232e-02,  3.5656e-02,
         6.3005e-03,  8.1323e-02, -6.5607e-02, -2.4467e-02,  5.2265e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0672,  0.0335,  0.0114, -0.0016, -0.0144, -0.0253, -0.1333, -0.0951,
        -0.0946, -0.3705, -0.0070, -0.0069, -0.0108,  0.0254,  0.0218, -0.0811,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1174, -0.2394,  0.0244, -0.0886, -0.0298, -0.0381, -0.1675, -0.0307,
         0.0096, -0.0106,  0.0064, -0.0317, -0.0221, -0.0046, -0.0067, -0.0021,
        -0.0084, -0.0490, -0.0313,  0.0058,  0.0004, -0.0425,  0.0329,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2234, -0.0226,  0.0352,  0.1197,  0.2661, -0.0305, -0.0628,  0.0168,
         0.0051, -0.0196,  0.0218, -0.0703,  0.0512,  0.0549,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.0523,  0.0814,  0.0891,  0.1874, -0.0148, -0.0242,  0.0804,  0.0010,
         0.0658,  0.0244,  0.0406,  0.0703, -0.0027,  0.0085, -0.0048, -0.0240,
         0.0175,  0.0503,  0.0153,  0.0844, -0.0008,  0.0201, -0.0061, -0.0338,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0685, -0.0116, -0.1472, -0.0670, -0.2007, -0.0277, -0.3119, -0.0236,
        -0.0504, -0.0438, -0.0476,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0054,  0.0340, -0.0140, -0.0004, -0.0186, -0.0361, -0.2094, -0.0127,
        -0.2068, -0.0540, -0.0185, -0.0221, -0.0364, -0.0057, -0.0683, -0.0191,
        -0.0043, -0.1253, -0.0486,  0.0360, -0.0242,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0669,  0.0693, -0.0686, -0.0057, -0.0372,  0.0041, -0.0288, -0.2254,
        -0.0950, -0.2032, -0.0345,  0.0082, -0.0324,  0.1209,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0153,  0.0949,  0.0429,  0.0078, -0.0144, -0.0226,  0.0176,  0.0121,
         0.0086, -0.0108,  0.0400,  0.1546,  0.3323,  0.0170,  0.0285,  0.0658,
         0.0198,  0.0257,  0.0147, -0.0251,  0.0294,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0223, -0.3663, -0.0231, -0.0821,  0.0863, -0.1416, -0.0601, -0.0611,
        -0.0288, -0.0489, -0.0406, -0.0287,  0.0101,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.8015e-03, -9.7229e-02, -1.3302e-02, -1.4651e-02, -1.3121e-02,
        -2.2290e-02, -4.9110e-02, -9.9964e-02, -3.4803e-02, -1.5318e-02,
        -9.1309e-03,  2.2596e-02,  1.2954e-02,  9.6021e-03, -2.6848e-03,
         5.7121e-03, -6.6066e-06,  2.0515e-03, -8.3705e-04,  2.5741e-02,
         3.5817e-03,  9.4253e-03,  3.9195e-03, -4.1850e-02, -1.8155e-01,
         1.2921e-02, -4.6595e-02, -6.4911e-02, -2.8608e-02, -3.9933e-05,
        -2.5946e-02,  9.7662e-03,  7.1292e-03, -1.8680e-02, -2.1052e-03,
        -6.0493e-03, -2.2123e-02, -8.4360e-04, -1.0153e-03, -6.5261e-03,
        -1.6112e-02, -1.2654e-03,  4.5769e-03,  1.8217e-02,  6.3397e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0451,  0.0787,  0.0483,  0.0179, -0.0116,  0.0144,  0.0160,  0.0209,
         0.0072,  0.0005,  0.0308,  0.0085,  0.0073,  0.0065,  0.0385,  0.0231,
         0.1258, -0.0243,  0.0039,  0.0119,  0.0370,  0.0782,  0.0308,  0.0874,
        -0.0096,  0.0050,  0.0285,  0.0312,  0.0787,  0.0134,  0.0167, -0.0156,
        -0.0228, -0.0038,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0295,  0.0421,  0.0004,  0.0438,  0.0230,  0.0029, -0.0273,  0.0203,
        -0.0074,  0.0377,  0.0097,  0.0206, -0.0043,  0.0278,  0.0208, -0.0284,
        -0.0178, -0.0150,  0.0259,  0.1777,  0.1842, -0.0635, -0.0238, -0.0257,
         0.1012,  0.0194,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0017, -0.3385,  0.0343, -0.0720,  0.0329, -0.0651, -0.0104, -0.0855,
        -0.0416, -0.0202,  0.0096,  0.0159, -0.0036,  0.0085, -0.0100, -0.0318,
        -0.1134,  0.0182, -0.0128,  0.0104,  0.0019,  0.0013,  0.0605,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0207, -0.1252, -0.0179, -0.0226,  0.0004, -0.0068, -0.0318, -0.0392,
        -0.0695, -0.0130,  0.0236, -0.0216, -0.0075, -0.0070, -0.1635,  0.0204,
         0.0068, -0.0306, -0.0695, -0.0177,  0.0006, -0.0141,  0.0055,  0.0278,
         0.0036, -0.0009, -0.0464, -0.0858,  0.0093, -0.0027,  0.0027, -0.0031,
         0.0044, -0.0011,  0.0052,  0.0070, -0.0281,  0.0037, -0.0108,  0.0218,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0356, -0.0604, -0.0127, -0.0107, -0.1182, -0.0141,  0.0818, -0.0930,
         0.0024, -0.0034, -0.0142, -0.0024, -0.0077, -0.0196, -0.0515, -0.0612,
        -0.0202, -0.0060, -0.0115, -0.0167, -0.0190, -0.0240, -0.0469,  0.0078,
        -0.0113, -0.0065, -0.0073, -0.0828, -0.0112,  0.0238, -0.0067, -0.0311,
        -0.0696,  0.0086,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0125, -0.0866, -0.0064, -0.0157, -0.0539,  0.0242,  0.0083, -0.0197,
         0.0068,  0.0082, -0.0033, -0.0081,  0.0156, -0.0132, -0.0926, -0.0062,
        -0.0224, -0.0076, -0.0582, -0.0195,  0.0018,  0.0068, -0.0441,  0.0031,
        -0.0378, -0.0909, -0.0227, -0.0293, -0.0135,  0.0115,  0.0234,  0.0014,
         0.0095,  0.0022, -0.0342,  0.0020, -0.0116, -0.0325, -0.0050, -0.0114,
        -0.0194, -0.0071, -0.0135, -0.0046, -0.0013,  0.0031, -0.0010,  0.0008,
        -0.0053, -0.0178,  0.0054, -0.0368], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0489, -0.0161, -0.0276, -0.0809, -0.0873,  0.0057,  0.0128,  0.0201,
         0.0113, -0.0215,  0.0060, -0.0043, -0.0232, -0.1527, -0.0219, -0.0532,
        -0.1185, -0.0636, -0.0499, -0.0274, -0.0100, -0.0308, -0.0093, -0.0501,
        -0.0016,  0.0097, -0.0083, -0.0011,  0.0210,  0.0037, -0.0016,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0783, -0.1917,  0.0269,  0.0070,  0.0191,  0.0112,  0.0122,  0.0018,
         0.0032,  0.0292, -0.0032, -0.0805, -0.0117, -0.0180, -0.0492, -0.0551,
         0.0058, -0.0534, -0.1095, -0.0927, -0.0441, -0.0042, -0.0245, -0.0423,
         0.0251,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1225,  0.0737, -0.0362, -0.0250, -0.1035, -0.1709, -0.3150, -0.0118,
        -0.0117, -0.0197, -0.0519,  0.0580,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0475, -0.3041, -0.0176, -0.0417, -0.0662,  0.0031, -0.0032, -0.0087,
         0.0018, -0.0082,  0.0146, -0.0141, -0.0385, -0.0991, -0.0114, -0.0407,
        -0.0050, -0.0107,  0.0086,  0.0017, -0.0298, -0.0150,  0.0025, -0.0515,
        -0.0320, -0.0855, -0.0129,  0.0238, -0.0004,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0008, -0.3521, -0.0368,  0.0346, -0.0173, -0.1292, -0.1622,  0.1321,
         0.0731,  0.0263,  0.0035, -0.0320,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0159, -0.3642,  0.0337, -0.1118,  0.0358, -0.1128, -0.1974,  0.0256,
        -0.0338,  0.0120,  0.0113, -0.0457,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1338,  0.0562, -0.1032, -0.0317, -0.1499, -0.0638,  0.0098, -0.0275,
        -0.1172,  0.0040, -0.0036, -0.0594, -0.0100,  0.0117, -0.0129, -0.0982,
         0.0505,  0.0218, -0.0069, -0.0279,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0094, -0.2258,  0.0246, -0.0867, -0.1327, -0.0449, -0.0222,  0.0124,
        -0.0256, -0.0178, -0.0194, -0.0289, -0.0040,  0.0043,  0.0213,  0.0089,
        -0.0171,  0.0107,  0.0057, -0.0125, -0.0234,  0.0321,  0.0189, -0.0028,
        -0.0065, -0.0097, -0.0100, -0.0222, -0.0108, -0.0006,  0.0080, -0.0014,
        -0.0378, -0.0639, -0.0034,  0.0058,  0.0080,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1954,  0.6000, -0.0601,  0.0195,  0.0310, -0.0036, -0.0414, -0.0288,
         0.0202,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0421,  0.0017,  0.0176, -0.1311, -0.0324, -0.0123, -0.0612, -0.1480,
         0.0324,  0.0115, -0.0264, -0.0312, -0.0239, -0.0808,  0.0017,  0.0151,
         0.0006, -0.0059, -0.0572, -0.0451,  0.0360,  0.0090, -0.0148, -0.0004,
         0.0049,  0.0240,  0.1328,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 6.3144e-02,  2.6644e-01, -1.9475e-02,  5.3233e-02,  1.6149e-02,
         1.8794e-02,  6.8179e-05,  1.2122e-02,  1.6175e-02,  1.9932e-02,
        -1.3121e-02,  6.2660e-02,  9.0722e-03, -1.7370e-02, -1.4407e-03,
         3.9958e-03,  3.2183e-02, -1.2151e-02, -6.1166e-03,  1.6674e-03,
        -1.8113e-02,  3.7795e-02,  3.1910e-02, -1.1063e-02, -9.5684e-04,
        -1.6118e-03, -1.1153e-03,  1.6234e-02,  1.4733e-01,  8.0831e-03,
         4.4531e-03, -4.3422e-03, -4.3743e-02,  2.7940e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0081, -0.0129, -0.0154, -0.0186, -0.0241, -0.0161,  0.0044, -0.0055,
        -0.0051,  0.0047, -0.0098, -0.0841, -0.0261, -0.0163,  0.0083, -0.0516,
        -0.1046, -0.0223, -0.0180,  0.0163,  0.0034,  0.0012, -0.0545, -0.0180,
        -0.0081, -0.0032, -0.0159, -0.0167, -0.0155, -0.0065, -0.0111, -0.0963,
        -0.0563, -0.0023, -0.0118, -0.0322, -0.0395,  0.0138,  0.0196, -0.0017,
        -0.0159, -0.0342, -0.0134, -0.0046, -0.0319,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0043, -0.0152, -0.0323, -0.0501,  0.0060, -0.1212, -0.2245, -0.0219,
         0.0530, -0.1444, -0.0741, -0.0137, -0.0749,  0.0316, -0.0382, -0.0945,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0012, -0.0138,  0.0511,  0.0348,  0.0469,  0.0257, -0.0123,  0.0229,
         0.1416,  0.1304,  0.0676,  0.1178,  0.0177,  0.0500,  0.0296, -0.0265,
        -0.0172, -0.0048, -0.0018,  0.0076, -0.0019, -0.0083, -0.0014,  0.0089,
        -0.0382,  0.0302,  0.0898,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0316, -0.0088, -0.0514, -0.0035, -0.0236, -0.0157, -0.0291, -0.0010,
        -0.0218, -0.0106, -0.0043, -0.0129, -0.0231, -0.0022, -0.0090, -0.0077,
        -0.0230,  0.0047, -0.0039, -0.0181,  0.0215, -0.0152, -0.0183, -0.1058,
         0.0092, -0.0077, -0.0422, -0.0532,  0.0047, -0.1002,  0.0037, -0.0576,
        -0.0332, -0.0099, -0.0131, -0.0297, -0.0043, -0.0346,  0.0291,  0.0193,
        -0.0013, -0.0429,  0.0006, -0.0069,  0.0197,  0.0100], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0073, -0.0433, -0.0032,  0.0198,  0.0073,  0.0075, -0.0089,  0.0509,
         0.0079,  0.0035,  0.0103,  0.0142, -0.0218, -0.0166,  0.0076, -0.0067,
        -0.0303, -0.0108,  0.1308,  0.2823,  0.1014,  0.0440,  0.0184,  0.0175,
        -0.0030, -0.0106,  0.0629,  0.0309,  0.0092,  0.0112,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1371,  0.0377, -0.0435,  0.0238,  0.0316,  0.0715,  0.0925,  0.2298,
         0.1704,  0.0028,  0.1594,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0028, -0.2785, -0.0021, -0.0616, -0.0357, -0.0110, -0.0520, -0.1637,
        -0.0126, -0.1017,  0.0012, -0.0142,  0.0115,  0.0055, -0.0652, -0.0603,
        -0.0211, -0.0357,  0.0017, -0.0620,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1000, -0.1925, -0.2168, -0.1266,  0.0074, -0.0514, -0.0848, -0.0213,
        -0.0822, -0.0450, -0.0719,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.6753e-02,  2.6666e-01,  1.4328e-01,  1.1385e-01, -1.1535e-02,
         2.1814e-02,  3.2735e-02,  4.6576e-02,  7.7374e-02, -9.3240e-03,
         4.0809e-02, -2.5690e-04,  1.9183e-02,  2.0344e-02,  3.6267e-03,
         2.4666e-02, -4.7354e-03,  1.3845e-02,  1.8789e-03,  4.0063e-02,
        -8.0692e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0142,  0.2102,  0.1698,  0.0176,  0.0121,  0.2455, -0.0628, -0.0280,
         0.0347,  0.0162, -0.0201,  0.0009, -0.0138,  0.0074,  0.0047, -0.0144,
        -0.0665,  0.0610,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0787, -0.0460, -0.1724, -0.0079,  0.0091,  0.0258,  0.0035,  0.0118,
         0.0058,  0.0020,  0.0165, -0.0002,  0.0160, -0.0409, -0.0581,  0.0227,
        -0.0042, -0.0594, -0.0188, -0.0155, -0.0063, -0.0354, -0.1247, -0.0115,
        -0.0349, -0.0050, -0.0330, -0.0076, -0.0138, -0.0145, -0.0010,  0.0008,
         0.0023, -0.0043, -0.0133, -0.0069, -0.0036,  0.0409, -0.0249,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0501,  0.3400,  0.0040,  0.0618,  0.0043, -0.0101, -0.0219, -0.0096,
         0.0244,  0.0042,  0.0245,  0.0633, -0.0263,  0.0021, -0.0063,  0.0199,
         0.0601,  0.0071, -0.0052,  0.0108,  0.0127,  0.0078, -0.0097, -0.0078,
        -0.0093,  0.0042,  0.0205,  0.0076,  0.0423,  0.0621,  0.0046,  0.0040,
         0.0067, -0.0447,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.1472, -0.0778, -0.2370, -0.1894, -0.0542, -0.1467, -0.0087, -0.0044,
        -0.1346,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.5866e-02, -1.8817e-01, -2.5526e-02, -7.8990e-02, -3.2344e-02,
        -1.1462e-01,  1.9995e-02,  1.5082e-02,  2.7919e-03,  1.4631e-02,
        -8.1800e-03, -7.3005e-03,  2.4237e-03, -5.1761e-03,  9.3259e-03,
        -2.7942e-02, -4.8560e-03, -1.5768e-01, -1.6641e-02,  9.7628e-03,
        -2.3299e-03,  1.2006e-02, -1.2114e-02, -8.4320e-03, -3.3924e-02,
        -1.0342e-02, -2.1049e-02,  1.3233e-02, -3.8365e-03,  8.3654e-03,
        -9.1296e-03, -3.5814e-02,  1.3352e-04, -1.9955e-02,  1.2202e-02,
        -1.0058e-03,  1.5774e-02, -1.2256e-03, -4.3925e-03, -7.4274e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0539, -0.1540, -0.1187, -0.0063, -0.0465, -0.0019,  0.0090, -0.0026,
        -0.0190, -0.0929, -0.0018,  0.0020, -0.0002,  0.0060,  0.0061,  0.0172,
        -0.0013, -0.0094, -0.0207,  0.0080, -0.0040, -0.0042,  0.0166, -0.0011,
        -0.0061, -0.0191, -0.0377,  0.0078, -0.0507, -0.0585, -0.0204, -0.0083,
         0.0004,  0.0025,  0.0079, -0.0041, -0.0339, -0.0442,  0.0032, -0.0024,
         0.0251,  0.0026, -0.0009,  0.0096, -0.0026,  0.0031,  0.0016,  0.0064,
         0.0045,  0.0038, -0.0032, -0.0259,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1328,  0.2273,  0.0405,  0.0357,  0.0826,  0.0100,  0.0464,  0.0099,
         0.0144, -0.0060, -0.0296,  0.0336,  0.0466, -0.0022,  0.0391,  0.0678,
         0.0271, -0.0133,  0.0872, -0.0145, -0.0202,  0.0133,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1408e-01,  4.2041e-01,  4.4540e-02,  3.1845e-03, -1.0548e-02,
         2.4558e-02,  1.6615e-02,  1.0009e-02,  1.2283e-02,  9.4295e-03,
         1.5739e-02,  3.8064e-02,  2.2314e-02,  3.4106e-02,  3.2294e-02,
         5.6095e-02, -5.5374e-05, -7.8737e-04, -2.9244e-02, -3.6048e-02,
        -6.9602e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0286, -0.1371, -0.0292, -0.0151,  0.0105, -0.0129,  0.0031,  0.0144,
        -0.0233, -0.0051, -0.0045, -0.0041, -0.0032, -0.0129,  0.0007, -0.0028,
         0.0092,  0.0003, -0.0121, -0.0050,  0.0080,  0.0217, -0.0066, -0.0077,
        -0.0039,  0.0020, -0.0073, -0.0253,  0.0053, -0.0103, -0.0307, -0.0675,
        -0.0033, -0.0293, -0.0273,  0.0117, -0.0069,  0.0030, -0.0016,  0.0025,
        -0.1641, -0.0117, -0.0165, -0.0015, -0.0234,  0.0016, -0.0348, -0.0183,
         0.0100,  0.0145, -0.0051, -0.0011,  0.0185, -0.0078, -0.0549],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0736, -0.0182,  0.0201, -0.0096, -0.0142,  0.0014, -0.1007, -0.0087,
        -0.0272,  0.0176, -0.0183, -0.0018, -0.0044, -0.0079, -0.0403, -0.0143,
        -0.0166, -0.0209, -0.0290, -0.0519, -0.0212, -0.0170, -0.0328,  0.0094,
         0.0076,  0.0093,  0.0112, -0.0445, -0.0136, -0.0003, -0.0006, -0.0075,
        -0.0283, -0.0747, -0.0018, -0.0285, -0.0148, -0.0313, -0.0388, -0.0201,
        -0.0096, -0.0007, -0.0059, -0.0553, -0.0184,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.6742e-02,  1.8223e-01,  8.0489e-02,  7.7054e-02, -2.1791e-02,
         2.2522e-02,  1.3640e-02, -4.2463e-03,  6.6176e-02, -1.8599e-02,
         3.6725e-02,  4.0397e-02, -1.0427e-05,  4.5561e-02,  5.9282e-02,
         1.9009e-02,  1.6060e-02,  7.2438e-02,  6.5820e-03,  1.9776e-02,
         7.6619e-02, -1.1499e-02,  1.8085e-02, -1.4463e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0505,  0.0477,  0.0160, -0.0042, -0.0076, -0.0150, -0.0485, -0.0760,
        -0.0013,  0.0342, -0.0061, -0.0163, -0.0010, -0.0754, -0.0037,  0.0238,
        -0.0390, -0.0524, -0.0404, -0.0533, -0.0103,  0.0228, -0.0339, -0.0480,
         0.0247, -0.0379, -0.0790, -0.0097, -0.0246, -0.0398, -0.0176,  0.0108,
        -0.0286,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1286,  0.3897,  0.0174,  0.0695, -0.0135,  0.0064,  0.0308, -0.0266,
         0.1489, -0.0044,  0.0047, -0.0116,  0.0075,  0.0178,  0.0046, -0.0220,
         0.0107, -0.0024,  0.0832,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0166, -0.0543,  0.0882, -0.0348, -0.1109, -0.0331, -0.0633, -0.0725,
        -0.0864, -0.0287, -0.0258,  0.0389,  0.0116,  0.0153, -0.0183, -0.0116,
        -0.0325, -0.0375, -0.0025, -0.0091, -0.0887, -0.0899, -0.0141, -0.0120,
        -0.0033,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0509, -0.3145, -0.0836, -0.0092, -0.0357, -0.0994, -0.1187,  0.0794,
        -0.0204, -0.0369, -0.0038,  0.0252, -0.0370, -0.0330, -0.0435, -0.0050,
        -0.0038,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.1656,  0.0006, -0.0146, -0.0063, -0.0150,  0.0042, -0.0084, -0.0082,
         0.0087, -0.0489, -0.0159, -0.0126,  0.0038, -0.0017, -0.0068, -0.0093,
        -0.0614, -0.1430, -0.0013, -0.0540, -0.0072, -0.1000, -0.1408,  0.0078,
        -0.0218,  0.0484, -0.0065, -0.0189, -0.0045, -0.0036, -0.0163,  0.0341,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.6304e-02, -1.0858e-01, -2.1282e-01, -1.8497e-02, -3.5446e-02,
        -7.3282e-03, -4.1979e-02, -8.2138e-02, -2.1060e-02,  4.4909e-03,
         1.8624e-02,  1.9181e-02,  8.4367e-03, -9.9050e-03, -3.5695e-02,
         1.1622e-02, -9.7198e-06, -1.8774e-03, -5.4965e-02,  3.0675e-02,
        -2.9413e-02,  1.4368e-03, -5.8977e-02, -4.6296e-02,  2.7996e-02,
        -1.1218e-02, -5.3219e-02, -1.1813e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0491, -0.1221, -0.0087,  0.0116,  0.0160, -0.0707, -0.1301, -0.0143,
        -0.0256,  0.0182,  0.0181, -0.0446, -0.0145, -0.0497, -0.0171, -0.1084,
        -0.0240, -0.0814, -0.0187,  0.0270, -0.0123,  0.0038, -0.0358, -0.0036,
        -0.0251,  0.0195,  0.0128, -0.0095,  0.0078,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0048, -0.0666, -0.0247, -0.0598, -0.1453, -0.0154, -0.0162, -0.0208,
        -0.0101,  0.0073, -0.0017,  0.0027, -0.0294, -0.1719,  0.0270, -0.0426,
        -0.0117, -0.1060,  0.0016, -0.0913, -0.0027, -0.0147, -0.0151, -0.0136,
        -0.0324, -0.0133, -0.0245, -0.0118, -0.0020,  0.0042, -0.0029,  0.0043,
        -0.0018,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0387, -0.1860, -0.0119, -0.0525, -0.0134, -0.1013, -0.0226, -0.0517,
        -0.0286, -0.0290, -0.0642, -0.0024, -0.0103, -0.0121, -0.0367,  0.0119,
        -0.0437,  0.0051, -0.1055, -0.0104, -0.0899,  0.0229, -0.0340, -0.0151,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0245, -0.2082, -0.0952, -0.0227, -0.0541,  0.0092, -0.0266, -0.0582,
         0.0044, -0.0713, -0.0257, -0.0399, -0.0425, -0.0281,  0.0022, -0.0211,
        -0.0145, -0.0636, -0.0087, -0.1138, -0.0260, -0.0340, -0.0056,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0368, -0.1560,  0.0059, -0.0017, -0.0103, -0.0419,  0.0024,  0.0002,
        -0.0224, -0.0032,  0.0169, -0.0075,  0.0017,  0.0268,  0.0176, -0.0203,
        -0.0474, -0.0192, -0.0350,  0.0010, -0.0059, -0.0408, -0.0066, -0.0143,
        -0.0438, -0.0260, -0.0068, -0.0181,  0.0059, -0.1151, -0.0723, -0.0072,
        -0.0046,  0.0119, -0.0173, -0.0438, -0.0241, -0.0346,  0.0268,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0097, -0.0953, -0.0237, -0.0130, -0.1432, -0.0178,  0.0053,  0.0039,
        -0.0040,  0.0011,  0.0044, -0.0005, -0.0052,  0.0010,  0.0053, -0.0398,
        -0.0335,  0.0109, -0.0139,  0.0058, -0.0468, -0.0012, -0.0026, -0.0284,
        -0.0744, -0.0980, -0.0172, -0.0268,  0.0010, -0.0457, -0.0443, -0.0250,
        -0.0120, -0.0113, -0.0038,  0.0026, -0.0082, -0.0477, -0.0103, -0.0013,
        -0.0121, -0.0185,  0.0232,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0482, -0.1811,  0.0127,  0.0062, -0.0223, -0.0069, -0.0309, -0.0032,
        -0.0510, -0.0093,  0.0052, -0.0036, -0.0199, -0.0041, -0.0052, -0.0114,
        -0.0363, -0.0466, -0.0051, -0.0119, -0.0013,  0.0042, -0.0176, -0.0028,
        -0.0018, -0.0064, -0.0096, -0.0112, -0.0083,  0.0098,  0.0023, -0.0235,
        -0.0743,  0.0107, -0.0267, -0.0062, -0.0247, -0.0220, -0.0401, -0.0118,
        -0.0171, -0.0267,  0.0004,  0.0181, -0.0025,  0.0159, -0.0018,  0.0022,
         0.0033, -0.0032,  0.0079,  0.0039,  0.0085, -0.0056, -0.0094,  0.0186,
        -0.0186], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0771, -0.4697, -0.0142, -0.0093, -0.0438, -0.0122, -0.0766, -0.0890,
        -0.0051,  0.0123, -0.0403,  0.0216, -0.0730,  0.0559,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0283, -0.1130, -0.0558,  0.0440, -0.0174, -0.0182, -0.0242, -0.0064,
        -0.0198, -0.0035, -0.0340, -0.1250,  0.0107, -0.0515, -0.0632,  0.0055,
        -0.0148, -0.0412,  0.0185, -0.0518, -0.0795, -0.0239, -0.0271, -0.0031,
         0.0338,  0.0175, -0.0012,  0.0112, -0.0396,  0.0162,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2914,  0.1819,  0.0195,  0.2289,  0.0018, -0.0204,  0.0808, -0.1752,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0601, -0.0816, -0.0318, -0.0235, -0.0274, -0.1043, -0.0021, -0.0199,
        -0.0021, -0.0293, -0.0048, -0.0243, -0.0508, -0.0821, -0.0190, -0.0095,
        -0.0282,  0.0014, -0.0068,  0.0030,  0.0205, -0.0118, -0.0145,  0.0121,
        -0.0311, -0.0091, -0.0229, -0.0190, -0.0753, -0.0783,  0.0361, -0.0406,
         0.0168,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0882, -0.1411, -0.0385, -0.0531,  0.0149, -0.0159, -0.0151, -0.0354,
         0.0280, -0.0149,  0.0023,  0.0027,  0.0034,  0.0064, -0.0040,  0.0025,
         0.0037,  0.0082, -0.0085,  0.0059, -0.0030,  0.0012, -0.0024,  0.0083,
         0.0144, -0.0013, -0.0190, -0.0198,  0.0075, -0.0010, -0.0009, -0.0114,
        -0.0352, -0.0462,  0.0174, -0.0040, -0.0154,  0.0052, -0.0002,  0.0022,
         0.0021, -0.0081, -0.0058, -0.0270,  0.0023, -0.0051,  0.0052,  0.0087,
        -0.0026, -0.0022, -0.0007, -0.0074, -0.0040, -0.0141, -0.0150,  0.0065,
        -0.0067,  0.0138, -0.0008, -0.0122, -0.0087, -0.0002, -0.0458, -0.0179,
        -0.0048, -0.0018, -0.0023,  0.0244,  0.0152,  0.0177,  0.0020],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.7379e-03, -8.0192e-03, -2.3271e-02, -3.1760e-03, -3.8905e-02,
        -7.6492e-03, -2.0694e-01, -1.1010e-02, -7.3906e-04,  1.0331e-03,
         3.3292e-03,  1.9271e-03,  9.4132e-03, -5.2140e-03, -2.0027e-02,
        -5.3913e-04, -8.1477e-03,  7.9892e-03, -2.1423e-02, -1.1226e-01,
        -7.9174e-04,  1.9074e-02,  1.9790e-02, -5.7668e-02,  3.0780e-02,
        -4.8064e-02, -2.7890e-02, -1.1886e-04, -2.8498e-03, -3.5276e-02,
         2.3509e-03, -1.8763e-02, -7.2035e-02, -4.1907e-02,  2.7616e-02,
        -1.5085e-02, -4.5550e-02, -5.2616e-03, -9.4383e-04, -1.6257e-02,
         1.3817e-02, -1.3665e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0603, -0.1625, -0.0227, -0.0075, -0.0066,  0.0107,  0.0128,  0.0181,
        -0.0063, -0.0970, -0.0255, -0.0191, -0.0844, -0.0037,  0.0027,  0.0005,
         0.0082, -0.0399, -0.0684,  0.0041, -0.0135, -0.0108,  0.0128,  0.0174,
        -0.0039, -0.0585, -0.0229, -0.0138, -0.0707, -0.0219,  0.0109, -0.0032,
        -0.0026, -0.0100, -0.0250, -0.0334,  0.0074,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0117, -0.5172, -0.0522, -0.0556,  0.0257,  0.0523,  0.0377,  0.0104,
         0.0105, -0.0162, -0.0388,  0.0098, -0.0767, -0.0735, -0.0117,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0550,  0.0615,  0.0140,  0.0296,  0.0688,  0.1284,  0.0521,  0.0578,
         0.1887,  0.0341, -0.0205,  0.0220,  0.0697,  0.0061, -0.0319, -0.0545,
         0.0204, -0.0098, -0.0247, -0.0504,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1431e-03,  2.3562e-02, -1.4930e-02,  1.2929e-02, -1.2314e-01,
        -6.0929e-03, -1.0489e-02,  5.9007e-05, -1.9242e-03,  2.2566e-03,
        -4.8826e-02, -6.3979e-02, -3.1632e-02, -3.3863e-02, -1.9053e-02,
        -2.4860e-02,  7.0055e-03, -1.6794e-02,  2.4902e-02, -3.2341e-02,
        -1.1296e-02, -6.3944e-03, -1.7739e-02, -1.8888e-03, -1.2746e-03,
        -3.7700e-02, -6.8811e-02, -1.1826e-02, -1.4522e-02, -1.7075e-02,
        -7.9848e-02, -1.8727e-02, -4.0870e-03, -1.4701e-02, -3.3506e-02,
        -4.7649e-02, -1.7077e-02, -2.1655e-02,  6.2020e-03,  1.3216e-03,
         1.6856e-02, -7.5665e-03, -4.2497e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0369, -0.0106,  0.0026,  0.0341,  0.0315,  0.0292,  0.0248,  0.1376,
         0.0157,  0.0143, -0.0238, -0.0254,  0.0058,  0.0393,  0.0978,  0.0140,
         0.0090,  0.0132, -0.0058,  0.0007,  0.0638,  0.0140,  0.0064,  0.0030,
        -0.0137, -0.0052,  0.0081,  0.0051,  0.0505,  0.0158,  0.0091,  0.0263,
         0.0211,  0.0373,  0.0440,  0.0209, -0.0406,  0.0428,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0006,  0.2887,  0.0823,  0.1119, -0.0443,  0.0599, -0.0389,  0.0122,
         0.0275,  0.0827, -0.0033,  0.0181,  0.0329,  0.0073, -0.0034,  0.0169,
        -0.0039,  0.0610, -0.0103, -0.0397, -0.0127, -0.0413,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1473e-02, -2.0663e-01, -4.2524e-02, -1.5840e-02,  2.3348e-02,
        -8.9931e-03, -1.5532e-02, -1.7008e-02, -3.1406e-02, -8.7143e-02,
        -1.0316e-02, -5.6729e-03, -1.0601e-02, -5.1620e-03, -1.7862e-02,
        -2.7719e-02, -7.3778e-02,  3.5480e-03, -1.3242e-02, -4.7023e-02,
         7.4794e-03,  1.2219e-02, -3.5721e-02, -1.7960e-02, -1.7361e-02,
        -5.2079e-03,  4.3445e-04,  1.1962e-04, -6.2717e-03, -4.9363e-02,
        -3.3086e-03, -4.8251e-03, -1.1976e-02,  3.1945e-03, -3.2053e-02,
        -8.9555e-02,  7.1307e-03, -1.1756e-02,  2.9442e-03, -6.3008e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.5445e-02, -8.6532e-03, -2.2193e-02, -2.2875e-02, -8.2111e-03,
        -3.1748e-02, -1.4190e-03, -1.8799e-02, -6.5901e-02, -1.2694e-02,
         6.0303e-03, -4.3865e-02, -1.2750e-03, -1.4510e-01, -1.0909e-01,
        -1.1846e-02, -2.8455e-03, -1.8547e-02,  1.5284e-03,  8.2860e-03,
        -1.6339e-02, -9.9636e-03,  1.0223e-04,  7.7078e-03, -1.7144e-02,
        -1.1487e-01, -1.5794e-03, -8.1309e-02, -8.6959e-02, -2.7751e-02,
         5.1512e-03, -2.4775e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0412, -0.3012, -0.0114,  0.0379, -0.0201, -0.0008, -0.0039, -0.0602,
        -0.0925, -0.0101,  0.0116, -0.0246,  0.0195, -0.0128,  0.0013,  0.0035,
        -0.0442,  0.0090,  0.0047, -0.0020, -0.0049, -0.0244, -0.0083, -0.0642,
         0.0046,  0.0237, -0.0402, -0.0028,  0.0139, -0.0331, -0.0004, -0.0007,
        -0.0054, -0.0194, -0.0413,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.1170, -0.0826, -0.0810, -0.0873, -0.0194, -0.0520, -0.1168, -0.1194,
         0.0232, -0.0470, -0.0144, -0.0331, -0.0069, -0.0394, -0.0135, -0.0090,
        -0.0322, -0.0062,  0.0231, -0.0272,  0.0494,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0614, -0.4272, -0.0013, -0.0491, -0.0449, -0.0488,  0.0031, -0.0612,
        -0.0938,  0.0081,  0.0160,  0.0191,  0.0133,  0.0298, -0.0292,  0.0121,
        -0.0212,  0.0033,  0.0571,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0835, -0.1243, -0.1283, -0.0208, -0.0310, -0.0020,  0.0069,  0.0149,
        -0.0193, -0.0927, -0.1616, -0.0474, -0.0225, -0.0479, -0.0061, -0.0298,
        -0.0193, -0.0538,  0.0135, -0.0745,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0533, -0.3181, -0.0227, -0.1011,  0.0389, -0.0092, -0.0171, -0.0576,
        -0.0206, -0.0134, -0.0290, -0.0136,  0.0194, -0.0587,  0.0248, -0.0320,
        -0.0757,  0.0020, -0.0285, -0.0143,  0.0080, -0.0227,  0.0194,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0523, -0.1723, -0.0424, -0.0469, -0.0168, -0.0659,  0.0009, -0.0386,
        -0.0892, -0.0844, -0.0430, -0.0559, -0.0461, -0.0525, -0.0068,  0.0156,
         0.0182, -0.0212, -0.0351,  0.0176, -0.0331,  0.0019, -0.0162, -0.0156,
        -0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0252, -0.3354, -0.0282, -0.0189, -0.0343, -0.0131,  0.0073, -0.0127,
        -0.0058,  0.0191, -0.0372, -0.0004,  0.0026,  0.0308,  0.0162,  0.0159,
        -0.0141,  0.0022,  0.0136, -0.0223, -0.0386, -0.0655, -0.0419, -0.0183,
        -0.0150, -0.0219, -0.0200,  0.0039, -0.0170, -0.0096,  0.0047,  0.0160,
        -0.0204, -0.0318,  0.0201,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0477, -0.0086,  0.0080, -0.0764, -0.0301,  0.0401, -0.0502, -0.0349,
        -0.1531, -0.0402, -0.0705, -0.0198,  0.0150, -0.0672, -0.0610, -0.1038,
        -0.0228, -0.0907,  0.0043,  0.0417, -0.0131,  0.0007,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0520,  0.0472, -0.0067,  0.0048,  0.0512,  0.0454,  0.0121,  0.0582,
        -0.0430,  0.1781, -0.0112,  0.0645, -0.0674,  0.0425,  0.0626,  0.0604,
        -0.0191,  0.0735, -0.0436, -0.0561, -0.0003,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0010, -0.4021, -0.0236, -0.0012, -0.0208,  0.0373, -0.0877, -0.1436,
         0.0081, -0.0231, -0.0679, -0.0387,  0.0006,  0.0068, -0.0268, -0.0424,
        -0.0087,  0.0492, -0.0103,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0147,  0.0200, -0.0018,  0.0125,  0.0065, -0.0313,  0.0073,  0.0002,
        -0.0005, -0.0057, -0.0063, -0.0075, -0.0003, -0.0264, -0.0027,  0.0097,
        -0.0197, -0.0396, -0.0162, -0.0237, -0.0045, -0.0270, -0.0281, -0.0082,
        -0.0337, -0.0024, -0.0131, -0.0551, -0.0350, -0.1940, -0.0246, -0.0768,
        -0.0295, -0.0469, -0.0083,  0.0061,  0.0066,  0.0106,  0.0273,  0.0035,
         0.0045,  0.0190,  0.0149,  0.0122,  0.0025, -0.0203, -0.0325],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.8437e-02, -2.6596e-01, -1.1434e-01, -7.9426e-03,  1.9768e-04,
         4.7615e-04, -6.6539e-04,  1.8289e-02, -2.9478e-02,  8.8622e-03,
         2.9402e-03,  7.3338e-03,  1.8978e-02, -1.7231e-02,  3.4490e-03,
         1.5248e-02, -6.3672e-02, -1.6807e-03,  1.3529e-02, -4.4812e-03,
         2.6246e-02, -2.3083e-02, -3.2804e-02, -8.3823e-02, -2.9987e-03,
         4.7056e-02, -1.7126e-02,  1.0147e-02,  3.6642e-03, -5.0074e-03,
         1.3242e-03,  1.9286e-02, -6.5587e-02,  2.8663e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0010, -0.3881, -0.0281,  0.0118,  0.0045, -0.0720, -0.0087,  0.0034,
        -0.0099, -0.0039,  0.0125,  0.0075, -0.0524, -0.0499, -0.0056,  0.0251,
        -0.0185, -0.0359, -0.0324,  0.0156,  0.0097,  0.0308, -0.0005,  0.0128,
         0.0067,  0.0207,  0.0217, -0.0029,  0.0090, -0.0160,  0.0065,  0.0433,
        -0.0192,  0.0135,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.1246, -0.2639, -0.0622, -0.0300, -0.0382, -0.0667, -0.0109,  0.0122,
        -0.0074, -0.0896, -0.0332,  0.0052, -0.0523, -0.1100, -0.0444, -0.0138,
        -0.0227, -0.0126,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0033,  0.0042,  0.0157, -0.0657,  0.0023, -0.0006,  0.0142, -0.0208,
         0.1657,  0.2711,  0.0083, -0.0252, -0.0013,  0.0664, -0.0005, -0.0023,
        -0.0101, -0.0052,  0.0278,  0.1473, -0.0166, -0.0170, -0.0024, -0.0213,
        -0.0107, -0.0305,  0.0432,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0040, -0.0804,  0.0142,  0.0048,  0.0164,  0.0114,  0.0325,  0.0211,
        -0.1218,  0.0120, -0.0066, -0.0225, -0.2184,  0.0158,  0.0169, -0.0132,
        -0.0078, -0.0235,  0.0111, -0.2495, -0.0118, -0.0201, -0.0063,  0.0243,
         0.0057,  0.0280,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0056, -0.2173,  0.0244, -0.0421, -0.1071, -0.0740, -0.0035, -0.0265,
        -0.0655, -0.0913, -0.0907, -0.0518, -0.0208,  0.0199, -0.0136, -0.0462,
        -0.0343, -0.0046,  0.0006, -0.0291,  0.0309,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.3944e-03, -3.2641e-01, -1.1648e-01, -1.7158e-01, -5.5581e-05,
        -7.5924e-03, -1.2149e-01, -1.9729e-02, -1.0894e-02,  4.8410e-02,
         7.0875e-04, -6.0424e-02,  4.0601e-02, -8.9632e-03,  6.4267e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1036,  0.0022, -0.0408,  0.0032,  0.0415,  0.0428,  0.0145,  0.0654,
        -0.0009,  0.0213,  0.0258, -0.0545,  0.4003,  0.0461,  0.0307,  0.0204,
         0.0344,  0.0515,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0299, -0.2694, -0.0868, -0.1422,  0.0034, -0.0365, -0.0012, -0.0170,
        -0.0270, -0.0723, -0.0609, -0.0462, -0.0131, -0.0679, -0.0041,  0.0345,
         0.0078, -0.0265,  0.0535,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0173, -0.1337, -0.0719, -0.0050, -0.0251, -0.0121,  0.0394,  0.0070,
        -0.0102, -0.0036,  0.0006, -0.0068, -0.0073,  0.0090,  0.0041,  0.0021,
        -0.0422,  0.0051,  0.0078,  0.0202, -0.0502,  0.0183, -0.0532, -0.0199,
        -0.0855, -0.0071, -0.0421,  0.0142,  0.0030, -0.0356,  0.0024, -0.0173,
        -0.0229, -0.0053, -0.0055, -0.0012, -0.0435,  0.0160, -0.0012,  0.0063,
         0.0061,  0.0182,  0.0178, -0.0769], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0580, -0.1692,  0.0053, -0.0490, -0.0072,  0.0043, -0.0229,  0.0064,
        -0.1150,  0.0035, -0.0036,  0.0009,  0.0005, -0.0033,  0.0295, -0.0065,
        -0.0402, -0.1165, -0.0141, -0.0312, -0.0362, -0.0109,  0.0080, -0.0069,
         0.0126, -0.0036, -0.0358, -0.0571,  0.0206, -0.0156, -0.0167, -0.0213,
        -0.0059, -0.0099,  0.0519,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0582, -0.3667, -0.0045, -0.0271, -0.0009, -0.0050, -0.1005, -0.0653,
         0.0047, -0.0084, -0.0064, -0.0259, -0.1051,  0.0184,  0.0096, -0.0509,
         0.0099, -0.0108,  0.0010,  0.0007,  0.0021,  0.0177, -0.0126,  0.0876,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.0117e-01, -2.1297e-01,  5.1954e-03, -5.3770e-02, -4.1316e-02,
         2.5147e-04,  1.2250e-02, -2.4696e-02, -7.3051e-02, -6.9446e-05,
        -1.4581e-02,  2.9509e-02, -9.2359e-03, -1.3124e-02, -7.4594e-02,
         1.4591e-02,  1.5375e-02, -7.2484e-03, -6.8812e-02,  8.8082e-03,
        -4.2463e-03, -4.7538e-02,  1.6809e-02, -6.6049e-04,  2.7830e-03,
        -3.7692e-02, -2.0165e-02, -8.0371e-03, -3.5359e-03, -5.5769e-03,
        -2.1631e-02,  6.0184e-03, -4.5723e-03,  4.0115e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0335,  0.0317, -0.0061,  0.0153,  0.0088,  0.0169, -0.0792, -0.0259,
         0.0005,  0.0058,  0.0159,  0.0107, -0.0795, -0.0062, -0.0840, -0.2434,
        -0.0096,  0.0026, -0.0214,  0.0035,  0.0067, -0.0284, -0.0013, -0.0674,
        -0.0579, -0.0255, -0.0707, -0.0237,  0.0003, -0.0176,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0496, -0.0195, -0.0584, -0.2303,  0.0083,  0.0539, -0.0679, -0.0197,
        -0.0531, -0.2570,  0.0660,  0.0541, -0.0623,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0093,  0.0281, -0.0220,  0.1019,  0.0212, -0.0561, -0.0202, -0.1431,
        -0.1296, -0.0137, -0.0090, -0.1596, -0.0044, -0.0577, -0.1164, -0.0310,
         0.0260, -0.0506,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0376, -0.2326, -0.3284,  0.0950, -0.1194,  0.0118, -0.0279,  0.1475,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0134, -0.1832, -0.0026, -0.0174, -0.0258, -0.0165, -0.0057, -0.0069,
         0.0095, -0.0008, -0.0067,  0.0229, -0.0061, -0.0066, -0.0015, -0.0012,
         0.0072,  0.0114,  0.0138, -0.0029,  0.0060, -0.0044, -0.0189, -0.0069,
        -0.0041, -0.0178, -0.0056, -0.0173, -0.0960, -0.0258,  0.0193, -0.0618,
        -0.0202, -0.1262, -0.0041, -0.0590, -0.0287, -0.0134,  0.0076,  0.0079,
        -0.0130, -0.0061, -0.0089,  0.0070,  0.0109, -0.0178,  0.0230],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0816, -0.0984,  0.1135, -0.2743,  0.0754, -0.0469,  0.0649,  0.0691,
         0.0371, -0.0238, -0.1149,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0203,  0.2582,  0.0401,  0.0687,  0.0275,  0.0199,  0.0498,  0.1287,
         0.0200,  0.0655,  0.0115,  0.0079,  0.0900, -0.0174, -0.0318,  0.0230,
         0.0345, -0.0853,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1905, -0.0530, -0.2529, -0.0302,  0.0475,  0.0139, -0.1644, -0.0318,
        -0.1100, -0.0127, -0.0027,  0.0082,  0.0012,  0.0402,  0.0410,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0818,  0.3010, -0.0166,  0.0592,  0.0142,  0.1316,  0.1052, -0.0082,
         0.0327,  0.1083,  0.0591, -0.0051, -0.0213, -0.0302, -0.0256,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0868,  0.0062, -0.0006,  0.0444,  0.0074, -0.0113, -0.0016, -0.0172,
        -0.0285, -0.0185, -0.0061, -0.0736, -0.1414, -0.0495, -0.0025, -0.0161,
        -0.0545, -0.0056, -0.0302, -0.0052, -0.0810, -0.1246,  0.0175,  0.0747,
        -0.0461, -0.0490,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0573, -0.0038, -0.0366,  0.0129, -0.0038, -0.0151,  0.0498, -0.2398,
        -0.3521, -0.0426,  0.0042, -0.0043, -0.0835,  0.0660, -0.0281,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0842,  0.0998,  0.0203,  0.0548,  0.0658,  0.0378,  0.0660,  0.0034,
         0.0122,  0.2176,  0.0205,  0.0051,  0.0505,  0.0251, -0.0040,  0.0150,
         0.0173,  0.0136,  0.0425, -0.0507,  0.0372, -0.0021, -0.0010, -0.0171,
        -0.0182,  0.0183,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0294, -0.0194, -0.0042, -0.0024, -0.0434, -0.0146, -0.0333, -0.0235,
         0.0023, -0.0418, -0.1238, -0.0317, -0.0271, -0.0329, -0.0918, -0.0675,
        -0.0139, -0.0347, -0.0066, -0.0872, -0.0061, -0.0394, -0.0159, -0.0716,
        -0.0051, -0.1303,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-5.4432e-02, -3.6061e-01,  5.1875e-02,  2.4310e-02,  1.1794e-02,
         3.1696e-02, -1.9128e-02, -4.0179e-02,  1.7130e-04, -3.6788e-02,
        -2.3513e-03, -1.0149e-02, -4.2916e-02,  1.0404e-02, -2.3238e-02,
        -5.9717e-03, -7.7315e-03, -3.9203e-03, -2.3520e-02, -1.4265e-02,
        -9.5240e-05, -2.5935e-02, -3.8852e-03, -2.5303e-02, -7.0312e-02,
        -9.8020e-04, -1.5100e-02, -1.0661e-02,  3.9606e-03,  5.9993e-03,
        -1.9818e-03, -1.7037e-02, -2.4813e-03,  2.6853e-03,  4.0921e-03,
         3.4043e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1049, -0.0729, -0.0396, -0.0138, -0.0321, -0.0228, -0.0340, -0.0643,
         0.0125, -0.0088, -0.0142, -0.0085, -0.0003,  0.0027, -0.0119, -0.0115,
         0.0075, -0.0035,  0.0096,  0.0085, -0.0095, -0.0449, -0.0238, -0.0074,
        -0.0518, -0.0851,  0.0044, -0.0528, -0.0622, -0.0083, -0.0042, -0.0607,
        -0.0506,  0.0040, -0.0409,  0.0054,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0963, -0.0792,  0.0515, -0.0541, -0.0694, -0.0856,  0.0219,  0.0117,
         0.0015,  0.0188, -0.0368, -0.0636, -0.2350, -0.0364, -0.0133, -0.0139,
        -0.0543, -0.0100, -0.0076, -0.0392,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1925, -0.0859, -0.0101, -0.0034, -0.0033, -0.0110, -0.0570, -0.0668,
         0.0011, -0.0079,  0.0145, -0.0011, -0.0075, -0.0076, -0.1043, -0.0552,
        -0.0201, -0.0426,  0.0030,  0.0012, -0.0033, -0.0021,  0.0006,  0.0124,
        -0.0450,  0.0070, -0.0064, -0.0012,  0.0098,  0.0016, -0.0254, -0.0045,
        -0.0086, -0.0651, -0.0115,  0.0184, -0.0682, -0.0128,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.4394e-02, -2.5301e-01, -8.4363e-02, -6.0945e-02, -1.1201e-02,
        -7.2685e-03,  1.4413e-04,  3.6119e-02, -2.3121e-02,  1.0008e-02,
        -3.4466e-02, -3.9181e-02, -1.3725e-02,  6.9997e-04, -1.6710e-03,
        -4.0721e-02,  2.6930e-03, -2.8973e-02,  1.2579e-03, -3.1936e-03,
        -1.3575e-03,  1.5508e-02,  1.3033e-02, -1.8664e-03,  1.6209e-03,
        -7.1060e-03, -1.3405e-02, -2.6306e-02, -6.9670e-02,  7.1356e-03,
        -1.6504e-02, -3.1828e-03,  1.4466e-02,  2.0093e-03, -2.5551e-02,
         8.1557e-03, -1.2388e-02, -2.3614e-04,  1.0217e-02, -5.6193e-02,
         2.6936e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0578, -0.3155,  0.0234, -0.0139, -0.0054, -0.0420, -0.0156, -0.0280,
        -0.0320,  0.0123, -0.0074, -0.0784, -0.0546, -0.0143, -0.0050, -0.0934,
        -0.0189, -0.1136,  0.0110, -0.0101, -0.0416,  0.0057,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0477,  0.0133,  0.0023, -0.0096,  0.0127, -0.0255, -0.0052,  0.0044,
        -0.0116, -0.0032, -0.0013, -0.0553, -0.0127, -0.0468, -0.0137, -0.0760,
         0.0074, -0.0336, -0.0027, -0.0919, -0.1734, -0.0117, -0.0218, -0.0216,
         0.0062,  0.0082,  0.0050, -0.0285, -0.0713, -0.0078, -0.0282, -0.0328,
        -0.0468,  0.0578, -0.0020,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.0873e-02, -1.2459e-01, -2.3201e-01, -6.4935e-02, -9.8534e-02,
        -4.8112e-03,  3.3511e-03,  2.8296e-03,  3.3955e-02, -8.0824e-03,
        -1.8009e-02,  3.0839e-04, -1.3221e-03,  1.0528e-02,  3.2463e-03,
        -1.9024e-02,  3.1766e-02, -1.6082e-02, -4.5295e-02,  6.0787e-03,
        -2.5174e-02,  1.0734e-02,  1.5791e-04, -1.5171e-02, -1.6958e-03,
        -1.2629e-02, -6.6138e-02,  1.7814e-02,  8.4854e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0250, -0.3342, -0.0746, -0.0738, -0.0229, -0.1029, -0.0094,  0.0098,
        -0.0217, -0.0048, -0.0152, -0.0474,  0.0043, -0.0331, -0.0066, -0.0450,
        -0.0130, -0.0216, -0.0247, -0.0005,  0.0060,  0.0021,  0.0239,  0.0040,
         0.0735,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0277,  0.0117, -0.0153, -0.0084, -0.1110, -0.0288, -0.0305, -0.0851,
        -0.0755, -0.0119,  0.0227,  0.0327,  0.0140,  0.1118, -0.0506, -0.0729,
        -0.0046, -0.0378, -0.0222, -0.0139, -0.0076, -0.0301, -0.1733,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0117, -0.0026,  0.0183, -0.0049,  0.0189,  0.0069,  0.0059,  0.0036,
        -0.0059, -0.0103, -0.0064, -0.0188, -0.0624, -0.1222, -0.0380, -0.0400,
        -0.0425, -0.0511, -0.0030, -0.0228,  0.0151,  0.0248,  0.0011, -0.0028,
        -0.0149,  0.0024, -0.1502,  0.0065, -0.0646, -0.0194, -0.0416, -0.0051,
        -0.0292, -0.0302, -0.0126, -0.0085, -0.0215,  0.0385,  0.0147,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0195, -0.0178,  0.0637, -0.1180, -0.2458, -0.0352,  0.0672, -0.0830,
        -0.1722,  0.0751,  0.0231,  0.0794,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.1252, -0.1118, -0.2633, -0.0079, -0.0052, -0.0200, -0.0522,  0.0198,
        -0.0186, -0.0170,  0.0046,  0.0160, -0.0111,  0.0038, -0.0096,  0.0038,
        -0.0097, -0.0250, -0.0023, -0.0331, -0.0156, -0.0209,  0.0053, -0.0039,
        -0.0020, -0.0033,  0.0134, -0.0156,  0.0196,  0.0178, -0.0056,  0.0100,
        -0.0133,  0.0017, -0.0116, -0.0046, -0.0145,  0.0017, -0.0006,  0.0006,
         0.0005,  0.0068,  0.0104,  0.0147, -0.0262,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0756,  0.4623,  0.0108,  0.0843, -0.0324, -0.0516, -0.0039,  0.0449,
         0.1255,  0.0417,  0.0346, -0.0325,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1495, -0.4515, -0.0376, -0.1567, -0.0253,  0.0318, -0.0570,  0.0903,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.9638e-02, -1.4989e-01, -1.4096e-02, -2.6150e-02, -6.1221e-02,
        -1.9143e-03, -5.6815e-03,  5.1086e-03, -3.7828e-02, -6.9908e-02,
        -2.2669e-02, -1.5302e-02, -6.8707e-03,  1.3558e-03,  5.0654e-03,
         3.7103e-03,  2.0015e-02, -7.6825e-04,  1.7489e-02, -9.9105e-03,
         4.7412e-03,  9.9954e-03, -1.5964e-02,  7.1134e-03, -5.2290e-03,
        -7.3360e-04,  3.7510e-03, -1.2002e-02, -3.3070e-02,  2.2891e-03,
        -8.3270e-03, -1.4275e-02, -7.4587e-04,  3.5052e-03, -5.2563e-03,
         6.0024e-03,  1.2556e-02, -3.2149e-03,  3.3721e-04,  4.3261e-03,
        -6.5295e-03,  6.3493e-04, -7.8389e-03, -1.0093e-05, -1.2254e-03,
         3.4321e-02, -6.7698e-02,  1.8067e-02, -5.1708e-02, -6.7363e-02,
        -1.6358e-02, -8.1012e-03, -1.9459e-02,  1.4630e-03, -2.1784e-02,
        -3.5153e-03, -2.3329e-02, -3.0217e-04,  1.3952e-03,  2.8029e-04,
         5.9297e-04], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0220, -0.0194,  0.0150, -0.0392, -0.0975,  0.0381, -0.0522,  0.0052,
         0.0103,  0.0195,  0.0219, -0.0811, -0.0074, -0.0553, -0.0749,  0.0042,
         0.0014, -0.0031, -0.0674, -0.0260, -0.0547, -0.0672, -0.0169, -0.0081,
        -0.0430, -0.0107, -0.0006, -0.0127, -0.0227, -0.0204, -0.0325, -0.0201,
        -0.0292,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1043,  0.0849,  0.1665,  0.0522,  0.0297,  0.0133,  0.0591, -0.0140,
        -0.0093, -0.0208, -0.0238,  0.0020,  0.0108,  0.0012, -0.0011,  0.0341,
        -0.0012,  0.0365,  0.2033, -0.0038, -0.0187, -0.0249, -0.0645,  0.0199,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0329, -0.1120, -0.0535, -0.0912, -0.0229, -0.1080, -0.0248, -0.0426,
         0.0118, -0.0620, -0.0535, -0.0518, -0.1645, -0.0105,  0.0128, -0.0069,
         0.0142, -0.0132,  0.0229,  0.0152, -0.0226, -0.0110,  0.0201, -0.0189,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0696,  0.0164, -0.0598, -0.0611,  0.0152,  0.0157, -0.0080, -0.0418,
        -0.0011,  0.0047,  0.0166, -0.0243, -0.0081, -0.0060, -0.0506, -0.1227,
        -0.1011,  0.0095,  0.0084, -0.0054, -0.0399,  0.0200, -0.0072, -0.0876,
        -0.0239,  0.0091, -0.0469, -0.0107,  0.0050, -0.0006, -0.0332,  0.0103,
         0.0314,  0.0280,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0335, -0.0311, -0.0338,  0.0203, -0.0770,  0.0537,  0.0010,  0.0574,
         0.0168,  0.0165, -0.0157,  0.0122, -0.0070, -0.1397, -0.0183, -0.0372,
        -0.1208, -0.0009, -0.0612, -0.0058, -0.0420, -0.1636,  0.0217,  0.0064,
        -0.0064,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0238, -0.0471, -0.0528, -0.0074, -0.0214, -0.0416, -0.1197, -0.0639,
        -0.0492, -0.3887, -0.0433, -0.0031, -0.0271, -0.0244,  0.0098, -0.0767,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0827, -0.2111, -0.0147, -0.0887, -0.0270, -0.0216, -0.1683,  0.0150,
         0.0369,  0.0084, -0.0055, -0.1118, -0.0476, -0.0059,  0.0161, -0.0119,
        -0.0130, -0.0131, -0.0096, -0.0020,  0.0004,  0.0817, -0.0070,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0294,  0.0395,  0.0589,  0.2148,  0.3152,  0.1078, -0.0379,  0.0351,
         0.0292,  0.0146, -0.0285, -0.0147, -0.0132,  0.0611,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0348, -0.1168, -0.0660, -0.1648, -0.0244, -0.0050, -0.0389, -0.0029,
        -0.0394, -0.0257, -0.0933, -0.0787, -0.0669, -0.0055, -0.0005, -0.0024,
         0.0309, -0.0699, -0.0221, -0.0655, -0.0104, -0.0068,  0.0183,  0.0101,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1668, -0.0579, -0.1418,  0.0022, -0.1534, -0.0211, -0.1949, -0.0481,
        -0.0929, -0.0783, -0.0426,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0400, -0.0128,  0.0142,  0.0005,  0.0155, -0.1447, -0.2412, -0.0234,
        -0.1330, -0.0638, -0.0168,  0.0064, -0.0377,  0.0101, -0.0662, -0.0027,
        -0.0292, -0.0493,  0.0512, -0.0106, -0.0306,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0196, -0.0107, -0.0081, -0.0205, -0.1342, -0.0508, -0.0679, -0.2326,
        -0.0345, -0.2902, -0.0524, -0.0165,  0.0260, -0.0360,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1635,  0.0160, -0.0121, -0.0048, -0.0124,  0.0096,  0.0103, -0.0126,
        -0.0200, -0.0165,  0.0485,  0.1320,  0.2251,  0.0130, -0.0056,  0.0209,
        -0.0899,  0.0712,  0.0031, -0.0177,  0.0950,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1024,  0.3736, -0.0107,  0.0753, -0.0702,  0.0836,  0.0093,  0.1516,
         0.0020,  0.0447,  0.0073, -0.0257, -0.0436,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.6482e-02, -1.0077e-01, -1.1964e-02, -1.9339e-03, -1.2877e-04,
        -2.8239e-04, -2.8146e-02, -3.8924e-02, -4.2997e-03, -7.8073e-03,
        -3.5134e-05, -1.0137e-02,  7.8197e-03, -7.8223e-03, -7.4529e-03,
        -1.2513e-02,  6.8811e-03,  5.9203e-03, -5.6504e-03,  2.1095e-03,
        -4.8023e-04,  2.2388e-02,  3.2383e-03, -3.3571e-02, -8.3354e-02,
        -8.2066e-03, -1.1181e-01, -1.0710e-01, -6.6373e-02, -1.1723e-02,
        -1.1687e-03, -6.3625e-02, -4.5594e-03,  3.6311e-04, -1.2821e-02,
        -2.2518e-02, -3.8268e-02, -1.1621e-02,  6.6256e-03, -1.1247e-02,
        -3.0091e-02, -1.1311e-02,  1.7184e-03, -2.4620e-02,  4.1146e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0014, -0.0294, -0.0865, -0.0131, -0.0039,  0.0229,  0.0019, -0.0120,
        -0.0043, -0.0140, -0.0670, -0.0018,  0.0062, -0.0067, -0.0363, -0.0065,
        -0.1564, -0.0180, -0.0152,  0.0129, -0.0349, -0.0904,  0.0164, -0.0784,
        -0.0012,  0.0059, -0.0247, -0.0425, -0.0800, -0.0503, -0.0008, -0.0399,
        -0.0006,  0.0175,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.6328e-02,  7.4993e-02, -2.7847e-02, -3.5128e-03, -7.7456e-03,
        -4.7774e-06, -8.8177e-03, -1.1020e-02, -2.6837e-02,  2.2848e-03,
        -1.5253e-02,  1.8020e-02, -5.2113e-03,  5.7600e-03, -1.8283e-02,
        -1.2781e-02, -1.6676e-02,  1.5122e-02,  7.4923e-03,  1.9966e-01,
         2.4239e-01,  5.4883e-03,  2.3136e-02,  5.5415e-02, -5.3556e-02,
         6.6366e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0646, -0.4474, -0.0448, -0.0611,  0.0223, -0.0098,  0.0089, -0.0234,
        -0.0320,  0.0043,  0.0026,  0.0029,  0.0018, -0.0141, -0.0079, -0.0470,
        -0.0851,  0.0063, -0.0345, -0.0062, -0.0199,  0.0094,  0.0436,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0812, -0.0670, -0.0024, -0.0340, -0.0127,  0.0210, -0.0205, -0.0388,
        -0.0697, -0.0122, -0.0031, -0.0250, -0.0002, -0.0148, -0.1708,  0.0083,
         0.0049, -0.0317, -0.0350, -0.0230,  0.0067,  0.0030,  0.0010,  0.0059,
         0.0088, -0.0048, -0.0535, -0.0682, -0.0185, -0.0307, -0.0117,  0.0199,
         0.0010, -0.0047, -0.0172,  0.0103,  0.0007,  0.0399, -0.0014,  0.0157,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0265, -0.0493, -0.0239, -0.0148, -0.0991, -0.0162, -0.0287, -0.0917,
         0.0083, -0.0009,  0.0025, -0.0137, -0.0405, -0.0109, -0.0377, -0.0379,
        -0.0562,  0.0114,  0.0066, -0.0214, -0.0076, -0.0836, -0.0485, -0.0151,
        -0.0048, -0.0098, -0.0081,  0.0029,  0.0016, -0.0335, -0.0967,  0.0220,
        -0.0554, -0.0119,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0186,  0.0144,  0.0122,  0.0016,  0.0862,  0.0050, -0.0040, -0.0096,
         0.0007, -0.0189,  0.0020,  0.0144, -0.0545,  0.0237,  0.1825, -0.0106,
        -0.0260,  0.0088,  0.0773,  0.0010, -0.0022,  0.0017,  0.0050, -0.0159,
         0.0419,  0.0838,  0.0010,  0.0121,  0.0109,  0.0165, -0.0007, -0.0008,
        -0.0226,  0.0023,  0.0315, -0.0027,  0.0157,  0.0230,  0.0007,  0.0081,
         0.0121,  0.0080,  0.0338,  0.0124, -0.0041, -0.0003, -0.0125, -0.0060,
         0.0087,  0.0122,  0.0023, -0.0165], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0471,  0.0064,  0.0220,  0.0343,  0.0758,  0.0111,  0.0137,  0.0060,
        -0.0029, -0.0026, -0.0037, -0.0143, -0.0093,  0.0826,  0.0403,  0.0687,
         0.1199,  0.0906,  0.0445,  0.0077,  0.0301,  0.0060,  0.0483,  0.1130,
        -0.0048, -0.0023,  0.0162, -0.0263, -0.0125,  0.0053, -0.0317,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.4242e-02,  1.3012e-01, -1.4752e-02,  9.1415e-03, -8.4176e-03,
        -1.8056e-03,  1.1380e-02,  1.3470e-04,  3.1162e-03, -2.5882e-02,
         1.4006e-02,  4.1922e-02, -4.2931e-02, -6.4528e-04,  5.0059e-02,
         1.0561e-01,  2.0369e-02,  5.8371e-02,  1.7804e-01,  9.4447e-02,
         4.8415e-02,  3.2886e-02,  2.7418e-02, -5.4101e-02, -1.7955e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2244,  0.0644, -0.0143,  0.0070,  0.0228,  0.2134,  0.2417, -0.0073,
        -0.0768, -0.0067,  0.0400,  0.0812,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0249, -0.3270, -0.0202, -0.0367, -0.0274, -0.0331,  0.0306,  0.0010,
        -0.0157, -0.0240,  0.0093,  0.0127, -0.0521, -0.0815, -0.0148, -0.0523,
        -0.0171,  0.0164,  0.0049, -0.0044, -0.0171,  0.0036, -0.0099,  0.0147,
         0.0112, -0.1086, -0.0015,  0.0179,  0.0096,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0548, -0.3309, -0.0274,  0.0314, -0.0664, -0.0945, -0.2573,  0.0161,
        -0.0139, -0.0304, -0.0118, -0.0650,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0218, -0.3469, -0.0920, -0.1406,  0.0111, -0.1280, -0.1697, -0.0059,
        -0.0482, -0.0293, -0.0011, -0.0053,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0513,  0.0588, -0.1201,  0.0247, -0.0964,  0.0099,  0.0136, -0.0303,
        -0.2090, -0.0147, -0.0314, -0.0991, -0.0173, -0.0288, -0.0135, -0.0476,
         0.0086, -0.0102, -0.0479,  0.0669,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0221, -0.2904, -0.0098, -0.0366, -0.0653, -0.0580, -0.0137,  0.0211,
        -0.0013,  0.0082, -0.0055, -0.0437, -0.0246, -0.0248,  0.0069, -0.0166,
        -0.0537, -0.0099,  0.0061, -0.0036, -0.0066,  0.0225, -0.0076, -0.0070,
        -0.0161,  0.0221, -0.0110, -0.0397, -0.0310,  0.0046,  0.0179, -0.0036,
        -0.0230, -0.0250, -0.0132, -0.0150,  0.0121,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1862,  0.6432, -0.0365, -0.0201,  0.0579, -0.0204, -0.0092, -0.0061,
         0.0205,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0655,  0.0615,  0.0453, -0.1076, -0.0202,  0.0050, -0.0568, -0.0558,
        -0.0143,  0.0049, -0.0179, -0.0175, -0.0620, -0.1228,  0.0119, -0.0213,
         0.0238,  0.0333, -0.0686, -0.0376, -0.0275,  0.0177, -0.0274,  0.0107,
        -0.0028,  0.0195, -0.0409,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0943,  0.1822, -0.0078,  0.0622,  0.0009,  0.0221,  0.0046,  0.0046,
        -0.0009,  0.0182,  0.0005,  0.0607,  0.0323,  0.0090, -0.0079, -0.0151,
         0.0675,  0.0004, -0.0169,  0.0173, -0.0043,  0.0619,  0.0462, -0.0268,
         0.0100, -0.0106, -0.0040,  0.0133,  0.0849,  0.0111,  0.0091, -0.0009,
         0.0046, -0.0869,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0438, -0.0045, -0.0066, -0.0072, -0.0194, -0.0096,  0.0109, -0.0077,
         0.0031,  0.0097, -0.0135, -0.0985,  0.0113, -0.0092, -0.0076, -0.0657,
        -0.1483, -0.0384, -0.0330,  0.0119,  0.0047, -0.0004, -0.0323, -0.0085,
        -0.0021, -0.0067, -0.0015, -0.0076,  0.0044, -0.0072, -0.0225, -0.0400,
        -0.0767, -0.0156, -0.0325, -0.0218, -0.0340, -0.0220,  0.0055, -0.0142,
        -0.0107, -0.0271,  0.0271,  0.0065,  0.0086,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0738,  0.0190, -0.0094,  0.0153, -0.0936, -0.1485, -0.2561, -0.0089,
         0.0115, -0.1388, -0.0332, -0.0025, -0.0846,  0.0371, -0.0297, -0.0380,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0239, -0.0096, -0.0041, -0.0395, -0.0185, -0.0378, -0.0026, -0.0455,
        -0.2248, -0.1603, -0.0412, -0.0987, -0.0150, -0.0497, -0.0633, -0.0014,
        -0.0272, -0.0062,  0.0119, -0.0299,  0.0050,  0.0040, -0.0033, -0.0254,
        -0.0276,  0.0075, -0.0160,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.6647e-02,  8.9971e-03, -7.4215e-02, -7.2936e-03, -1.5294e-02,
        -1.3098e-03, -1.5802e-02, -4.6626e-03, -1.9397e-02, -7.7335e-03,
         2.9968e-03, -1.6276e-02, -1.8141e-02, -8.4838e-03, -1.4600e-02,
        -3.5842e-03, -4.1258e-02,  8.8575e-05, -1.2438e-02, -1.9360e-02,
         3.6566e-02, -2.3756e-02, -5.6677e-03, -6.3399e-02, -6.6844e-03,
         6.6508e-03, -4.0479e-02, -9.0203e-02, -1.5474e-04, -8.7908e-02,
         1.2849e-02, -3.2718e-02, -2.7734e-02,  5.3769e-03, -8.1652e-03,
        -2.2724e-02, -1.1166e-02, -4.8230e-02, -8.6826e-03,  2.9947e-03,
        -1.3350e-02, -3.2830e-02, -2.0873e-02,  1.0183e-03,  1.3995e-03,
         1.9843e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0096,  0.0240, -0.0154, -0.0223, -0.0201, -0.0387, -0.0134, -0.0802,
        -0.0452, -0.0139, -0.0229, -0.0089, -0.0125, -0.0064,  0.0049, -0.0184,
         0.0118, -0.0305, -0.1067, -0.1505, -0.0143, -0.0791, -0.0148, -0.0303,
         0.0059, -0.0044, -0.0570,  0.1043, -0.0239, -0.0098,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0087, -0.0210,  0.0397,  0.0500,  0.0211,  0.0752,  0.2059,  0.4320,
         0.0521,  0.0662, -0.0281,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1238, -0.0330,  0.0063,  0.0164, -0.0206, -0.0042, -0.0276, -0.0090,
         0.0198,  0.0372,  0.0481,  0.0474,  0.0559, -0.0129, -0.1764, -0.0752,
        -0.0573, -0.0276,  0.0178, -0.1834,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0268,  0.5093,  0.1001,  0.0294, -0.0520,  0.0609,  0.0384, -0.0256,
        -0.0137,  0.0756,  0.0682,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0481,  0.2633,  0.0548,  0.1127, -0.0995, -0.0586,  0.0102,  0.0411,
         0.0815,  0.0323,  0.0164,  0.0426, -0.0328,  0.0058,  0.0101,  0.0235,
         0.0024, -0.0078,  0.0054, -0.0464,  0.0045,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0334,  0.1948,  0.1333, -0.0037,  0.0013,  0.3273, -0.0568, -0.0143,
         0.0222,  0.0405,  0.0079,  0.0217,  0.0065,  0.0039,  0.0038, -0.0036,
        -0.0634,  0.0613,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0020, -0.0455, -0.1782,  0.0366,  0.0288,  0.0293,  0.0045,  0.0082,
        -0.0048, -0.0066,  0.0004,  0.0189, -0.0253, -0.0714, -0.1193,  0.0045,
        -0.0126, -0.0586, -0.0128,  0.0153, -0.0087, -0.0280, -0.0710, -0.0051,
        -0.0227,  0.0191, -0.0304, -0.0132, -0.0222, -0.0144,  0.0082, -0.0078,
        -0.0019,  0.0097, -0.0137, -0.0023,  0.0074,  0.0102, -0.0204,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.7120e-02,  3.0986e-01, -7.0862e-03,  8.5719e-02,  5.7833e-04,
         1.0254e-02, -1.2342e-02,  2.4311e-02, -4.4734e-03,  6.9669e-03,
         1.1843e-02,  6.1409e-02, -1.6178e-02, -1.1607e-02, -3.4934e-02,
        -4.3162e-03,  8.3066e-02,  4.9368e-03,  2.0812e-03,  9.2406e-04,
         4.5004e-03, -1.3954e-02, -9.5119e-03, -4.4467e-03, -7.4721e-03,
         8.9059e-03,  1.8447e-02, -1.2095e-04,  5.2087e-02,  1.0373e-01,
         9.7156e-04,  9.4047e-03,  4.2027e-03,  2.2239e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.1467, -0.3057, -0.1811, -0.1890,  0.0347, -0.0235,  0.0401, -0.0290,
        -0.0502,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.2855e-02, -1.7550e-01,  1.6739e-02, -5.2455e-02, -1.1464e-02,
        -7.7184e-02,  2.8284e-03, -1.9178e-02,  1.3016e-02,  8.9360e-03,
        -3.4639e-02, -9.6791e-03,  9.1552e-03, -1.1628e-02, -2.3197e-03,
        -7.0361e-02, -4.9452e-03, -1.0176e-01,  7.0848e-03,  2.3511e-02,
        -3.4662e-03,  1.4300e-02,  2.6896e-03,  6.4811e-03, -5.3383e-02,
         6.9881e-03, -1.5792e-02,  3.3024e-05,  1.6624e-02,  8.7902e-04,
        -1.3525e-02, -5.5494e-02, -3.6863e-03, -3.6194e-02, -1.2996e-02,
         7.3974e-03,  5.1933e-03, -2.7065e-02,  2.6829e-02, -1.5747e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.0477e-02, -8.6438e-02, -1.3946e-01, -8.9145e-03, -4.8695e-02,
        -5.7406e-03,  4.8976e-03, -1.8821e-02, -3.5015e-02, -5.8166e-02,
         1.5832e-03, -1.6316e-03,  1.8960e-03,  3.3666e-03,  3.5304e-03,
         1.7124e-02,  8.3934e-03,  4.3673e-02, -3.0186e-02,  2.9452e-03,
        -1.8067e-02,  1.6001e-03, -3.3254e-03, -5.7290e-03,  1.1720e-02,
        -1.3543e-02, -2.9902e-02,  1.5640e-02, -2.8720e-02, -7.2639e-02,
         1.9459e-03, -1.3238e-02,  6.1005e-03, -2.7458e-03,  7.3704e-03,
         7.4943e-03, -2.5459e-02, -6.0785e-02,  5.9996e-03, -2.3067e-05,
         2.1148e-02,  1.2424e-02, -8.7305e-04, -1.4218e-04, -3.2198e-03,
        -2.1177e-03, -2.6472e-03,  1.0021e-02, -2.2274e-04,  1.2235e-02,
        -1.3278e-02,  8.6715e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0234, -0.3724, -0.0326, -0.0177, -0.0189, -0.0301, -0.0565, -0.0117,
        -0.0248,  0.0010,  0.0071, -0.0139, -0.0777,  0.0062, -0.0128, -0.0937,
        -0.0604,  0.0059, -0.0458, -0.0135,  0.0076,  0.0663,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0861,  0.4729, -0.0024,  0.0114,  0.0110,  0.0166, -0.0090,  0.0592,
         0.0274,  0.0523,  0.0179,  0.0261,  0.0091,  0.0316, -0.0046,  0.0758,
         0.0113, -0.0221, -0.0216,  0.0224,  0.0094,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.9654e-03, -2.2147e-01, -1.0350e-02, -2.4878e-02,  1.6820e-02,
        -3.7345e-03, -2.1549e-03,  3.5129e-02, -4.6266e-03,  1.5710e-02,
         1.8453e-03, -1.2771e-03, -1.4874e-03, -1.9945e-02,  4.2551e-03,
        -7.7732e-04,  1.0783e-02,  1.5116e-02, -1.4115e-02,  4.6188e-03,
         8.5163e-03, -8.8529e-03, -7.1396e-03,  8.9506e-04,  6.3042e-03,
         2.0007e-02, -1.3074e-02, -2.9568e-02,  6.1496e-03, -2.9875e-02,
        -8.5032e-02, -7.9690e-02, -3.0996e-05, -1.1964e-02, -2.3421e-02,
         7.5229e-04,  6.4030e-03,  4.6106e-03,  1.2450e-04,  2.0441e-03,
        -4.6783e-02,  3.6839e-03, -4.5702e-03,  1.0398e-02, -2.0608e-02,
         1.0521e-02, -2.2431e-02, -5.6883e-03,  5.7807e-03,  7.1092e-03,
        -1.2333e-02,  1.0386e-02,  1.6916e-02,  2.8425e-02, -3.7855e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.8839e-03, -1.9009e-03, -5.0955e-03, -2.4819e-02, -2.8408e-02,
        -4.4879e-02, -9.7718e-02, -9.5069e-03,  1.4185e-02, -5.5455e-03,
        -1.4633e-02,  1.4095e-03, -9.4545e-03, -2.0215e-02, -3.0897e-02,
         8.5934e-05,  2.0431e-04, -1.6131e-02, -3.0275e-03, -6.1680e-03,
        -8.1948e-03, -6.6113e-03, -5.3439e-02,  1.0861e-02,  9.6498e-03,
         5.9720e-03, -1.5095e-02, -8.7493e-02, -1.2608e-02, -5.6552e-03,
         1.5282e-02,  1.0139e-02, -2.1956e-02, -1.4175e-01, -1.2707e-03,
        -4.6031e-02,  5.3568e-03, -4.0288e-02, -4.6531e-02, -3.1140e-02,
        -1.1721e-02, -9.5952e-03,  6.5358e-03, -1.2283e-02, -4.5377e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0100, -0.1972, -0.0792, -0.0988,  0.0141, -0.0101, -0.0129, -0.0090,
        -0.0608,  0.0112,  0.0081, -0.0475, -0.0225, -0.0658, -0.0622,  0.0387,
        -0.0357, -0.0315, -0.0079, -0.0079, -0.0794,  0.0139, -0.0361,  0.0397,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0722, -0.0061,  0.0013,  0.0383,  0.0183,  0.0044, -0.0487, -0.0305,
         0.0221,  0.0012,  0.0077, -0.0032, -0.0168, -0.0462, -0.0076, -0.0133,
        -0.0668, -0.0443, -0.0514, -0.0692, -0.0096, -0.0052, -0.0498, -0.0638,
        -0.0079, -0.0651, -0.0437, -0.0284, -0.0337, -0.0559, -0.0112,  0.0115,
        -0.0445,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0891, -0.3763, -0.0529, -0.0920, -0.0094, -0.0509, -0.0359,  0.0149,
        -0.0958, -0.0075,  0.0090, -0.0160,  0.0444, -0.0194,  0.0136, -0.0073,
        -0.0307, -0.0308, -0.0040,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0101, -0.0205,  0.0566, -0.0245, -0.0683, -0.0118, -0.0818, -0.0639,
        -0.1281,  0.0203,  0.0261, -0.0232,  0.0268, -0.0409,  0.0011, -0.0065,
        -0.0264, -0.0479, -0.0256, -0.0218, -0.0441, -0.1017, -0.0233, -0.0182,
        -0.0806,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0427, -0.2535, -0.0942, -0.0386, -0.0420, -0.0945, -0.1881,  0.0310,
        -0.0204, -0.0220, -0.0089,  0.0096, -0.0401, -0.0234,  0.0139,  0.0211,
         0.0562,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.0031, -0.0091, -0.0284, -0.0012, -0.0203, -0.0124, -0.0050, -0.0068,
         0.0131, -0.0632, -0.0309, -0.0386,  0.0114, -0.0011,  0.0223, -0.0292,
        -0.0694, -0.1412, -0.0137, -0.0558, -0.0007, -0.1641, -0.1448, -0.0110,
         0.0046, -0.0248,  0.0119, -0.0040,  0.0197,  0.0125,  0.0231, -0.0023,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0373, -0.1577, -0.1278, -0.0471, -0.0123,  0.0027, -0.0827, -0.1243,
         0.0014, -0.0079, -0.0005, -0.0095, -0.0215, -0.0185, -0.0081,  0.0322,
        -0.0015, -0.0081, -0.0777, -0.0205, -0.0331,  0.0009, -0.0408, -0.0841,
        -0.0110, -0.0119,  0.0081,  0.0110,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0711,  0.1410,  0.0122, -0.0374,  0.0378,  0.0526,  0.1201,  0.0419,
         0.0028,  0.0539,  0.0107,  0.0487, -0.0111,  0.0276, -0.0070,  0.0610,
         0.0139,  0.0793,  0.0131, -0.0151,  0.0052,  0.0015,  0.0138, -0.0076,
         0.0381,  0.0100,  0.0125, -0.0086, -0.0443,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0125, -0.0453, -0.0337, -0.0453, -0.1597, -0.0273, -0.0330, -0.0055,
        -0.0142,  0.0077, -0.0002, -0.0275, -0.0376, -0.1306,  0.0556, -0.0683,
         0.0003, -0.0920, -0.0244, -0.0176,  0.0027,  0.0146, -0.0383, -0.0205,
        -0.0149, -0.0069, -0.0132,  0.0013, -0.0019, -0.0082,  0.0041,  0.0223,
        -0.0129,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0188, -0.3066, -0.0018, -0.0254,  0.0161, -0.0506,  0.0519, -0.0115,
        -0.0374, -0.0489, -0.0568, -0.0128, -0.0061, -0.0028, -0.0081,  0.0093,
        -0.0330, -0.0020, -0.0799,  0.0237, -0.0442, -0.0430, -0.0341,  0.0752,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0962, -0.1720, -0.1249, -0.0152, -0.0412, -0.0509, -0.0257, -0.0505,
         0.0331, -0.0521, -0.0242, -0.0204, -0.0225, -0.0096, -0.0009, -0.0004,
         0.0028, -0.0266, -0.0106, -0.1380, -0.0347, -0.0079,  0.0397,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.6756e-02, -2.6705e-01, -2.1602e-02,  7.9435e-03, -2.4522e-02,
        -5.9547e-02,  2.0847e-02,  1.5217e-02, -1.0278e-02, -2.4940e-02,
         9.6017e-03, -9.7067e-04, -8.5888e-03,  2.8310e-02, -4.4108e-04,
         1.7606e-02, -3.7078e-02, -1.8253e-02, -6.0468e-03,  2.8046e-02,
         4.8046e-03, -2.4474e-04,  4.1277e-04, -9.5265e-03, -2.6018e-02,
        -4.5093e-03, -1.5853e-03,  1.7593e-03, -8.8481e-03, -7.0842e-02,
        -7.7023e-03, -1.9052e-02, -2.5170e-02, -4.4954e-03, -8.0334e-03,
        -4.1372e-02, -3.4643e-02,  1.8703e-02,  3.8633e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.6837e-02, -8.0342e-02,  2.0812e-02, -1.9495e-02, -1.3050e-01,
        -9.8583e-03,  9.5291e-03,  6.7998e-03,  1.2133e-02,  1.6486e-02,
         1.0165e-04, -5.8140e-03,  3.8558e-03, -8.9066e-03, -5.6275e-03,
        -7.6714e-02, -4.0476e-02,  6.4688e-03, -2.8967e-03, -3.7273e-02,
        -7.4325e-02, -1.1532e-02, -2.8342e-02, -2.8772e-02, -5.4523e-02,
        -2.9355e-02, -1.1203e-02, -9.1522e-04, -1.3837e-02, -2.2720e-02,
        -3.2592e-02, -4.8713e-03, -9.1162e-03, -1.2025e-02,  2.6504e-03,
         3.6667e-03, -1.2932e-02, -6.0903e-02, -4.4928e-03, -1.1124e-02,
        -1.5017e-02,  9.7395e-03,  1.4421e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0424, -0.2018,  0.0123,  0.0135, -0.0115,  0.0212,  0.0095, -0.0034,
        -0.0502,  0.0073,  0.0375, -0.0027, -0.0193, -0.0022,  0.0039,  0.0092,
        -0.0341, -0.0609, -0.0086, -0.0059,  0.0045, -0.0073, -0.0118,  0.0018,
         0.0038,  0.0028,  0.0087,  0.0023,  0.0035,  0.0091,  0.0313, -0.0259,
        -0.0664,  0.0214, -0.0108, -0.0010, -0.0070, -0.0016, -0.0260, -0.0119,
        -0.0256, -0.0465, -0.0039,  0.0134,  0.0042, -0.0112, -0.0007, -0.0008,
        -0.0028,  0.0024,  0.0084, -0.0025, -0.0062,  0.0043, -0.0036,  0.0285,
        -0.0185], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0270,  0.4753,  0.0628, -0.0186,  0.0535, -0.0017,  0.0444,  0.1185,
         0.0056,  0.0129,  0.0653, -0.0460, -0.0535, -0.0150,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0502, -0.1759,  0.0581,  0.0215, -0.0077, -0.0568, -0.0257, -0.0038,
        -0.0221, -0.0025, -0.0582, -0.1274,  0.0339, -0.0179, -0.0263,  0.0036,
        -0.0081, -0.0118,  0.0240, -0.0519, -0.0986, -0.0124, -0.0286,  0.0009,
         0.0269,  0.0144,  0.0027, -0.0102, -0.0052, -0.0128,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0953,  0.0926,  0.0367,  0.4724,  0.1061,  0.0821,  0.0682, -0.0465,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0726, -0.1488,  0.0147, -0.0189, -0.0532, -0.0887, -0.0141,  0.0229,
        -0.0089, -0.0137,  0.0122,  0.0072, -0.0396, -0.0819, -0.0085,  0.0006,
         0.0145, -0.0047, -0.0020,  0.0099, -0.0053,  0.0039,  0.0113,  0.0123,
        -0.0348, -0.0290, -0.0477, -0.0132, -0.0805, -0.0860, -0.0108, -0.0035,
        -0.0242,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0473, -0.2067, -0.0825, -0.0519, -0.0026, -0.0017, -0.0208, -0.0095,
        -0.0056,  0.0017, -0.0073,  0.0034, -0.0007, -0.0045, -0.0026, -0.0132,
         0.0039, -0.0020, -0.0028,  0.0054,  0.0129, -0.0040,  0.0018,  0.0072,
         0.0063, -0.0091, -0.0140, -0.0183, -0.0009,  0.0055, -0.0043, -0.0074,
        -0.0276, -0.0300, -0.0023, -0.0119, -0.0171, -0.0061, -0.0032,  0.0028,
         0.0047, -0.0064, -0.0205, -0.0123, -0.0020, -0.0079, -0.0061,  0.0163,
        -0.0074,  0.0039, -0.0220, -0.0042,  0.0011, -0.0073, -0.0006, -0.0007,
        -0.0070, -0.0158, -0.0036, -0.0220, -0.0064, -0.0022, -0.0218,  0.0082,
        -0.0023,  0.0003, -0.0058, -0.0142,  0.0035, -0.0517, -0.0423],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0853, -0.0100, -0.0270, -0.0029, -0.0481, -0.0119, -0.1202, -0.0141,
         0.0080,  0.0097, -0.0155, -0.0019,  0.0019, -0.0152, -0.0159, -0.0069,
        -0.0148, -0.0074, -0.0148, -0.0783, -0.0306, -0.0034,  0.0264, -0.0411,
         0.0147, -0.0346, -0.0214, -0.0071, -0.0121, -0.0179, -0.0123, -0.0232,
        -0.0434, -0.0446, -0.0184, -0.0191, -0.0530, -0.0051, -0.0027, -0.0160,
        -0.0206,  0.0226,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0575, -0.1693,  0.0188,  0.0074, -0.0042,  0.0160,  0.0207,  0.0048,
        -0.0296, -0.0563, -0.0313, -0.0335, -0.0630,  0.0009,  0.0021, -0.0067,
         0.0104, -0.0507, -0.0531, -0.0049, -0.0024, -0.0152, -0.0022, -0.0169,
         0.0050, -0.0951, -0.0487, -0.0165, -0.0543, -0.0025,  0.0125, -0.0095,
        -0.0063, -0.0088, -0.0124,  0.0023, -0.0480,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1299, -0.4655,  0.0534, -0.0409,  0.0099,  0.0717, -0.0203,  0.0171,
        -0.0316, -0.0176, -0.0153,  0.0067, -0.0510, -0.0230,  0.0463,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0382, -0.0630,  0.0289,  0.0280, -0.0995, -0.1243, -0.0164, -0.0807,
        -0.1432, -0.0271,  0.0246, -0.0291, -0.0264, -0.0251,  0.0956,  0.0328,
        -0.0285, -0.0022,  0.0160,  0.0705,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0137,  0.0034,  0.0223, -0.0184, -0.1288, -0.0214, -0.0130, -0.0091,
        -0.0145, -0.0068, -0.0382, -0.0523, -0.0270, -0.0101, -0.0360, -0.0265,
         0.0033, -0.0272,  0.0180, -0.0410, -0.0016, -0.0173, -0.0142, -0.0004,
        -0.0130, -0.0461, -0.0406, -0.0053, -0.0035, -0.0388, -0.0509, -0.0161,
        -0.0121, -0.0104, -0.0326, -0.0560, -0.0065, -0.0298,  0.0045, -0.0010,
         0.0075,  0.0191, -0.0417,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1115,  0.0285, -0.0158,  0.0031, -0.0241, -0.0368, -0.0547, -0.1173,
        -0.0182,  0.0004, -0.0018, -0.0130, -0.0120, -0.0793, -0.0819, -0.0070,
         0.0093, -0.0017, -0.0103,  0.0061, -0.0576, -0.0133, -0.0272,  0.0149,
         0.0078, -0.0111,  0.0005, -0.0114, -0.0462, -0.0069, -0.0032, -0.0335,
        -0.0133, -0.0425, -0.0320, -0.0009,  0.0200,  0.0249,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0636, -0.2875, -0.0742, -0.0822,  0.0090, -0.0666, -0.0133,  0.0397,
        -0.0281, -0.0932, -0.0203, -0.0088, -0.0449,  0.0042,  0.0068, -0.0265,
        -0.0123, -0.0738,  0.0137,  0.0266, -0.0042, -0.0006,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1261, -0.1531, -0.0090,  0.0025,  0.0075, -0.0080, -0.0052,  0.0288,
         0.0035, -0.1265, -0.0181,  0.0015, -0.0083,  0.0264, -0.0244, -0.0563,
        -0.1109, -0.0016, -0.0264, -0.0300,  0.0155, -0.0041, -0.0181, -0.0193,
        -0.0104,  0.0102,  0.0014,  0.0034,  0.0076, -0.0417,  0.0008,  0.0071,
         0.0090,  0.0070, -0.0127, -0.0174,  0.0226, -0.0017, -0.0088, -0.0071,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0300,  0.0360, -0.0146,  0.0009, -0.0287, -0.0150,  0.0143, -0.0090,
        -0.0737,  0.0005, -0.0350, -0.0168, -0.0153, -0.1139, -0.1257, -0.0143,
        -0.0050,  0.0155, -0.0473, -0.0209, -0.0200,  0.0163, -0.0293,  0.0008,
        -0.0142, -0.0724, -0.0215, -0.0537, -0.0732,  0.0032,  0.0249, -0.0378,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0551, -0.2834, -0.0113,  0.0097, -0.0174,  0.0031, -0.0248, -0.0984,
        -0.1110, -0.0178, -0.0166, -0.0474, -0.0196, -0.0153,  0.0187, -0.0124,
        -0.0474, -0.0051, -0.0037,  0.0005,  0.0012,  0.0095, -0.0038, -0.0549,
        -0.0034,  0.0021, -0.0311, -0.0041,  0.0021, -0.0193, -0.0026,  0.0025,
        -0.0146,  0.0033, -0.0268,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0878, -0.0758, -0.0590, -0.1068,  0.0050,  0.0086, -0.1222, -0.0794,
        -0.0685, -0.1109, -0.0217, -0.0329,  0.0038, -0.0327, -0.0300, -0.0113,
        -0.0462, -0.0139,  0.0047,  0.0347,  0.0442,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1025, -0.2878, -0.0706, -0.0685, -0.0703, -0.0874, -0.0019, -0.0653,
        -0.0729,  0.0065,  0.0135,  0.0031, -0.0054,  0.0135, -0.0315,  0.0082,
        -0.0275, -0.0052,  0.0584,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0107, -0.1832, -0.1186, -0.0095, -0.0440, -0.0013,  0.0014, -0.0376,
        -0.0026, -0.1180, -0.1743, -0.0385, -0.0313, -0.1082,  0.0102, -0.0334,
         0.0024, -0.0492, -0.0017, -0.0240,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.8694e-02, -3.4230e-01, -1.3087e-02, -8.1772e-02, -1.9566e-04,
        -1.1098e-02, -9.1065e-03, -5.1599e-02, -9.3085e-03,  4.0448e-03,
        -7.1869e-03, -7.3309e-03,  3.4069e-03, -4.0637e-02, -1.5369e-02,
        -3.5762e-02, -6.8947e-02,  9.7692e-05, -3.4501e-02, -5.7985e-02,
         1.7596e-02,  9.0062e-02, -1.9913e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0370, -0.1922, -0.0682, -0.0093,  0.0066, -0.0524, -0.0248, -0.0608,
        -0.0854, -0.0377, -0.0321, -0.0684, -0.0369, -0.0560, -0.0110, -0.0255,
         0.0019, -0.0075, -0.0138,  0.0275, -0.0045,  0.0031,  0.0030, -0.0754,
         0.0590,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0533, -0.1527, -0.0199,  0.0180,  0.0185,  0.0023, -0.0393,  0.0311,
         0.0748, -0.0806, -0.1199, -0.0057, -0.0119, -0.0301,  0.0048, -0.0430,
         0.0083, -0.0009,  0.0035, -0.0069, -0.0362, -0.0250, -0.0277, -0.0141,
        -0.0072, -0.0230, -0.0212,  0.0097, -0.0289,  0.0278, -0.0016, -0.0035,
        -0.0209, -0.0233,  0.0046,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0512,  0.0045,  0.0367, -0.0599, -0.0083, -0.0146, -0.0756, -0.0192,
        -0.1060, -0.0153, -0.0113, -0.0021, -0.0182, -0.0836, -0.0990, -0.2147,
        -0.0630, -0.0433, -0.0387, -0.0018,  0.0086,  0.0245,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0122, -0.0449, -0.0048,  0.0203, -0.0082,  0.0179, -0.0017,  0.0137,
         0.0338,  0.2179,  0.0774, -0.0448,  0.0397,  0.0539,  0.0477,  0.0873,
         0.0194,  0.0902,  0.0023, -0.0732,  0.0888,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0447, -0.4534, -0.0081, -0.0195, -0.0048,  0.0288, -0.0578, -0.1102,
        -0.0042, -0.0233, -0.0130, -0.0019,  0.0167, -0.0642,  0.0057, -0.0490,
         0.0089,  0.0223, -0.0635,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0568, -0.0238, -0.0277,  0.0087, -0.0108, -0.0458, -0.0053,  0.0030,
         0.0097, -0.0027, -0.0248,  0.0007,  0.0009, -0.0085,  0.0045,  0.0084,
        -0.0121, -0.0471, -0.0031, -0.0334, -0.0251, -0.0506, -0.0411, -0.0024,
        -0.0213, -0.0222, -0.0013, -0.0533, -0.0099, -0.1176, -0.0074, -0.0280,
        -0.0616, -0.0979,  0.0113, -0.0015,  0.0024,  0.0059,  0.0125,  0.0087,
         0.0255,  0.0268, -0.0005,  0.0034,  0.0020,  0.0144,  0.0079],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0282, -0.1537, -0.0581, -0.0002,  0.0088,  0.0155, -0.0177,  0.0171,
        -0.0356,  0.0028,  0.0072, -0.0020,  0.0304, -0.0148,  0.0130, -0.0086,
        -0.0786, -0.0124, -0.0045, -0.0167,  0.0077, -0.0635, -0.0950, -0.1148,
        -0.0138,  0.0382,  0.0028, -0.0080,  0.0061,  0.0042,  0.0121,  0.0328,
        -0.0304,  0.0446,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0869, -0.2996,  0.0136,  0.0179,  0.0012, -0.0251,  0.0228,  0.0241,
        -0.0094, -0.0211,  0.0150,  0.0022, -0.0554, -0.0645, -0.0007, -0.0127,
        -0.0018,  0.0072, -0.0244, -0.0464, -0.0129, -0.0573, -0.0084, -0.0389,
        -0.0088, -0.0029, -0.0113, -0.0212, -0.0067, -0.0207, -0.0098,  0.0090,
         0.0165,  0.0236,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0148, -0.3799, -0.0266,  0.0180, -0.0324, -0.0757, -0.0014,  0.0511,
         0.0233, -0.0960,  0.0072, -0.0407, -0.0373, -0.1007, -0.0202, -0.0366,
         0.0172, -0.0211,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0409, -0.0012, -0.0077, -0.0265, -0.0102,  0.0192, -0.0040, -0.0255,
        -0.1474, -0.2687, -0.0287, -0.0560, -0.0147,  0.0027,  0.0256,  0.0065,
        -0.0126,  0.0044, -0.0016, -0.1257,  0.0297, -0.0141, -0.0151, -0.0116,
        -0.0145,  0.0051, -0.0800,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0057, -0.0020, -0.0152,  0.0176, -0.0402, -0.0194,  0.0097,  0.0233,
         0.1472,  0.0077, -0.0070,  0.0405,  0.2055,  0.0645,  0.0167, -0.0322,
         0.0174, -0.0087, -0.0167,  0.2075,  0.0336, -0.0109,  0.0115,  0.0090,
         0.0010, -0.0294,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0475, -0.3288,  0.0246, -0.0873, -0.0939, -0.0088, -0.0289, -0.0021,
        -0.0419, -0.0420, -0.0696, -0.0385, -0.0203, -0.0129, -0.0408, -0.0374,
        -0.0133, -0.0101, -0.0167, -0.0288,  0.0058,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0683, -0.2579, -0.1923, -0.2209, -0.0643, -0.0172, -0.0716,  0.0158,
         0.0006,  0.0031,  0.0114, -0.0272,  0.0092, -0.0032, -0.0370,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1408, -0.0101,  0.0213,  0.0474,  0.0399, -0.0092,  0.0064, -0.0338,
         0.0009, -0.1001, -0.0190,  0.0295, -0.2662, -0.0783,  0.0379, -0.0741,
         0.0680,  0.0171,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0284, -0.3179, -0.0981, -0.1159, -0.0602, -0.0117, -0.0301, -0.0368,
        -0.0293, -0.0713, -0.0603, -0.0314, -0.0125, -0.0521,  0.0041,  0.0204,
        -0.0100, -0.0036, -0.0058,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0470, -0.1594, -0.0200, -0.0027, -0.0299, -0.0298, -0.0175,  0.0100,
        -0.0107,  0.0083, -0.0008,  0.0075, -0.0032,  0.0012,  0.0007,  0.0119,
        -0.0356,  0.0100,  0.0058, -0.0270, -0.1009,  0.0008, -0.0362, -0.0263,
        -0.1050, -0.0106, -0.0627, -0.0024, -0.0147, -0.0244, -0.0027,  0.0108,
        -0.0285, -0.0112, -0.0320, -0.0166, -0.0345,  0.0076, -0.0047, -0.0096,
        -0.0017,  0.0019,  0.0028, -0.0125], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0344, -0.1824, -0.0185, -0.0445, -0.0421, -0.0184, -0.0181,  0.0053,
        -0.0836, -0.0163, -0.0074,  0.0068, -0.0228, -0.0078,  0.0169,  0.0116,
         0.0092, -0.0860,  0.0166, -0.0411, -0.0315,  0.0046,  0.0043, -0.0056,
         0.0125, -0.0099, -0.0172, -0.1107,  0.0215, -0.0096, -0.0034, -0.0007,
        -0.0226,  0.0176,  0.0388,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0105, -0.3854, -0.0273, -0.0405, -0.0065, -0.0059, -0.0546, -0.0340,
        -0.0103,  0.0041,  0.0057, -0.0199, -0.0681,  0.0136,  0.0024, -0.0594,
        -0.0129, -0.0035,  0.0125,  0.0070,  0.0019, -0.0032,  0.0632,  0.1476,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0360, -0.1855, -0.0153, -0.0287, -0.0066, -0.0023, -0.0116, -0.0519,
        -0.0716, -0.0185, -0.0327, -0.0238,  0.0184, -0.0332, -0.0412, -0.0003,
         0.0157, -0.0088, -0.0390, -0.0026, -0.0102, -0.0408, -0.0109, -0.0355,
         0.0035, -0.0365, -0.0263, -0.0220, -0.0140, -0.0237, -0.0253, -0.0055,
        -0.0324, -0.0697,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0049,  0.0294, -0.0103, -0.0295,  0.0106,  0.0072, -0.1831, -0.0371,
         0.0215, -0.0192, -0.0002, -0.0390, -0.0553, -0.0167, -0.1103, -0.2027,
        -0.0009, -0.0139, -0.0075, -0.0122,  0.0100, -0.0020, -0.0199, -0.0542,
        -0.0258, -0.0165, -0.0404, -0.0031,  0.0085, -0.0080,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0898,  0.0079, -0.0491, -0.2531, -0.0309, -0.0357, -0.0334, -0.0552,
        -0.0447, -0.1333,  0.0589, -0.0723, -0.1357,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1105,  0.0044,  0.0674,  0.0603,  0.0167, -0.0705, -0.0977, -0.0664,
        -0.1707,  0.0327, -0.0098, -0.0751, -0.0292, -0.0229, -0.0761, -0.0312,
         0.0508, -0.0075,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0434, -0.2612, -0.2515,  0.0300, -0.1395, -0.0717, -0.0894, -0.1134,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0393, -0.1889,  0.0057,  0.0045, -0.0239, -0.0139, -0.0066, -0.0099,
        -0.0090, -0.0037, -0.0073,  0.0027, -0.0132, -0.0196, -0.0004,  0.0006,
         0.0081,  0.0007,  0.0038,  0.0007,  0.0108, -0.0050, -0.0286, -0.0157,
        -0.0043, -0.0204,  0.0005,  0.0045, -0.1067, -0.0447,  0.0152, -0.0466,
         0.0062, -0.0959, -0.0012, -0.0515, -0.0377, -0.0141,  0.0230, -0.0030,
        -0.0239, -0.0105, -0.0134,  0.0066,  0.0248,  0.0003,  0.0221],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0499,  0.2152, -0.0944,  0.1767,  0.0501, -0.0492, -0.0509,  0.0765,
        -0.0236,  0.0960, -0.1174,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0078,  0.3393, -0.0589,  0.1169, -0.0020,  0.0305,  0.0680,  0.1315,
         0.0183,  0.0301,  0.0007, -0.0081,  0.0233, -0.0080,  0.0658,  0.0405,
         0.0328,  0.0172,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0767, -0.1031, -0.3966,  0.0851,  0.0117, -0.0237, -0.0523, -0.0245,
        -0.0876, -0.0394,  0.0032, -0.0217,  0.0369, -0.0151, -0.0223,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0712, -0.3624, -0.0064, -0.0151, -0.0296, -0.0919, -0.1614,  0.0099,
         0.0093, -0.0873, -0.1036,  0.0005,  0.0150, -0.0042, -0.0324,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0309, -0.0001,  0.0191,  0.0587, -0.0085,  0.0328,  0.0115,  0.0043,
        -0.0630, -0.0453, -0.0164, -0.1102, -0.1404, -0.0077, -0.0054, -0.0393,
        -0.0593,  0.0242, -0.0131, -0.0067, -0.0669, -0.1006, -0.0074,  0.0368,
        -0.0329, -0.0585,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0149,  0.0036,  0.0426, -0.0522,  0.0120, -0.0261,  0.0415, -0.2477,
        -0.4230,  0.0051, -0.0137, -0.0189,  0.0193,  0.0578, -0.0215,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0084, -0.3924, -0.0193, -0.0415, -0.0408, -0.0190, -0.0377, -0.0042,
        -0.0092, -0.1496,  0.0073, -0.0190, -0.0014, -0.0072,  0.0203, -0.0155,
        -0.0178,  0.0136, -0.0419,  0.0236, -0.0240,  0.0187,  0.0120,  0.0043,
         0.0092,  0.0422,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0537, -0.0174,  0.0030,  0.0082, -0.0261, -0.0449,  0.0045,  0.0114,
        -0.0166, -0.0511, -0.1515, -0.0233, -0.0013, -0.0420, -0.1474, -0.0816,
         0.0113, -0.0499, -0.0098, -0.0520,  0.0125, -0.0165, -0.0181, -0.0291,
         0.0928, -0.0241,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.0193, -0.2712, -0.0077,  0.0007, -0.0169, -0.0006, -0.0387, -0.0534,
        -0.0019, -0.0484, -0.0077, -0.0388, -0.0617,  0.0391, -0.0248, -0.0023,
        -0.0097,  0.0056, -0.0313,  0.0011, -0.0138, -0.0243, -0.0108, -0.0417,
        -0.0608,  0.0168, -0.0346, -0.0009, -0.0063, -0.0011, -0.0044, -0.0360,
        -0.0049,  0.0106,  0.0293, -0.0232,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0074, -0.1063, -0.0511, -0.0484, -0.0366, -0.0076, -0.0789, -0.0592,
         0.0027,  0.0150, -0.0309, -0.0110, -0.0206,  0.0158, -0.0071, -0.0044,
        -0.0051, -0.0020,  0.0234, -0.0007, -0.0221, -0.0462,  0.0055, -0.0040,
        -0.0500, -0.0829,  0.0017, -0.0263, -0.0394, -0.0244, -0.0091, -0.0506,
        -0.0712,  0.0074,  0.0069, -0.0183,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0725, -0.1273, -0.0506, -0.0257, -0.0143, -0.0995, -0.0637, -0.0010,
         0.0163, -0.0208, -0.0254, -0.0226, -0.2200, -0.0116,  0.1274,  0.0024,
        -0.0228,  0.0276, -0.0148, -0.0336,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0538, -0.0661, -0.0108, -0.0243,  0.0131,  0.0173, -0.0202, -0.0298,
        -0.0100, -0.0677, -0.0127,  0.0019, -0.0020, -0.0164, -0.0986, -0.0115,
        -0.0243, -0.0388,  0.0063,  0.0029,  0.0078,  0.0085,  0.0059,  0.0136,
        -0.0124,  0.0066, -0.0184, -0.0162, -0.0174, -0.0466, -0.0364, -0.0215,
        -0.0255, -0.1115, -0.0111,  0.0493, -0.0217,  0.0415,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0168, -0.2134, -0.0579, -0.0960, -0.0019,  0.0099, -0.0147,  0.0297,
        -0.0125, -0.0177, -0.0574, -0.0671,  0.0026,  0.0121,  0.0044, -0.0540,
         0.0056, -0.0513,  0.0019,  0.0028,  0.0117,  0.0170,  0.0137,  0.0070,
         0.0196, -0.0021, -0.0070, -0.0140, -0.0503, -0.0077, -0.0130, -0.0089,
         0.0095, -0.0052, -0.0141,  0.0111,  0.0075,  0.0021,  0.0041,  0.0071,
        -0.0376], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0437, -0.4618,  0.0216, -0.0143, -0.0177, -0.0908,  0.0072, -0.0283,
        -0.0226, -0.0039, -0.0073, -0.0377, -0.0484, -0.0206,  0.0093, -0.0603,
        -0.0033, -0.0752, -0.0070,  0.0087, -0.0048, -0.0053,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0458, -0.0008, -0.0042, -0.0082,  0.0190, -0.0075, -0.0048, -0.0059,
         0.0041,  0.0156, -0.0045, -0.1005, -0.0123, -0.0096, -0.0068, -0.0762,
         0.0045, -0.0517, -0.0008, -0.0760, -0.0956,  0.0096,  0.0266, -0.0435,
        -0.0021,  0.0089, -0.0115, -0.0222, -0.1581, -0.0121, -0.0161, -0.0451,
        -0.0056,  0.0748,  0.0091,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1201,  0.0385,  0.1959,  0.0573,  0.0892, -0.0130,  0.0076,  0.0018,
         0.0468,  0.0075,  0.0309, -0.0068,  0.0244, -0.0108, -0.0085,  0.0174,
        -0.0020,  0.0025,  0.0588,  0.0083,  0.0367,  0.0086,  0.0251,  0.0555,
         0.0064, -0.0201,  0.0607,  0.0154, -0.0232,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1151, -0.3897, -0.0452, -0.0189, -0.0423, -0.0410,  0.0007,  0.0119,
         0.0068,  0.0328, -0.0128, -0.0187, -0.0124,  0.0136,  0.0239, -0.0438,
        -0.0288,  0.0140, -0.0218, -0.0115, -0.0108, -0.0131, -0.0020,  0.0038,
         0.0644,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0729,  0.0351, -0.0089, -0.0188, -0.1383,  0.0688, -0.0325, -0.1063,
        -0.1620,  0.0064, -0.0116,  0.0523,  0.0172,  0.0095, -0.0851, -0.0846,
         0.0116, -0.0096,  0.0239,  0.0025, -0.0060, -0.0166, -0.0195,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0344, -0.0101,  0.0041,  0.0043,  0.0062,  0.0179,  0.0067,  0.0023,
         0.0034, -0.0060,  0.0079,  0.0005, -0.0395, -0.1542, -0.0048, -0.0261,
        -0.0095, -0.0195, -0.0036, -0.0059,  0.0087,  0.0077,  0.0088, -0.0330,
         0.0221, -0.0172, -0.1417, -0.0377, -0.0821, -0.0451, -0.0757, -0.0152,
        -0.0310, -0.0432, -0.0167, -0.0232,  0.0030, -0.0051,  0.0159,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0620, -0.0595,  0.0440, -0.1306, -0.2133, -0.0531, -0.0146, -0.1628,
        -0.1225,  0.0377, -0.0813,  0.0185,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-1.9762e-02, -1.3316e-01, -2.5132e-01, -3.7024e-02, -4.5658e-03,
        -4.2784e-02, -5.8899e-02, -3.5959e-03,  2.0387e-02, -6.5961e-03,
         9.7854e-03,  3.9951e-03, -2.8278e-03, -1.0048e-02, -1.3736e-02,
         1.0852e-02, -8.9703e-03, -4.2564e-02, -1.1370e-03,  1.6912e-03,
        -6.8665e-03, -2.0541e-02,  8.5022e-03, -4.3132e-03, -6.3850e-03,
        -2.6180e-03, -2.7173e-03,  2.3598e-02,  1.0869e-02, -2.5568e-02,
        -1.1537e-02,  2.8150e-02, -5.9795e-02, -4.9685e-03, -3.4354e-03,
        -6.6982e-03, -2.8123e-02, -1.1652e-03,  1.3419e-04, -5.8000e-03,
         1.2723e-02,  1.8417e-03, -2.6722e-04, -2.0626e-02,  1.9053e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0558, -0.3241, -0.0995, -0.0899, -0.1119, -0.0185,  0.0176, -0.0326,
        -0.1716, -0.0291,  0.0222, -0.0273,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0657, -0.4888, -0.1331, -0.1630, -0.0837,  0.0020, -0.0144, -0.0493,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.2716e-02,  1.7494e-01,  1.2628e-02,  3.5529e-03,  2.3039e-02,
         8.3587e-03,  4.6831e-03,  8.4425e-03,  2.3972e-02,  5.6282e-02,
         4.6520e-03,  1.8170e-02,  9.6332e-03,  4.0324e-03,  3.1739e-03,
        -5.4200e-03, -1.2048e-02,  1.0299e-03, -1.6185e-02,  1.9125e-03,
         4.2381e-04, -3.9036e-03,  1.1321e-02, -1.4668e-03,  1.0026e-02,
         2.1409e-03,  1.1464e-04,  1.0617e-02,  2.1392e-02,  6.9873e-03,
         4.9378e-03,  7.4634e-03,  4.3479e-04, -2.4302e-03,  3.4838e-03,
        -1.8867e-03,  1.8639e-03,  3.3772e-03,  8.2068e-03, -7.1429e-04,
        -1.9078e-03,  1.1231e-02,  7.3783e-04,  1.0546e-02,  3.8411e-03,
        -3.6685e-02,  6.4456e-02,  1.4845e-02,  5.1149e-02,  6.6245e-02,
         3.7253e-02,  1.2271e-02,  2.7296e-02,  1.0442e-02,  1.5549e-02,
        -1.1388e-02,  6.9114e-03,  4.1039e-03, -3.5766e-03,  7.2070e-02,
        -2.9430e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0155, -0.0035, -0.0041, -0.0232, -0.0468,  0.0053, -0.0307,  0.0309,
         0.0157,  0.0206, -0.0008, -0.0899, -0.0123, -0.0453, -0.0621,  0.0093,
        -0.0244,  0.0088, -0.0633, -0.0398, -0.1004, -0.0591, -0.0239, -0.0136,
        -0.0639, -0.0103, -0.0049, -0.0177, -0.0296, -0.0205,  0.0256,  0.0607,
         0.0173,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0409, -0.0247, -0.2412, -0.0675, -0.0389, -0.0269, -0.0667, -0.0062,
         0.0186, -0.0206, -0.0014,  0.0098, -0.0077,  0.0189,  0.0015,  0.0474,
        -0.0015, -0.0404, -0.2004,  0.0145, -0.0134,  0.0230,  0.0530,  0.0147,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0312, -0.0996, -0.0769, -0.0416, -0.0108, -0.0339,  0.0079, -0.0492,
         0.0184, -0.0402, -0.0850, -0.0786, -0.2169,  0.0268,  0.0009,  0.0042,
         0.0144,  0.0007,  0.0401,  0.0160, -0.0192, -0.0222,  0.0193,  0.0462,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.4530e-02,  1.5444e-02, -3.9839e-02, -4.3685e-02,  2.0137e-03,
        -1.3906e-02, -1.4847e-02, -8.9754e-02, -8.6668e-03, -1.5714e-02,
         1.5092e-02, -2.3213e-02,  2.2430e-02, -7.3686e-04, -5.7046e-02,
        -1.0961e-01, -8.0514e-02, -4.8358e-03, -6.7122e-03, -2.8271e-02,
        -3.6460e-02, -1.1645e-02, -4.3378e-02, -8.3081e-02,  2.0609e-02,
         3.8491e-03, -7.0193e-02, -1.5681e-02,  5.4821e-03, -6.4122e-05,
        -2.1813e-02, -2.3634e-02,  9.8695e-03, -7.3774e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0650,  0.0006, -0.0102,  0.0096,  0.0533,  0.0013,  0.0062, -0.0095,
        -0.0245,  0.0104, -0.0027, -0.0204,  0.0454,  0.0968, -0.0532, -0.0707,
         0.0943, -0.0011,  0.0631,  0.0274,  0.0564,  0.1882, -0.0482,  0.0381,
         0.0033,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1112, -0.0205,  0.0395,  0.0078,  0.0331,  0.0339,  0.1414,  0.0598,
         0.0974,  0.2871, -0.0295,  0.0541,  0.0261,  0.0164,  0.0397,  0.0022,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0058, -0.2436, -0.0038, -0.0744, -0.0553, -0.0386, -0.1898, -0.0103,
        -0.0144,  0.0063, -0.0018, -0.0443, -0.0436, -0.0090, -0.0219,  0.0159,
        -0.0493, -0.0534, -0.0215, -0.0044,  0.0106,  0.0042,  0.0777,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0836, -0.0070, -0.0551, -0.1408, -0.3706, -0.0338,  0.1083, -0.0123,
        -0.0124,  0.0221, -0.0329,  0.0432, -0.0147,  0.0630,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0544, -0.1440, -0.0768, -0.1318, -0.0047,  0.0170, -0.0499,  0.0229,
        -0.0353, -0.0225, -0.0603, -0.0573, -0.0029, -0.0077, -0.0108,  0.0261,
         0.0007, -0.0947, -0.0252, -0.1113,  0.0142,  0.0038, -0.0198,  0.0057,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1332,  0.0059, -0.1406, -0.0003, -0.2047, -0.0219, -0.3318, -0.0019,
         0.0416,  0.0285, -0.0896,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1057, -0.0060,  0.0671, -0.0038,  0.0069, -0.0584, -0.1116, -0.0006,
        -0.1210, -0.0564, -0.0507,  0.0231, -0.0410,  0.0376, -0.0864, -0.0224,
        -0.0115, -0.0973,  0.0090,  0.0383,  0.0452,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0821,  0.0724, -0.0139,  0.0449,  0.0100, -0.0387, -0.0653, -0.1496,
        -0.1057, -0.2222, -0.0057,  0.0734,  0.0340, -0.0821,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.2169e-01,  4.7166e-02,  2.1160e-02,  1.7408e-02,  1.1107e-02,
         1.4206e-02,  2.4190e-02,  1.9498e-03, -1.1704e-02,  1.6493e-04,
         1.5097e-01,  5.0081e-02,  3.0531e-01,  2.3648e-02,  1.0032e-02,
         1.8005e-02,  1.0238e-01, -1.9092e-02,  3.2362e-02,  1.7002e-02,
        -3.7447e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0766,  0.3527, -0.0378,  0.0337, -0.0420,  0.0990,  0.0373,  0.0839,
         0.0217,  0.0285,  0.0689, -0.1134,  0.0046,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.2886e-02,  9.5116e-02,  1.4964e-02,  5.5803e-03,  1.2346e-02,
         1.1139e-02,  4.0830e-02,  4.3947e-02,  2.4359e-02,  2.9842e-03,
         6.1202e-03, -2.5242e-02,  9.4374e-05, -9.8964e-03, -3.3686e-03,
         8.1602e-03, -4.4628e-03,  7.6104e-03, -4.2270e-03, -2.8300e-02,
         4.1336e-03, -1.6544e-02, -2.6144e-02, -2.1209e-02,  1.6978e-01,
        -9.2429e-03,  1.0897e-01,  3.8606e-02,  3.6340e-02,  1.7917e-03,
         8.9936e-03, -1.3886e-03, -5.0929e-03, -5.3888e-04,  1.2323e-02,
         3.9272e-03,  1.8879e-02,  6.3834e-03,  2.2519e-03,  1.2617e-02,
         2.2634e-02,  3.6213e-03, -9.8135e-04, -2.6133e-02,  9.8392e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0178,  0.0508,  0.0340,  0.0097,  0.0135, -0.0110,  0.0227,  0.0399,
         0.0046,  0.0155,  0.0634, -0.0240, -0.0102,  0.0140, -0.0026,  0.0345,
         0.1225, -0.0131, -0.0017,  0.0182, -0.0036,  0.0716, -0.0117,  0.0415,
        -0.0133, -0.0256,  0.0356,  0.0426,  0.1105,  0.0119, -0.0043,  0.0118,
         0.0604,  0.0320,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0445, -0.0584,  0.0498,  0.0725,  0.0176, -0.0068, -0.0194,  0.0043,
        -0.0728,  0.0016, -0.0117,  0.0025, -0.0144, -0.0039, -0.0131, -0.0307,
         0.0146, -0.0041,  0.0439,  0.1359,  0.2473, -0.0223,  0.0158,  0.0056,
         0.0076,  0.0789,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0264, -0.2573,  0.0425, -0.0657,  0.0071,  0.0016,  0.0176, -0.0398,
        -0.0387, -0.0533, -0.0120,  0.0057, -0.0147, -0.0116, -0.0069, -0.0722,
        -0.1330,  0.0199, -0.0249, -0.0184, -0.0095, -0.0177,  0.1035,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0421, -0.0833, -0.0059, -0.0226, -0.0007, -0.0024, -0.0237, -0.0163,
        -0.0531, -0.0095,  0.0059, -0.0147,  0.0100, -0.0057, -0.1907, -0.0202,
        -0.0010, -0.0312, -0.0527, -0.0185, -0.0008,  0.0037,  0.0106,  0.0187,
         0.0152, -0.0006, -0.0789, -0.0777,  0.0055, -0.0240, -0.0181,  0.0032,
         0.0139, -0.0043,  0.0140, -0.0267, -0.0096,  0.0220, -0.0175,  0.0248,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0598,  0.0399,  0.0249,  0.0214,  0.0604,  0.0218,  0.0163,  0.0790,
        -0.0332,  0.0075, -0.0052,  0.0048,  0.0324,  0.0153,  0.0191,  0.0669,
         0.0474,  0.0225,  0.0106,  0.0177,  0.0091,  0.0492,  0.0409,  0.0026,
         0.0153,  0.0025, -0.0025,  0.0291,  0.0317,  0.0230,  0.0402,  0.0419,
         0.0740,  0.0319,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0142, -0.0386, -0.0195, -0.0149, -0.0856,  0.0107,  0.0032, -0.0166,
         0.0021,  0.0096, -0.0048, -0.0040,  0.0346, -0.0044, -0.1639,  0.0423,
         0.0174,  0.0020, -0.0578, -0.0009,  0.0031,  0.0051,  0.0193,  0.0094,
        -0.0196,  0.0179,  0.0139,  0.0063, -0.0026,  0.0050,  0.0244,  0.0055,
         0.0052, -0.0223, -0.0303, -0.0095, -0.0259, -0.0353, -0.0172, -0.0406,
        -0.0352, -0.0026, -0.0259, -0.0119, -0.0062,  0.0021,  0.0003, -0.0028,
        -0.0092, -0.0072, -0.0265,  0.0045], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.8504e-02,  4.8021e-03, -6.9066e-04, -1.9143e-02, -1.3014e-01,
        -8.7483e-03,  5.9634e-03,  2.3176e-02,  2.1869e-03, -5.3695e-04,
        -6.9198e-03, -1.1102e-02, -3.8781e-02, -1.8107e-01, -6.1010e-02,
        -5.7656e-02, -1.2314e-01,  1.2595e-02, -3.6660e-02,  4.5133e-02,
        -7.9709e-03, -3.1957e-02, -2.1701e-02, -3.9942e-02,  1.3400e-02,
         8.0719e-03,  1.2155e-02, -5.3306e-04,  8.9494e-05, -1.5390e-03,
        -3.4675e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0019, -0.2879, -0.0094, -0.0149,  0.0036,  0.0087, -0.0104,  0.0022,
        -0.0185,  0.0301, -0.0146, -0.0904, -0.0272, -0.0206, -0.0433, -0.0697,
        -0.0103, -0.0344, -0.0961, -0.0827, -0.0332, -0.0019, -0.0268, -0.0076,
         0.0534,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1397, -0.0092, -0.0097,  0.0309, -0.0808, -0.2098, -0.4097, -0.0360,
         0.0184, -0.0140, -0.0247, -0.0172,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0508,  0.3390,  0.0342,  0.0508,  0.0615,  0.0056,  0.0201, -0.0005,
         0.0030, -0.0171,  0.0071,  0.0194,  0.0520,  0.0821,  0.0257,  0.0659,
         0.0218,  0.0086, -0.0049, -0.0205,  0.0169,  0.0039, -0.0004, -0.0198,
        -0.0074,  0.0398,  0.0025,  0.0004, -0.0182,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0085,  0.4226, -0.0018, -0.0506, -0.0078,  0.1259,  0.2000, -0.0193,
         0.0008, -0.0033, -0.0640,  0.0954,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0612, -0.4778, -0.0505, -0.1305,  0.0183, -0.0749, -0.0725,  0.0237,
        -0.0601, -0.0194, -0.0062, -0.0048,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0053,  0.0486,  0.0088,  0.0217, -0.0680, -0.0113, -0.0196,  0.0097,
        -0.1368,  0.0403, -0.0295, -0.1576, -0.0255, -0.0122, -0.0360, -0.0932,
         0.0407, -0.0219, -0.0394,  0.1739,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.6311e-02,  2.0968e-01,  5.8489e-03,  4.5998e-02,  1.7710e-01,
         4.1961e-02,  6.0375e-03, -4.9762e-02,  2.2835e-02,  6.3312e-03,
         6.2840e-03,  3.0377e-02,  1.9461e-02,  1.1576e-03, -4.8437e-03,
        -2.2758e-02,  3.5054e-02, -4.8521e-03,  2.3262e-03,  9.4202e-03,
        -2.0855e-03, -7.5815e-03, -6.3901e-04, -4.2001e-05, -4.5323e-03,
        -1.3338e-03, -3.5285e-02,  1.5487e-02,  1.8825e-02, -2.2128e-02,
        -3.5490e-03,  3.2771e-03,  2.6364e-02,  6.9876e-02,  1.0682e-02,
        -3.1850e-02, -2.8068e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.3186, -0.5300, -0.0063, -0.0009,  0.0058,  0.0007, -0.0165,  0.0794,
        -0.0417,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0028,  0.0191,  0.0461, -0.1570, -0.0105,  0.0268, -0.0995, -0.0845,
         0.0216, -0.0049,  0.0054, -0.0019, -0.0805, -0.0662, -0.0206,  0.0253,
        -0.0124, -0.0151, -0.0580, -0.0454,  0.0219, -0.0025,  0.0050, -0.0025,
        -0.0195,  0.0623, -0.0826,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.1687e-02,  2.9069e-01, -1.1229e-02,  5.1635e-02,  1.1561e-03,
         9.9644e-03, -2.8635e-03,  6.4352e-03,  2.8059e-03,  1.1143e-03,
         1.1004e-02,  5.9389e-02,  1.1827e-02,  5.2680e-03, -2.6532e-03,
         1.6124e-02,  1.1619e-01, -1.2219e-02, -5.6519e-03,  1.6544e-02,
        -3.0271e-02,  4.5372e-02,  5.0840e-02,  6.6952e-03, -3.9017e-03,
         1.3952e-03,  8.3472e-03,  3.0152e-02,  1.0804e-01,  1.3252e-04,
        -2.0419e-04,  5.7518e-03,  2.3142e-03,  1.2857e-04,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0162, -0.0034, -0.0039, -0.0051, -0.0224, -0.0119,  0.0033,  0.0163,
         0.0033,  0.0236, -0.0418, -0.0832, -0.0107, -0.0063, -0.0068, -0.0592,
        -0.0858,  0.0162, -0.0185, -0.0066,  0.0298,  0.0081, -0.0441, -0.0247,
         0.0120, -0.0028, -0.0126, -0.0094, -0.0044, -0.0041, -0.0217, -0.0484,
        -0.0837, -0.0299, -0.0208, -0.0163, -0.0606, -0.0095,  0.0020,  0.0063,
        -0.0175, -0.0489,  0.0094, -0.0031, -0.0251,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0018, -0.0411, -0.0441, -0.0180, -0.0716, -0.1129, -0.2957, -0.0104,
        -0.0536, -0.1320, -0.0767, -0.0101, -0.0599, -0.0004, -0.0353, -0.0364,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0362, -0.0097,  0.0273,  0.0279,  0.0323, -0.0121, -0.0185,  0.0407,
         0.2024,  0.1769,  0.0761,  0.1172, -0.0277,  0.0556,  0.0351,  0.0188,
         0.0025, -0.0017, -0.0140,  0.0068,  0.0025,  0.0048, -0.0090,  0.0078,
        -0.0211, -0.0090, -0.0059,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.9209e-02,  2.5473e-02, -2.3523e-02, -2.7012e-03, -2.6271e-02,
        -1.5653e-02, -2.1068e-02, -7.8761e-03, -3.1248e-02, -3.5105e-03,
         8.0958e-03, -1.2435e-02, -1.9081e-02, -3.2957e-03, -8.6447e-03,
        -7.2262e-03, -2.8387e-02, -6.9562e-03,  1.1110e-02, -5.4247e-03,
         2.4155e-02, -2.6361e-02, -2.4695e-02, -1.3131e-01,  1.8713e-02,
        -1.4932e-02, -2.9470e-02, -1.0102e-01, -1.7200e-02, -6.1124e-02,
         9.8821e-04, -3.8634e-02, -6.6071e-02, -5.8408e-03, -1.2038e-02,
        -7.1606e-03,  1.4503e-03, -4.2559e-02,  1.3906e-02,  1.1873e-02,
        -1.5237e-03, -1.1808e-02, -1.2058e-04,  5.8866e-03,  2.6874e-02,
         1.7099e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1464, -0.0096, -0.0020, -0.0073,  0.0056, -0.0194, -0.0167, -0.1917,
         0.0313, -0.0015, -0.0394, -0.0023,  0.0243, -0.0214, -0.0190,  0.0071,
         0.0003, -0.0105, -0.1079, -0.1815, -0.0393, -0.0137, -0.0007, -0.0022,
        -0.0022, -0.0018, -0.0570,  0.0095, -0.0043,  0.0244,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1448,  0.0262, -0.0435,  0.0437,  0.0622,  0.0750, -0.0168,  0.4329,
        -0.0787,  0.0466,  0.0296,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0235, -0.2941,  0.0031, -0.0292, -0.0925, -0.0501, -0.0805, -0.1443,
         0.0139,  0.0112,  0.0167,  0.0030, -0.0643, -0.0270, -0.0107, -0.0760,
        -0.0206, -0.0271, -0.0058, -0.0065,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0215, -0.4595, -0.1193, -0.0489,  0.0894, -0.0595, -0.0969,  0.0131,
         0.0085, -0.0399,  0.0435,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0382,  0.2906,  0.1853,  0.1070,  0.0876,  0.0282,  0.0159,  0.0362,
         0.0462, -0.0114,  0.0117,  0.0217, -0.0197,  0.0029, -0.0026,  0.0259,
        -0.0052,  0.0040, -0.0125,  0.0011, -0.0463,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1320, -0.1708, -0.1429, -0.0013,  0.0185, -0.2708,  0.0499, -0.0077,
         0.0183, -0.0247, -0.0007, -0.0007,  0.0068,  0.0108, -0.0052,  0.0106,
         0.0877, -0.0407,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0228, -0.0860, -0.1513,  0.0165,  0.0221,  0.0131,  0.0066,  0.0132,
        -0.0058,  0.0015, -0.0032,  0.0056, -0.0049, -0.0378, -0.0520, -0.0076,
        -0.0085, -0.0400, -0.0270,  0.0094, -0.0059, -0.0481, -0.0827,  0.0045,
        -0.0744, -0.0144, -0.0559, -0.0259, -0.0262, -0.0524, -0.0030, -0.0063,
         0.0002,  0.0045, -0.0074,  0.0127,  0.0123,  0.0126,  0.0160,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0292, -0.2970, -0.0509, -0.0688, -0.0201, -0.0016, -0.0073, -0.0028,
        -0.0342, -0.0162, -0.0213, -0.0630, -0.0164, -0.0095,  0.0249, -0.0065,
        -0.0525, -0.0045, -0.0015, -0.0178, -0.0105, -0.0030,  0.0062,  0.0093,
         0.0038, -0.0069, -0.0158, -0.0036, -0.0458, -0.0707, -0.0049,  0.0035,
        -0.0188, -0.0511,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.0982, -0.2700, -0.1042, -0.1452,  0.0531, -0.0725,  0.1420,  0.0052,
        -0.1097,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0083,  0.1762,  0.0746,  0.0656,  0.0290,  0.0412,  0.0100, -0.0005,
         0.0096, -0.0298,  0.0356,  0.0137, -0.0008,  0.0099,  0.0144,  0.0557,
         0.0120,  0.0923,  0.0163,  0.0083, -0.0139, -0.0031, -0.0172, -0.0058,
         0.0256,  0.0182,  0.0042, -0.0020,  0.0015, -0.0088,  0.0299,  0.0458,
        -0.0055,  0.0345, -0.0248, -0.0085, -0.0058,  0.0040,  0.0220, -0.0152,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0088, -0.0720, -0.0973, -0.0157, -0.0521, -0.0069,  0.0035, -0.0063,
        -0.0381, -0.0490, -0.0020,  0.0032, -0.0005, -0.0027, -0.0119,  0.0008,
        -0.0073,  0.0310, -0.0084,  0.0186, -0.0031,  0.0082,  0.0012, -0.0221,
         0.0129, -0.0176, -0.0374,  0.0097, -0.0737, -0.0539, -0.0129, -0.0455,
         0.0013, -0.0152,  0.0131, -0.0305, -0.0344, -0.0548, -0.0007,  0.0061,
         0.0042, -0.0039,  0.0023,  0.0042, -0.0058,  0.0073, -0.0136,  0.0138,
         0.0101, -0.0094,  0.0219, -0.0128,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0690,  0.3431,  0.0034,  0.0138,  0.0008, -0.0168,  0.0398,  0.0099,
        -0.0050, -0.0297, -0.0289, -0.0117,  0.0752,  0.0065,  0.0342,  0.0699,
         0.0277, -0.0206,  0.0586, -0.0169,  0.0024,  0.1162,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0213,  0.5006,  0.0244,  0.0202, -0.0216, -0.0227,  0.0261,  0.0234,
        -0.0251,  0.0170,  0.0241,  0.0094,  0.0346,  0.0163,  0.0045,  0.0874,
         0.0511,  0.0213,  0.0043,  0.0118, -0.0328,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0232, -0.2015, -0.0362, -0.0291, -0.0028, -0.0009, -0.0097,  0.0114,
        -0.0132,  0.0078, -0.0045, -0.0080,  0.0065,  0.0168,  0.0021,  0.0092,
         0.0218,  0.0042, -0.0119, -0.0016,  0.0057,  0.0058, -0.0025,  0.0175,
         0.0015,  0.0030, -0.0187, -0.0245, -0.0080, -0.0349, -0.0633, -0.0576,
        -0.0099, -0.0371, -0.0202, -0.0093, -0.0023, -0.0084, -0.0008, -0.0094,
        -0.0693, -0.0013,  0.0158, -0.0117, -0.0258,  0.0055, -0.0107, -0.0115,
         0.0044,  0.0141, -0.0338, -0.0103, -0.0043,  0.0109, -0.0078],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0063, -0.0378, -0.0012, -0.0437, -0.0237, -0.0230, -0.0531, -0.0019,
         0.0073, -0.0079, -0.0150,  0.0011, -0.0198, -0.0232, -0.0195, -0.0032,
         0.0143, -0.0137, -0.0061, -0.0322, -0.0062, -0.0122, -0.0263,  0.0125,
        -0.0126,  0.0119, -0.0166, -0.0494, -0.0083,  0.0130,  0.0200, -0.0008,
        -0.0169, -0.1156, -0.0069, -0.0499, -0.0270, -0.0557, -0.0421, -0.0758,
        -0.0213, -0.0211,  0.0002, -0.0123, -0.0114,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0514, -0.1967, -0.0778, -0.0965,  0.0163, -0.0248, -0.0147, -0.0402,
        -0.0360,  0.0211, -0.0114, -0.0037, -0.0089, -0.0958, -0.0279,  0.0684,
         0.0135, -0.0583,  0.0047,  0.0017, -0.0610,  0.0037,  0.0394,  0.0262,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0369,  0.0151,  0.0087,  0.0465, -0.0018, -0.0148, -0.0572, -0.0670,
         0.0099, -0.0029, -0.0141,  0.0105, -0.0171, -0.0337, -0.0006, -0.0110,
        -0.0706, -0.0134, -0.0391, -0.0454, -0.0149,  0.0041, -0.0443, -0.0925,
        -0.0619, -0.0405, -0.0668, -0.0018, -0.0152, -0.0800, -0.0094, -0.0191,
        -0.0333,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0195, -0.4352, -0.0382, -0.1141, -0.0099, -0.0261,  0.0085, -0.0097,
        -0.1175,  0.0154, -0.0101, -0.0185,  0.0285, -0.0562,  0.0135,  0.0094,
        -0.0272,  0.0220, -0.0207,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0230, -0.0404,  0.0148, -0.0132, -0.1386,  0.0162, -0.0754, -0.0557,
        -0.1263, -0.0074, -0.0042,  0.0004,  0.0068,  0.0144, -0.0002,  0.0074,
        -0.0445, -0.0598, -0.0247,  0.0280, -0.0833, -0.1079, -0.0447,  0.0023,
         0.0607,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0622, -0.3064, -0.0758, -0.0300, -0.0603, -0.1358, -0.1575,  0.0178,
        -0.0047,  0.0125,  0.0036, -0.0061, -0.0429, -0.0138, -0.0094,  0.0188,
         0.0425,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-1.5945e-02,  9.7205e-03, -4.3439e-02, -7.2088e-03, -5.4168e-04,
         4.0976e-06, -6.3616e-03, -5.6985e-03,  6.0078e-03, -6.8325e-02,
        -2.4693e-02, -3.8536e-03,  1.3094e-02,  1.7315e-03,  4.3248e-03,
         1.7081e-03, -6.7659e-02, -6.3662e-02, -7.2080e-03, -5.6385e-02,
        -1.7991e-02, -1.4978e-01, -2.1389e-01, -1.5261e-02, -3.0813e-02,
        -3.0881e-02, -1.0021e-03,  2.9594e-03, -1.8011e-02,  1.1460e-02,
         5.9656e-02, -4.0719e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0510,  0.1022,  0.0846,  0.0452,  0.0195, -0.0071,  0.0724,  0.1323,
         0.0213,  0.0091, -0.0063,  0.0097, -0.0041,  0.0222,  0.0198, -0.0233,
         0.0104,  0.0299,  0.0421,  0.0181,  0.0308, -0.0026,  0.0686,  0.0923,
        -0.0215,  0.0107, -0.0317,  0.0111,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0872, -0.1241, -0.0131, -0.0022, -0.0005, -0.0541, -0.1525, -0.0233,
        -0.0208, -0.0281, -0.0048, -0.0531,  0.0047, -0.0206,  0.0015, -0.0719,
        -0.0229, -0.0916, -0.0345,  0.0027, -0.0056, -0.0019, -0.0095, -0.0069,
        -0.0443, -0.0061, -0.0006,  0.0148,  0.0961,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0928,  0.0319, -0.0056,  0.0422,  0.1417,  0.0589,  0.0053,  0.0007,
         0.0057, -0.0104,  0.0043,  0.0122,  0.0479,  0.1188,  0.0134,  0.0308,
        -0.0279,  0.1068,  0.0211,  0.0777,  0.0034,  0.0150,  0.0380,  0.0059,
         0.0191,  0.0193,  0.0090,  0.0064, -0.0103,  0.0068, -0.0003,  0.0014,
         0.0090,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0676, -0.2873,  0.0026, -0.0711, -0.0075, -0.0398,  0.0031, -0.0647,
         0.0014, -0.0279, -0.0343,  0.0136,  0.0057,  0.0016, -0.0056,  0.0152,
        -0.0381,  0.0030, -0.1093,  0.0170, -0.0652, -0.0079, -0.0608,  0.0496,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0674, -0.1644, -0.1221, -0.0100, -0.0668, -0.0125,  0.0112, -0.0368,
         0.0081, -0.0555,  0.0028, -0.0545, -0.0566, -0.0227,  0.0079,  0.0047,
         0.0238, -0.0292, -0.0131, -0.1290, -0.0388, -0.0055,  0.0565,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0152, -0.0995,  0.0172,  0.0030, -0.0056, -0.0286,  0.0160,  0.0065,
         0.0079, -0.0212,  0.0142, -0.0227, -0.0142,  0.0017,  0.0013, -0.0448,
        -0.1135, -0.0508, -0.0330, -0.0070, -0.0120, -0.0568, -0.0153, -0.0047,
        -0.0549, -0.0011, -0.0030,  0.0018,  0.0062, -0.0648, -0.0389,  0.0072,
        -0.0240,  0.0049, -0.0163, -0.0639, -0.0383, -0.0491, -0.0131,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0502,  0.0369, -0.0017,  0.0099,  0.0729,  0.0109,  0.0040, -0.0100,
         0.0075, -0.0069, -0.0113, -0.0034, -0.0047, -0.0024, -0.0194,  0.0566,
        -0.0032, -0.0162,  0.0035, -0.0109,  0.0587, -0.0008,  0.0117,  0.0345,
         0.1234,  0.1251,  0.0354,  0.0127, -0.0162,  0.0442,  0.0090, -0.0084,
         0.0026,  0.0088,  0.0073, -0.0138,  0.0133,  0.0709,  0.0021,  0.0075,
         0.0153,  0.0072, -0.0283,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.3000e-02, -1.6739e-01,  4.6247e-03,  6.5421e-03,  7.1896e-03,
         7.1738e-03, -1.4564e-02,  1.2852e-02, -5.4448e-02, -6.2623e-03,
        -2.0323e-04, -7.5606e-04, -9.9482e-03, -9.4718e-03, -5.5779e-03,
        -6.6633e-03, -2.3385e-02, -5.2296e-02, -6.1984e-03, -3.1844e-04,
         1.5916e-02,  2.4777e-02, -1.1716e-03,  1.4048e-03,  5.8353e-03,
         2.9615e-04, -5.6509e-03,  1.9876e-03,  5.2934e-03,  4.4271e-03,
         7.6954e-03, -5.2352e-02, -5.8609e-02, -2.0692e-02, -1.3202e-02,
        -3.8839e-02, -1.0904e-02, -5.2559e-02, -6.5731e-02, -2.9816e-02,
        -3.0578e-02, -3.8257e-02, -1.9202e-03,  2.4726e-04, -2.4041e-03,
        -5.5929e-04,  1.1807e-03,  5.5624e-03,  2.4727e-03,  8.7619e-05,
        -1.9002e-04,  8.0224e-03,  7.7859e-03,  5.5746e-03, -1.6098e-02,
         8.6281e-03, -1.4412e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0495, -0.3350,  0.0202, -0.0395, -0.0512,  0.0604, -0.0669, -0.1777,
        -0.0640, -0.0215, -0.0579,  0.0126, -0.0168,  0.0268,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0757,  0.1146,  0.0637,  0.0041,  0.0070,  0.0064, -0.0032, -0.0004,
         0.0304,  0.0016,  0.0243,  0.1076,  0.0917,  0.0123,  0.0563,  0.0414,
         0.0378,  0.0551, -0.0280,  0.0467,  0.0808,  0.0077,  0.0297, -0.0006,
        -0.0024,  0.0068,  0.0034, -0.0088,  0.0052, -0.0462,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0201,  0.2731, -0.0540,  0.2635,  0.2128,  0.0993,  0.0504, -0.0267,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([-0.0032, -0.1208, -0.0175,  0.0054, -0.0432, -0.1036,  0.0067, -0.0003,
         0.0157,  0.0101,  0.0107, -0.0180, -0.0586, -0.0807, -0.0034, -0.0034,
        -0.0026, -0.0061,  0.0055,  0.0128, -0.0160, -0.0124,  0.0124, -0.0100,
        -0.0704, -0.0259, -0.0580, -0.0177, -0.0604, -0.0868, -0.0131, -0.0480,
        -0.0406,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.7884e-02, -2.0017e-01, -7.4815e-02, -2.7680e-02,  3.1200e-03,
        -3.7874e-02, -1.7255e-02, -1.7220e-02, -6.0064e-03, -3.2508e-02,
        -5.4403e-03, -1.6839e-04, -3.9545e-03, -1.0877e-03,  1.5871e-02,
        -6.4020e-03,  1.4578e-03,  6.2485e-03, -1.0362e-02,  7.2371e-03,
        -6.4557e-03, -7.0413e-03,  1.8880e-03, -2.1982e-03,  1.3039e-03,
        -3.9228e-03, -2.0056e-03, -4.1408e-02,  3.1893e-03,  1.6767e-03,
        -5.5541e-03, -5.1216e-03, -5.5279e-02, -5.4715e-02,  3.9311e-03,
        -2.7263e-02, -1.3096e-02,  3.8319e-03,  6.5143e-04, -4.5484e-04,
         5.0643e-04,  8.4134e-03, -4.4198e-03, -8.5590e-03,  2.1000e-03,
         1.4592e-02, -1.8418e-03, -4.7255e-03, -2.2675e-03,  4.3742e-03,
        -2.6208e-03, -1.4447e-02, -1.7790e-03, -1.0027e-02, -1.0659e-02,
         1.6844e-03,  4.8993e-04, -6.0090e-03, -1.0257e-02, -1.5525e-02,
        -1.3388e-02, -1.4041e-03, -6.2365e-02, -9.3911e-03,  3.4927e-03,
        -1.4082e-03, -1.1209e-03,  4.0569e-03, -4.9521e-03, -1.8052e-02,
        -1.1321e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0310, -0.0218, -0.0153, -0.0055, -0.0261, -0.0355, -0.1609, -0.0078,
        -0.0052,  0.0085, -0.0017, -0.0103,  0.0060,  0.0037, -0.0056, -0.0028,
        -0.0038, -0.0096, -0.0214, -0.1161, -0.0283,  0.0085,  0.0194, -0.0576,
         0.0081, -0.0403, -0.0355, -0.0061, -0.0031, -0.0135,  0.0060, -0.0080,
        -0.0554, -0.0418, -0.0198, -0.0156, -0.0712, -0.0106,  0.0036,  0.0056,
         0.0155,  0.0279,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.2445e-02, -2.0628e-01,  1.8045e-02,  7.1047e-03, -1.9214e-04,
         1.0165e-02,  1.3708e-02,  6.2753e-03,  4.6119e-04, -4.2043e-02,
        -3.0793e-02, -3.1209e-02, -8.7822e-02,  2.8335e-03, -6.2050e-03,
         1.6918e-02,  3.9068e-03, -8.5497e-02, -9.9982e-02,  3.0349e-04,
        -1.3217e-02, -3.0742e-02, -4.4051e-03, -1.5286e-02, -6.6865e-03,
        -6.2496e-02, -3.5456e-02, -8.3408e-04, -5.6612e-02, -1.4793e-02,
        -1.0508e-02, -5.2954e-03,  3.6989e-03, -2.4837e-03,  4.8737e-03,
        -4.0526e-03,  1.6372e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0971, -0.4391,  0.0017, -0.0651,  0.0111,  0.0314, -0.0079,  0.0403,
         0.0341, -0.0182, -0.0362,  0.0284, -0.1172, -0.0445, -0.0279,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0151,  0.0345, -0.0030,  0.0290,  0.1394,  0.1338, -0.0243,  0.0443,
         0.1905,  0.0090, -0.0512, -0.0175,  0.0770, -0.0216, -0.0719, -0.0017,
         0.0383, -0.0596, -0.0078,  0.0305,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0076,  0.0090,  0.0043,  0.0086, -0.1620, -0.0360,  0.0015, -0.0177,
        -0.0241, -0.0084, -0.0369, -0.0725, -0.0222, -0.0159, -0.0296, -0.0177,
        -0.0006, -0.0152,  0.0131, -0.0309, -0.0112,  0.0004, -0.0080,  0.0065,
        -0.0086, -0.0416, -0.0866, -0.0006,  0.0010, -0.0383, -0.0718, -0.0044,
        -0.0077, -0.0076, -0.0221, -0.0267,  0.0163, -0.0093, -0.0139,  0.0085,
         0.0237, -0.0123, -0.0393,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0078, -0.0122, -0.0219,  0.0145, -0.0338, -0.0150, -0.0521, -0.1577,
        -0.0118,  0.0158,  0.0172,  0.0188, -0.0035, -0.0673, -0.0705, -0.0093,
        -0.0007,  0.0167, -0.0256,  0.0042, -0.1083,  0.0005,  0.0316,  0.0212,
        -0.0051,  0.0011, -0.0039,  0.0096, -0.0149, -0.0066, -0.0102, -0.0318,
         0.0091, -0.0654, -0.0653, -0.0238,  0.0036, -0.0115,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0055, -0.3005, -0.1041, -0.0742,  0.0137, -0.0520,  0.0288, -0.0077,
        -0.0182, -0.0775, -0.0161,  0.0065, -0.0510, -0.0014,  0.0068, -0.0402,
         0.0069, -0.0625, -0.0171,  0.0499, -0.0188,  0.0409,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0036,  0.0902, -0.0066, -0.0019, -0.0539,  0.0085,  0.0034,  0.0094,
         0.0286,  0.1235,  0.0242, -0.0058,  0.0245,  0.0232,  0.0194,  0.0318,
         0.0885,  0.0401,  0.0137,  0.0450, -0.0045, -0.0041,  0.0159,  0.0074,
        -0.0029, -0.0052,  0.0126, -0.0117,  0.0103,  0.0411, -0.0152,  0.0014,
        -0.0092,  0.0135,  0.0227,  0.0848, -0.0069,  0.0259, -0.0205,  0.0385,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0397,  0.0074,  0.0113,  0.0076, -0.0091,  0.0101,  0.0303, -0.0193,
        -0.0993, -0.0088, -0.0102, -0.0249,  0.0020, -0.1406, -0.0832,  0.0020,
         0.0095, -0.0174, -0.0412,  0.0305,  0.0103,  0.0156, -0.0021, -0.0017,
        -0.0111, -0.0769, -0.0351, -0.0459, -0.0856,  0.0086,  0.0442, -0.0586,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0127, -0.3371, -0.0307, -0.0008, -0.0219, -0.0072, -0.0094, -0.0476,
        -0.0951,  0.0143, -0.0085, -0.0461, -0.0194, -0.0015,  0.0036, -0.0044,
        -0.0357,  0.0064, -0.0007,  0.0181, -0.0032,  0.0172,  0.0129, -0.0639,
         0.0028,  0.0164, -0.0199,  0.0053,  0.0071, -0.0132,  0.0063,  0.0015,
        -0.0150,  0.0489, -0.0454,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0888, -0.0579, -0.1174, -0.0597, -0.0107, -0.0259, -0.1880, -0.0889,
         0.0894, -0.0568,  0.0248, -0.0454,  0.0119, -0.0318,  0.0020, -0.0025,
        -0.0284, -0.0030,  0.0168, -0.0042, -0.0456,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0393, -0.4055,  0.0196, -0.0419, -0.0942, -0.1062, -0.0203, -0.0574,
        -0.0930, -0.0043,  0.0024, -0.0131,  0.0031,  0.0053, -0.0267,  0.0157,
        -0.0130,  0.0227,  0.0163,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0822, -0.0869, -0.0964,  0.0229, -0.0254, -0.0227,  0.0370, -0.0228,
         0.0024, -0.0627, -0.1604, -0.0022, -0.0289, -0.0990, -0.0289, -0.0132,
        -0.0214, -0.0515,  0.1311, -0.0021,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0430, -0.3851, -0.0231, -0.1226,  0.0167, -0.0223, -0.0051, -0.0527,
        -0.0193, -0.0159, -0.0056,  0.0056, -0.0019, -0.0480, -0.0094, -0.0179,
        -0.0365, -0.0282, -0.0323, -0.0377,  0.0254, -0.0109,  0.0346,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0186, -0.1635, -0.0408, -0.0349,  0.0113, -0.0568, -0.0112, -0.0554,
        -0.0533, -0.1065, -0.0032, -0.0707, -0.0441, -0.1180,  0.0444,  0.0162,
        -0.0155, -0.0219, -0.0338, -0.0005, -0.0225,  0.0010, -0.0040, -0.0184,
         0.0338,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0005, -0.3985, -0.0642, -0.0048, -0.0273, -0.0230,  0.0012, -0.0038,
         0.0486, -0.0183, -0.0959,  0.0076, -0.0032,  0.0063, -0.0019,  0.0130,
        -0.0060,  0.0048,  0.0037, -0.0007, -0.0263, -0.0268, -0.0236, -0.0200,
        -0.0120, -0.0059,  0.0083,  0.0047, -0.0162, -0.0113,  0.0045,  0.0168,
        -0.0110,  0.0068,  0.0725,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0633,  0.0005, -0.0052, -0.0437, -0.0066,  0.0305, -0.0883, -0.0355,
        -0.1300, -0.0427, -0.0690, -0.0106,  0.0024, -0.0661, -0.0482, -0.0930,
        -0.0349, -0.1089,  0.0464,  0.0263,  0.0223,  0.0257,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0861,  0.0144,  0.0944,  0.0046,  0.0396,  0.0792,  0.0028,  0.0194,
        -0.0018,  0.2540,  0.1469,  0.0620,  0.0377,  0.0293,  0.0040,  0.0380,
         0.0062,  0.0331,  0.0323, -0.0100,  0.0042,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0525, -0.3947, -0.0085, -0.0196,  0.0164, -0.0117, -0.0799, -0.1004,
        -0.0033, -0.0015, -0.0424, -0.0350, -0.0015, -0.1077, -0.0036, -0.0608,
        -0.0306,  0.0290,  0.0008,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0439, -0.0118, -0.0148,  0.0172,  0.0098, -0.0957,  0.0022, -0.0115,
         0.0078, -0.0093, -0.0144, -0.0300, -0.0161, -0.0269, -0.0043, -0.0062,
        -0.0554, -0.0560,  0.0008, -0.0362, -0.0209, -0.0155, -0.0401, -0.0006,
        -0.0218,  0.0100, -0.0070, -0.0735, -0.0358, -0.0864,  0.0063, -0.0072,
        -0.0204, -0.0842, -0.0046, -0.0051,  0.0150,  0.0148,  0.0151,  0.0018,
         0.0002,  0.0086, -0.0033, -0.0041,  0.0075,  0.0107,  0.0088],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0495, -0.1886, -0.0718, -0.0154,  0.0218,  0.0210, -0.0034,  0.0044,
        -0.0437,  0.0049,  0.0087, -0.0124, -0.0034, -0.0467,  0.0015, -0.0056,
        -0.1133, -0.0015,  0.0042, -0.0089,  0.0151, -0.0225, -0.0640, -0.0521,
         0.0104,  0.0283, -0.0170, -0.0094, -0.0002, -0.0059, -0.0053,  0.0201,
        -0.1142,  0.0045,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0376, -0.4161,  0.0167,  0.0093, -0.0231, -0.0248, -0.0071,  0.0243,
        -0.0090, -0.0109, -0.0017, -0.0139, -0.0420, -0.0626,  0.0005, -0.0122,
        -0.0032,  0.0328, -0.0113, -0.0965, -0.0188,  0.0037, -0.0042,  0.0113,
         0.0049,  0.0105,  0.0091, -0.0185, -0.0117, -0.0342,  0.0082, -0.0019,
        -0.0067,  0.0009,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.1092,  0.2916,  0.0378,  0.0063,  0.0697,  0.1096,  0.0502, -0.0064,
         0.0102,  0.0812,  0.0339,  0.0220,  0.0274,  0.0761,  0.0179,  0.0242,
         0.0066,  0.0196,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1014,  0.0005, -0.0055,  0.0211, -0.0076,  0.0065,  0.0097, -0.0040,
        -0.0520, -0.2584, -0.0132, -0.0146, -0.0026, -0.0372, -0.0033, -0.0080,
        -0.0325,  0.0339, -0.0174, -0.1877,  0.0336, -0.0237, -0.0390,  0.0064,
        -0.0104,  0.0541, -0.0159,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.7790e-02, -8.8949e-05, -3.2554e-02,  1.2229e-02,  1.9765e-02,
         6.7527e-03,  2.5604e-02, -9.0274e-03,  6.3524e-02,  9.5602e-03,
        -1.2252e-02,  2.7220e-02,  2.4163e-01,  4.8371e-02, -2.1908e-03,
         3.7418e-03,  2.7404e-02,  2.7976e-02, -2.0340e-02,  2.1729e-01,
         2.0624e-02, -8.3221e-03,  4.0366e-02, -1.9493e-02,  8.4369e-03,
        -1.7453e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0215, -0.2501,  0.0414, -0.0701, -0.1537, -0.0222,  0.0061,  0.0124,
        -0.0638, -0.0467, -0.0590, -0.0352,  0.0147,  0.0469,  0.0102, -0.0788,
        -0.0253, -0.0025, -0.0368,  0.0023,  0.0003,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1031, -0.2145, -0.1666, -0.1643, -0.0327, -0.0096, -0.1289, -0.0052,
        -0.0232,  0.0348,  0.0007, -0.0293,  0.0285,  0.0075, -0.0511,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0934, -0.0179, -0.0186, -0.0011, -0.0376,  0.0310,  0.0180,  0.0917,
         0.0591, -0.0356, -0.0199,  0.0527,  0.2972,  0.0898, -0.0632, -0.0131,
        -0.0078,  0.0523,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0054,  0.2698,  0.0995,  0.1205,  0.0076,  0.0338,  0.0155,  0.0339,
         0.0159,  0.0579,  0.0676,  0.0233, -0.0355,  0.0474,  0.0153, -0.0191,
        -0.0097,  0.0150, -0.1074,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0594, -0.0922, -0.0345,  0.0020, -0.0109, -0.0212,  0.0069, -0.0013,
        -0.0057,  0.0014, -0.0006,  0.0143,  0.0021, -0.0059, -0.0052,  0.0127,
        -0.0524, -0.0040,  0.0066, -0.1106, -0.0412, -0.0090, -0.0456, -0.0072,
        -0.0645, -0.0061, -0.0527,  0.0139, -0.0034, -0.0440, -0.0085,  0.0358,
        -0.0230, -0.0059, -0.0094, -0.0059, -0.0694, -0.0009,  0.0067, -0.0069,
         0.0042,  0.0088,  0.0048, -0.0724], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.1464e-02, -2.8808e-01, -3.9195e-02, -3.6003e-02, -1.0713e-02,
        -8.5533e-03, -1.7783e-02,  1.1239e-02, -5.5979e-02, -1.4135e-02,
        -4.1915e-02,  1.0906e-02, -2.2687e-02, -2.3682e-03,  1.3652e-02,
         1.0495e-02,  7.8825e-03, -9.0816e-02,  4.0672e-03, -1.0864e-02,
        -3.7128e-02,  2.2218e-04,  4.3777e-03, -1.2037e-03, -2.3112e-02,
         6.2841e-03,  5.8830e-03, -6.9297e-02, -1.8575e-02, -1.5359e-02,
        -7.7853e-03, -1.7843e-02, -1.7103e-05, -3.8537e-02, -3.5581e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0209,  0.4070, -0.0161,  0.0128,  0.0105, -0.0238,  0.0699,  0.0470,
        -0.0327, -0.0051,  0.0059,  0.0285,  0.0401,  0.0005,  0.0028,  0.0639,
         0.0112,  0.0025, -0.0234, -0.0042, -0.0137,  0.0183, -0.0398, -0.0993,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.4233e-02, -2.1837e-01, -2.9130e-02, -4.3221e-02, -2.4032e-02,
        -2.1454e-02, -1.7811e-02, -1.9627e-02, -9.0579e-02, -9.2175e-03,
        -1.9644e-04,  5.1546e-03, -3.0098e-03, -3.3324e-02, -5.0238e-02,
        -1.7101e-03,  1.2588e-02, -1.3982e-03, -6.6255e-02,  2.4203e-02,
         1.2046e-02, -3.3502e-02, -1.8996e-03, -2.3907e-02,  6.3958e-03,
        -4.1786e-02, -2.4523e-02, -1.3753e-02, -1.7152e-02, -4.8070e-03,
        -4.0294e-02, -5.9733e-03, -2.9128e-02, -2.9079e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0377, -0.0166,  0.0133,  0.0268, -0.0191, -0.0298,  0.1535,  0.0059,
        -0.0156, -0.0020,  0.0067,  0.0160,  0.0941,  0.0193,  0.0306,  0.1425,
         0.0214,  0.0058,  0.0020, -0.0260, -0.0259, -0.0045, -0.0232,  0.0576,
         0.0700,  0.0198,  0.0383,  0.0027, -0.0357,  0.0378,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.1475,  0.0758, -0.0429, -0.2382, -0.0395, -0.0067, -0.0388, -0.0213,
        -0.0899, -0.1307, -0.0404, -0.0151, -0.1131,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0629,  0.0121, -0.0117,  0.0778,  0.0028, -0.0309, -0.0669, -0.0986,
        -0.1692,  0.0154, -0.0003, -0.0818,  0.0644, -0.0773, -0.1500, -0.0079,
         0.0410, -0.0291,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0541,  0.1596,  0.2993,  0.0386,  0.2346, -0.0606, -0.0379,  0.1154,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.5025e-02, -1.8316e-01,  5.0175e-03,  2.9215e-03, -2.3410e-02,
        -1.1486e-02, -1.1447e-02, -3.5407e-03,  1.0696e-02,  7.6475e-03,
         6.4329e-03,  2.0776e-02, -1.3262e-02, -1.4327e-03, -5.3339e-03,
        -6.6980e-03, -2.7679e-03,  1.3702e-04,  5.7766e-04, -1.3296e-02,
         1.4199e-02, -6.5855e-03, -4.6213e-02, -3.8267e-02, -1.2534e-02,
        -2.4295e-02, -1.0874e-02, -8.9118e-03, -9.0142e-02, -3.3358e-02,
        -1.1565e-02, -4.1004e-02,  2.2621e-03, -1.0559e-01, -2.9616e-03,
        -4.3155e-02, -3.2113e-02, -1.2253e-02,  1.6572e-02,  1.0876e-03,
        -7.2101e-03, -2.9816e-03, -7.7318e-03,  7.6050e-04,  2.3486e-02,
         1.1236e-02, -4.7588e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1882,  0.0998,  0.0419,  0.1256, -0.0170,  0.0916,  0.1142,  0.1010,
         0.0764,  0.0020, -0.1421,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1411,  0.2838,  0.0003,  0.0858, -0.0285, -0.0188, -0.0107,  0.0562,
         0.0158,  0.0531,  0.0173,  0.0036,  0.0805,  0.0115, -0.0362,  0.0418,
         0.0768, -0.0381,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0231, -0.0787, -0.3374,  0.0385,  0.0372, -0.0384, -0.2237, -0.0440,
        -0.0520, -0.0122,  0.0255,  0.0093, -0.0329, -0.0205, -0.0268,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1334, -0.1661,  0.0160, -0.0095,  0.0019, -0.0942, -0.2187, -0.0408,
        -0.0206, -0.0807, -0.1158,  0.0112,  0.0053,  0.0326, -0.0531,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0842,  0.0196, -0.0184,  0.0880,  0.0347, -0.0227, -0.0446, -0.0279,
        -0.0350, -0.0078,  0.0145, -0.0530, -0.1214, -0.0141,  0.0053, -0.0502,
        -0.0439,  0.0310,  0.0027,  0.0101, -0.0832, -0.0939,  0.0096,  0.0283,
        -0.0162,  0.0399,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1789,  0.0587,  0.0044, -0.0210, -0.0166,  0.0497,  0.0658,  0.1891,
         0.2718,  0.0011,  0.0339, -0.0088,  0.0377, -0.0384,  0.0240,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0060, -0.2365, -0.0147, -0.0747, -0.1033, -0.0250, -0.0804, -0.0041,
        -0.0106, -0.1236,  0.0131, -0.0282, -0.0145, -0.0195, -0.0007, -0.0017,
        -0.0266,  0.0011, -0.0408,  0.0511, -0.0423,  0.0294,  0.0139, -0.0085,
         0.0079,  0.0219,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0271,  0.0029,  0.0383,  0.0131, -0.0064, -0.0195,  0.0019,  0.0172,
        -0.0207, -0.0407, -0.2554, -0.0286, -0.0441,  0.0422, -0.0891, -0.0708,
        -0.0279, -0.0223, -0.0046, -0.0498, -0.0226, -0.0193,  0.0210, -0.0574,
        -0.0234,  0.0337,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 1.8768e-02, -3.0074e-01,  7.7734e-03,  1.3380e-02, -4.2954e-03,
        -6.2361e-05, -3.3325e-02, -2.3044e-02, -2.5866e-03, -2.4176e-02,
        -2.9701e-03, -2.0410e-02, -5.6522e-02,  1.4965e-02, -3.3421e-02,
         1.1540e-02,  9.6357e-04,  1.9587e-02, -2.6849e-02, -2.8654e-02,
        -1.9758e-02, -3.3725e-02, -1.8509e-02, -7.7445e-02, -5.9914e-02,
         1.0354e-02, -2.4078e-02, -1.3095e-02, -2.2061e-02,  2.3611e-04,
         8.4731e-03, -2.0985e-02, -3.1584e-03,  2.1721e-03, -1.3727e-02,
         2.8281e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0063, -0.1262, -0.0547,  0.0096, -0.0610, -0.0253, -0.0812, -0.0826,
        -0.0042,  0.0116, -0.0190,  0.0041, -0.0143, -0.0002, -0.0074, -0.0029,
         0.0272, -0.0008,  0.0125,  0.0143, -0.0060, -0.0390, -0.0094, -0.0005,
        -0.0322, -0.0622,  0.0053, -0.0426, -0.0538, -0.0154,  0.0110, -0.0483,
        -0.0714,  0.0177, -0.0003, -0.0195,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0438, -0.0153, -0.0436, -0.0250, -0.0126, -0.3081, -0.1193, -0.0411,
         0.0477,  0.0438, -0.0137, -0.0422, -0.1191,  0.0085,  0.0119, -0.0224,
        -0.0100, -0.0184, -0.0532, -0.0006,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0037, -0.1210,  0.0029, -0.0214, -0.0046, -0.0182, -0.0058, -0.0383,
         0.0015, -0.0469, -0.0091, -0.0023, -0.0132, -0.0008, -0.0583, -0.0130,
        -0.0175, -0.0573, -0.0124, -0.0038,  0.0104,  0.0080,  0.0031, -0.0273,
        -0.0145, -0.0023, -0.0020, -0.0105, -0.0133, -0.0351, -0.0851, -0.0635,
        -0.0426, -0.1067,  0.0096, -0.0257, -0.0460, -0.0424,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0359, -0.1859, -0.0550, -0.0667, -0.0048,  0.0113, -0.0291,  0.0119,
        -0.0337, -0.0141, -0.0558, -0.0474, -0.0047,  0.0155,  0.0068, -0.0449,
         0.0130, -0.0356, -0.0054,  0.0092,  0.0108,  0.0103,  0.0134, -0.0049,
        -0.0121, -0.0015, -0.0063, -0.0237, -0.0773,  0.0038, -0.0315,  0.0014,
         0.0187, -0.0032, -0.0175,  0.0058, -0.0028,  0.0122, -0.0006, -0.0221,
        -0.0334], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0204, -0.3632,  0.0183, -0.0384,  0.0373, -0.0496,  0.0111, -0.0410,
        -0.0287, -0.0209, -0.0122, -0.0552, -0.0697, -0.0263,  0.0188, -0.0476,
        -0.0006, -0.0656,  0.0091, -0.0180, -0.0119,  0.0361,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0248,  0.0031,  0.0046,  0.0013, -0.0102, -0.0390,  0.0013, -0.0122,
        -0.0089,  0.0087,  0.0043, -0.0594, -0.0083, -0.0181,  0.0179, -0.0852,
        -0.0017,  0.0005, -0.0317, -0.0412, -0.1206,  0.0235,  0.0554, -0.0494,
         0.0057,  0.0222, -0.0086, -0.0386, -0.1144, -0.0335, -0.0482, -0.0485,
        -0.0098, -0.0173,  0.0218,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1350,  0.0236,  0.2512,  0.0471,  0.0704,  0.0039, -0.0020, -0.0197,
        -0.0042,  0.0121,  0.0149,  0.0033,  0.0198,  0.0012,  0.0085,  0.0482,
        -0.0331, -0.0093,  0.0487, -0.0046,  0.0477, -0.0089,  0.0087,  0.0300,
         0.0109, -0.0033,  0.0486,  0.0213, -0.0598,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0828, -0.3019, -0.1011, -0.0448,  0.0036, -0.0684, -0.0153,  0.0038,
         0.0149, -0.0435, -0.0154, -0.0270,  0.0044,  0.0100,  0.0047, -0.0285,
        -0.0369, -0.0103, -0.0244,  0.0197,  0.0023,  0.0009,  0.0296,  0.0520,
         0.0538,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1426,  0.0090,  0.0152, -0.0207, -0.1633, -0.0130, -0.0165, -0.0985,
        -0.0980,  0.0131,  0.0208,  0.0111,  0.0014, -0.0138, -0.0672, -0.1513,
         0.0190, -0.0339,  0.0154,  0.0108, -0.0145, -0.0230,  0.0279,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0637, -0.0129, -0.0049,  0.0001,  0.0082,  0.0258,  0.0166,  0.0087,
        -0.0061,  0.0075,  0.0036,  0.0013, -0.0225, -0.0874,  0.0008, -0.0313,
        -0.0114, -0.0460, -0.0103, -0.0141, -0.0045, -0.0061, -0.0154,  0.0101,
        -0.0626, -0.0108, -0.1178, -0.0422, -0.1003, -0.0229, -0.0371, -0.0300,
        -0.0362, -0.0203, -0.0387, -0.0037, -0.0219,  0.0081, -0.0279,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0200,  0.0628,  0.0098, -0.2272, -0.2796,  0.0242,  0.0262, -0.1193,
        -0.1183,  0.0389, -0.0470,  0.0268,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-2.2366e-02,  5.2521e-02,  2.6064e-01, -7.1452e-03,  2.2701e-02,
         7.3042e-02,  4.1801e-02, -2.9179e-02,  3.3987e-03,  2.3558e-02,
        -4.5853e-03, -2.7718e-02,  7.3167e-03, -1.1612e-02,  4.0832e-03,
        -1.4892e-02,  1.0497e-02,  5.1150e-02, -1.4104e-02,  2.8030e-02,
         2.6138e-02,  3.6141e-02,  3.5444e-03,  6.4088e-03,  6.0177e-03,
         3.9960e-03,  5.9159e-03,  1.4237e-02,  9.6277e-03, -5.1347e-03,
        -3.1241e-03, -5.6705e-03,  4.4022e-02,  9.4748e-05,  1.0519e-04,
         2.6885e-03,  1.3989e-02,  4.9216e-04,  4.7434e-03, -2.1661e-03,
        -1.4538e-02, -1.1419e-02, -1.7600e-02,  1.5060e-02, -3.6786e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.7027e-05,  4.1683e-01,  1.3217e-02,  8.4628e-02, -3.6903e-02,
         3.8132e-02, -6.3720e-02,  4.4958e-02,  1.7806e-01, -1.8382e-02,
        -4.9069e-02, -5.6054e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0120,  0.5651,  0.1155,  0.1722,  0.0301, -0.0277,  0.0353, -0.0422,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0122, -0.1678, -0.0020, -0.0144, -0.0411,  0.0056, -0.0093, -0.0133,
         0.0011, -0.0557,  0.0128, -0.0081, -0.0073,  0.0038,  0.0081,  0.0111,
         0.0133, -0.0029,  0.0029, -0.0108,  0.0088,  0.0092, -0.0123,  0.0045,
        -0.0096,  0.0062, -0.0003, -0.0167, -0.0397, -0.0176, -0.0045, -0.0067,
         0.0097, -0.0002, -0.0057, -0.0010,  0.0009, -0.0053,  0.0050, -0.0005,
        -0.0114,  0.0012, -0.0023, -0.0078,  0.0074,  0.0556, -0.0811,  0.0238,
        -0.0506, -0.0732, -0.0369, -0.0029, -0.0215,  0.0074, -0.0104,  0.0025,
        -0.0043, -0.0034,  0.0171,  0.0053,  0.0058], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.0667e-02, -5.3271e-03, -2.1145e-04,  3.1615e-02,  1.2835e-01,
         2.5470e-03,  3.6638e-02, -9.1010e-03,  7.5320e-03, -1.4503e-02,
         1.7069e-02,  1.0732e-01, -9.2509e-03,  6.3604e-02,  7.0576e-02,
         8.6089e-05, -1.1046e-04,  9.3822e-03,  1.8634e-02,  2.1416e-02,
         5.0485e-02,  1.3078e-02,  4.2947e-02, -1.9993e-02,  5.0430e-02,
         2.3150e-02,  3.1588e-02,  1.6071e-02,  2.9227e-02,  2.7599e-02,
         2.0526e-02, -7.2636e-02,  3.8334e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0095,  0.0847,  0.2056,  0.0956,  0.0030,  0.0316,  0.0644,  0.0061,
         0.0050,  0.0216, -0.0094,  0.0036,  0.0287,  0.0093,  0.0182, -0.0231,
         0.0313,  0.0327,  0.1470, -0.0142, -0.0456,  0.0383, -0.0357, -0.0359,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0077, -0.1269, -0.0813, -0.0382, -0.0034, -0.0761, -0.0156, -0.0516,
        -0.0120, -0.0432, -0.0735, -0.0510, -0.1357,  0.0101,  0.0027,  0.0023,
        -0.0047, -0.0057,  0.0783,  0.0495, -0.0238,  0.0162,  0.0351,  0.0556,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0380, -0.0006, -0.0480, -0.1117, -0.0096, -0.0252, -0.0357, -0.0856,
         0.0109, -0.0266,  0.0109, -0.0132, -0.0057,  0.0044, -0.0704, -0.0536,
        -0.0634,  0.0167, -0.0082, -0.0501, -0.0474, -0.0050, -0.0341, -0.0615,
         0.0040, -0.0096, -0.0373, -0.0074, -0.0063,  0.0071, -0.0246, -0.0258,
         0.0212, -0.0200,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0561, -0.0184,  0.0189,  0.0074, -0.0766,  0.0077, -0.0116, -0.0082,
        -0.0186, -0.0255, -0.0194,  0.0425,  0.0317, -0.1328, -0.0381, -0.0386,
        -0.1149,  0.0213, -0.0505, -0.0153, -0.0605, -0.0445,  0.0219, -0.0635,
        -0.0554,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1273, -0.0320, -0.0249, -0.0024, -0.0320,  0.0359, -0.1099, -0.0646,
        -0.0159, -0.4370,  0.0243, -0.0301, -0.0055,  0.0093, -0.0034,  0.0455,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0351,  0.2731,  0.0135,  0.1156,  0.0674,  0.0283,  0.1595,  0.0322,
        -0.0198, -0.0072, -0.0089,  0.0234,  0.0284, -0.0171, -0.0135,  0.0004,
         0.0297,  0.0327,  0.0079, -0.0037, -0.0078,  0.0306,  0.0441,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1193,  0.0571,  0.0701,  0.1590,  0.3518,  0.0447,  0.0129,  0.0297,
         0.0421, -0.0056, -0.0118,  0.0060, -0.0258,  0.0641,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0030,  0.0544,  0.0652,  0.1506,  0.0248, -0.0445,  0.0372, -0.0194,
         0.0524,  0.0351,  0.0587,  0.0786,  0.0184,  0.0043, -0.0025,  0.0105,
        -0.0265,  0.0686,  0.0206,  0.0959,  0.0175, -0.0118, -0.0251, -0.0744,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1431, -0.0870,  0.2091, -0.0307,  0.1127,  0.0883,  0.2258, -0.0335,
         0.0063, -0.0250,  0.0385,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0689,  0.0235,  0.0203, -0.0022, -0.0303, -0.1029, -0.1185,  0.0341,
        -0.2734, -0.0591, -0.0246, -0.0022, -0.0129,  0.0616, -0.0443, -0.0092,
         0.0138, -0.0363,  0.0011,  0.0556, -0.0052,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1481, -0.0377,  0.0434,  0.0935,  0.0285,  0.0126, -0.0150, -0.2387,
         0.0050, -0.1909, -0.0192,  0.0242,  0.0024, -0.1407,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1165,  0.0561,  0.0590, -0.0168,  0.0089,  0.0197,  0.0067,  0.0076,
        -0.0038,  0.0550,  0.0641, -0.0353,  0.2755,  0.0649,  0.0223, -0.0257,
         0.0466,  0.0224,  0.0399,  0.0244,  0.0288,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0010, -0.3717, -0.0534, -0.0332,  0.1532, -0.0986,  0.0246, -0.0763,
        -0.0221, -0.0922, -0.0271,  0.0026,  0.0440,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0205, -0.0682,  0.0011, -0.0130, -0.0099, -0.0031, -0.0251, -0.0717,
        -0.0480, -0.0077, -0.0084,  0.0210,  0.0205,  0.0091,  0.0093, -0.0155,
         0.0068, -0.0095,  0.0108,  0.0259,  0.0016, -0.0018,  0.0018, -0.0010,
        -0.1517,  0.0101, -0.1150, -0.0609, -0.0604, -0.0265,  0.0070,  0.0071,
        -0.0155,  0.0015,  0.0088,  0.0083, -0.0338,  0.0002,  0.0063, -0.0074,
        -0.0213, -0.0052,  0.0004,  0.0228, -0.0184], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0200, -0.0621, -0.0920, -0.0404, -0.0104,  0.0077, -0.0180, -0.0432,
        -0.0061, -0.0059, -0.0479, -0.0014,  0.0181, -0.0144, -0.0222, -0.0173,
        -0.1457,  0.0107, -0.0002,  0.0067,  0.0026, -0.0372, -0.0051, -0.0619,
         0.0249,  0.0013, -0.0117, -0.0778, -0.1296, -0.0088,  0.0094, -0.0072,
        -0.0070, -0.0250,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0307,  0.1013,  0.0063, -0.0365, -0.0103,  0.0078,  0.0090, -0.0088,
         0.0371, -0.0124, -0.0062, -0.0064,  0.0431, -0.0064, -0.0060, -0.0155,
         0.0233,  0.0187, -0.0667, -0.2129, -0.2224,  0.0230,  0.0106, -0.0188,
         0.0009, -0.0586,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0256, -0.2897, -0.0216, -0.0785,  0.0389, -0.0404,  0.0025, -0.0389,
        -0.0163, -0.0231,  0.0150,  0.0016,  0.0094,  0.0180, -0.0293, -0.0988,
        -0.1019, -0.0054, -0.0442, -0.0119, -0.0339, -0.0325, -0.0228,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.6730e-02, -1.8350e-01, -3.8613e-02, -3.3683e-02,  3.3803e-03,
        -1.1858e-02, -1.9674e-02, -4.1115e-02, -8.8074e-02, -1.5670e-02,
         6.7702e-03,  1.8260e-02,  4.4372e-03, -8.9905e-03, -1.5619e-01,
         4.5710e-03, -1.1294e-02, -2.0386e-02, -4.3531e-02, -1.7033e-02,
         2.0124e-03, -2.0345e-03,  1.0512e-02,  2.8942e-03,  5.3801e-03,
         7.3960e-04, -5.4687e-02, -3.1250e-02, -1.6346e-02, -1.9974e-02,
         3.0985e-05,  1.9034e-02, -6.2557e-03,  7.8767e-03, -3.1988e-03,
         5.5400e-03, -1.4770e-02,  2.6065e-03,  4.6677e-03,  2.6429e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0818, -0.0040, -0.0316, -0.0303, -0.0581, -0.0037,  0.0061, -0.0499,
         0.0068, -0.0004, -0.0138, -0.0154, -0.0346, -0.0151, -0.0488, -0.0668,
        -0.0731, -0.0130, -0.0243, -0.0290,  0.0047, -0.0785, -0.0547,  0.0006,
        -0.0352, -0.0153, -0.0126, -0.0709, -0.0013, -0.0107, -0.0456,  0.0025,
        -0.0538, -0.0073,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0840, -0.0446, -0.0016, -0.0104, -0.0690, -0.0086,  0.0179, -0.0034,
         0.0088,  0.0019, -0.0068, -0.0229,  0.0200,  0.0062, -0.0675,  0.0214,
         0.0056, -0.0058, -0.0206,  0.0027, -0.0009,  0.0044, -0.0016,  0.0115,
        -0.0436, -0.0615, -0.0120, -0.0218, -0.0176,  0.0136,  0.0055, -0.0007,
         0.0095, -0.0227, -0.0368, -0.0120, -0.0276, -0.0602, -0.0154, -0.0528,
        -0.0377, -0.0245, -0.0293, -0.0099, -0.0014, -0.0035, -0.0024, -0.0017,
        -0.0066, -0.0021,  0.0070, -0.0127], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0285, -0.0112, -0.0046, -0.0141, -0.1382,  0.0049, -0.0189,  0.0546,
        -0.0079, -0.0274,  0.0069,  0.0151,  0.0018, -0.1648, -0.0160, -0.0342,
        -0.0944,  0.0215, -0.0532, -0.0034, -0.0357, -0.0078, -0.0245, -0.1131,
         0.0020,  0.0023, -0.0043, -0.0135, -0.0234, -0.0125,  0.0397,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0578, -0.1424,  0.0180,  0.0158,  0.0149,  0.0069, -0.0064,  0.0069,
        -0.0111,  0.0536, -0.0073, -0.0658,  0.0226, -0.0158, -0.0769, -0.0701,
         0.0228, -0.0560, -0.1073, -0.0998, -0.0234,  0.0161, -0.0542, -0.0021,
        -0.0261,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0018,  0.0486,  0.0828,  0.0444,  0.0775,  0.1034,  0.4527,  0.0952,
         0.0203, -0.0043,  0.0299,  0.0392,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.2760e-04, -3.4395e-01,  1.6687e-03, -3.8328e-02, -8.3989e-02,
        -2.5571e-02, -1.7843e-02, -1.2094e-02,  1.2363e-03,  1.0003e-03,
         9.4569e-04, -1.6297e-02, -5.6977e-02, -4.2616e-02, -2.4683e-02,
        -4.2844e-02, -1.5381e-02, -2.0268e-02, -2.7128e-05,  1.6541e-02,
        -3.5030e-02, -1.0569e-02,  3.3269e-03, -3.4422e-02,  2.0955e-02,
        -6.5630e-02, -1.5940e-02, -3.4839e-02, -1.6597e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0133, -0.3488,  0.0022, -0.0492, -0.0105, -0.1091, -0.2422,  0.1259,
        -0.0013, -0.0191,  0.0415, -0.0369,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0254, -0.3386,  0.0941, -0.0760,  0.0477, -0.1279, -0.1481,  0.0095,
        -0.0606, -0.0401,  0.0009, -0.0312,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0272,  0.0750, -0.0587, -0.0048, -0.0669,  0.0010,  0.0047, -0.0157,
        -0.2148, -0.0073, -0.0197, -0.2308, -0.0235, -0.0172, -0.0320, -0.1204,
        -0.0080,  0.0067, -0.0275, -0.0379,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.6343e-02, -2.7516e-01, -1.0274e-02, -7.7258e-02, -1.3222e-01,
        -1.5754e-02, -4.2865e-02, -1.0090e-02, -1.7973e-02, -1.6721e-03,
        -1.2910e-03, -4.6183e-02, -7.1785e-03, -3.1375e-03, -1.8522e-03,
         1.0034e-03, -1.4652e-02,  1.0913e-02,  2.6922e-03, -6.5745e-03,
        -1.3794e-02,  2.4626e-02, -1.0064e-02, -1.9848e-03, -1.4463e-02,
         1.9846e-02, -2.5508e-03, -3.8279e-02, -3.5466e-02,  7.2700e-03,
        -3.2699e-03, -6.8529e-03, -1.1412e-02, -5.7761e-02, -2.1288e-03,
        -1.4980e-02, -1.6694e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0752,  0.7022, -0.0277,  0.0183, -0.0442,  0.0392,  0.0418,  0.0301,
        -0.0212,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0705, -0.0061,  0.0191, -0.1348, -0.0392, -0.0617, -0.0837, -0.0828,
        -0.0098,  0.0014,  0.0041, -0.0220, -0.0620, -0.1008,  0.0111, -0.0081,
         0.0174,  0.0205, -0.0427, -0.0588,  0.0453,  0.0162,  0.0122,  0.0070,
         0.0167,  0.0262, -0.0197,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.0638e-04, -3.1246e-01,  1.2181e-02, -4.8858e-02, -1.7438e-02,
        -1.0383e-02, -7.1544e-04,  7.7595e-03, -6.4581e-03,  1.8088e-02,
         2.7871e-02, -7.9897e-02, -1.4542e-02, -1.5960e-03, -6.6535e-03,
        -6.5939e-03, -4.9476e-02,  7.0023e-03, -1.7169e-02,  8.1603e-03,
         1.6436e-03, -1.3020e-02, -4.5907e-02,  6.7580e-03, -1.7404e-02,
        -5.0069e-02, -4.8886e-03, -1.3646e-02, -1.0749e-01, -1.2028e-02,
        -1.0286e-05, -1.6823e-02,  2.8361e-02, -2.8246e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 7.3209e-02, -1.7928e-03, -1.2411e-02, -9.0860e-03, -2.0641e-02,
         7.5954e-03, -1.2246e-03, -1.1452e-03,  1.5023e-02,  2.9587e-02,
        -3.1223e-02, -9.2016e-02,  1.2821e-02, -1.9734e-03,  9.3906e-03,
        -3.9330e-02, -9.1432e-02, -2.3533e-02, -1.3839e-02,  3.8136e-03,
         1.6298e-02,  9.5847e-03, -2.1227e-02, -6.4814e-03,  1.0539e-02,
         7.2170e-03,  1.2097e-03,  9.3261e-05, -1.2516e-03, -4.4128e-03,
         2.3508e-03, -6.0212e-02, -1.0460e-01, -1.4444e-02, -2.3018e-02,
        -2.8776e-02, -5.8349e-02, -3.2892e-03,  1.1978e-02,  2.4574e-03,
        -1.1240e-02, -7.2328e-02,  9.7124e-03,  1.2304e-02, -1.5541e-02,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0782, -0.0119, -0.0274, -0.0151, -0.0726, -0.1783, -0.2599,  0.0133,
         0.0076, -0.1333, -0.0228, -0.0152, -0.1099,  0.0427, -0.0065, -0.0051,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0153, -0.0353,  0.0171,  0.0300,  0.0175,  0.0318, -0.0192,  0.0456,
         0.2013,  0.1733,  0.0263,  0.1210,  0.0167,  0.0540,  0.0438, -0.0126,
         0.0045,  0.0032, -0.0227,  0.0152,  0.0052, -0.0191, -0.0153, -0.0265,
         0.0087, -0.0180,  0.0007,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0003, -0.0076, -0.0685,  0.0023, -0.0090,  0.0171, -0.0170, -0.0031,
        -0.0219, -0.0129, -0.0026, -0.0108, -0.0435, -0.0139, -0.0044, -0.0101,
        -0.0306,  0.0123, -0.0112, -0.0442,  0.0220, -0.0281, -0.0210, -0.0767,
        -0.0057, -0.0016, -0.0327, -0.0817, -0.0146, -0.0879,  0.0140, -0.0581,
        -0.0348,  0.0180, -0.0200, -0.0026, -0.0106, -0.0396,  0.0088,  0.0282,
        -0.0057, -0.0311,  0.0021,  0.0025,  0.0056, -0.0033], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0794, -0.0173,  0.0007,  0.0157, -0.0049,  0.0122,  0.0042,  0.1070,
        -0.0315,  0.0161, -0.0327, -0.0464, -0.0095, -0.0007, -0.0111,  0.0177,
         0.0026,  0.0358,  0.1076,  0.1953,  0.0255, -0.0138, -0.0075,  0.0225,
         0.0014,  0.0415,  0.0592,  0.0391, -0.0231, -0.0175,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0858,  0.0888,  0.0455, -0.0368, -0.0422,  0.0277,  0.0346,  0.4722,
         0.0068,  0.1301, -0.0295,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.8328e-02, -3.8001e-01, -1.9270e-02, -4.1117e-02, -4.8529e-02,
        -1.2121e-02, -4.2355e-02, -1.2883e-01, -9.7571e-03, -3.5522e-02,
        -6.9714e-03,  5.2635e-05, -2.3219e-02,  5.0310e-04, -2.3196e-03,
        -9.5898e-02, -5.1507e-02, -4.3870e-02, -3.8829e-02, -9.8810e-04,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1412,  0.3387,  0.2053,  0.0814, -0.0374,  0.0253,  0.1133, -0.0171,
         0.0283, -0.0008,  0.0112,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0399, -0.2415, -0.0910, -0.1622,  0.0338,  0.0070, -0.0123, -0.0516,
        -0.0904, -0.0425, -0.0171, -0.0140, -0.0101, -0.0150, -0.0093, -0.0498,
         0.0009, -0.0267, -0.0125,  0.0106,  0.0618,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0797,  0.3665,  0.0762,  0.0298,  0.0235,  0.2626, -0.0744,  0.0030,
        -0.0059, -0.0054,  0.0067,  0.0066, -0.0079,  0.0008, -0.0162, -0.0163,
        -0.0093,  0.0090,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0014, -0.0950, -0.1847, -0.0066,  0.0021,  0.0036, -0.0036,  0.0056,
         0.0033, -0.0021,  0.0100,  0.0157,  0.0031, -0.0420, -0.0612, -0.0145,
        -0.0305, -0.0383, -0.0265, -0.0210, -0.0097, -0.0488, -0.1104,  0.0081,
        -0.0378, -0.0069, -0.0271, -0.0154, -0.0103, -0.0710, -0.0117, -0.0101,
        -0.0139,  0.0030, -0.0197, -0.0030, -0.0030, -0.0051,  0.0140,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0537, -0.3571, -0.0185, -0.0925, -0.0093, -0.0047,  0.0131, -0.0127,
        -0.0053, -0.0212, -0.0249, -0.0762,  0.0019,  0.0090, -0.0067, -0.0135,
        -0.0435, -0.0082, -0.0073, -0.0099,  0.0014, -0.0011,  0.0020,  0.0019,
        -0.0035,  0.0012, -0.0257, -0.0014, -0.0340, -0.0924, -0.0073, -0.0243,
         0.0114, -0.0031,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0149,  0.2450,  0.2340,  0.2665,  0.0335,  0.0290,  0.0568, -0.0066,
         0.1138,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.2552e-02, -1.8141e-01, -5.8084e-03, -5.2886e-02, -1.4662e-02,
        -9.1789e-02, -1.7523e-03, -3.3274e-02, -1.6501e-03, -5.9267e-04,
        -1.4925e-02, -9.6574e-03,  4.9076e-04, -1.2069e-02, -1.8448e-03,
        -6.9088e-02,  5.1465e-04, -1.5189e-01, -7.2585e-03, -2.0095e-02,
         5.1674e-03,  3.4335e-03,  1.0871e-02,  7.7572e-03, -4.5296e-02,
        -2.5879e-03,  5.0268e-03, -7.6856e-03, -1.7075e-05, -9.5733e-03,
        -2.4637e-02, -6.9247e-02,  6.7477e-03, -5.7858e-02,  1.5052e-02,
        -2.2343e-03,  2.9154e-03,  3.5383e-02,  2.3156e-03, -1.9847e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0569, -0.0860, -0.2658, -0.0161, -0.0285, -0.0136, -0.0024, -0.0072,
        -0.0192, -0.0601, -0.0007,  0.0069,  0.0020,  0.0028, -0.0049,  0.0095,
         0.0024, -0.0047, -0.0323,  0.0034, -0.0143,  0.0055, -0.0089, -0.0072,
        -0.0036, -0.0320, -0.0310, -0.0094, -0.0213, -0.0662, -0.0152,  0.0034,
         0.0105,  0.0014,  0.0011, -0.0012, -0.0149, -0.0061,  0.0024,  0.0036,
         0.0173, -0.0129,  0.0042,  0.0084,  0.0032, -0.0026,  0.0039, -0.0045,
         0.0090, -0.0032,  0.0311, -0.0154,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0312, -0.3008, -0.0393, -0.0212, -0.0616, -0.0188, -0.0703, -0.0164,
        -0.0385,  0.0067, -0.0309, -0.0282, -0.0710, -0.0331, -0.0124, -0.0662,
        -0.0360, -0.0149, -0.0559,  0.0091,  0.0020,  0.0353,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0489, -0.5027,  0.0278, -0.0301,  0.0219, -0.0147,  0.0178, -0.0329,
        -0.0089, -0.0291, -0.0141, -0.0736, -0.0113, -0.0251, -0.0030, -0.0859,
        -0.0059, -0.0078,  0.0344,  0.0013, -0.0027,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0072, -0.1152, -0.0352, -0.0262,  0.0072, -0.0051, -0.0199,  0.0183,
        -0.0195, -0.0078,  0.0008, -0.0048,  0.0035,  0.0011, -0.0011, -0.0031,
         0.0135, -0.0074, -0.0358, -0.0037, -0.0005, -0.0005, -0.0155,  0.0011,
         0.0031, -0.0136, -0.0382, -0.0470, -0.0131, -0.0372, -0.0682, -0.0388,
        -0.0026, -0.0451, -0.0225,  0.0041,  0.0020, -0.0030, -0.0049,  0.0069,
        -0.1003, -0.0023, -0.0060, -0.0057, -0.0461,  0.0022, -0.0309, -0.0033,
        -0.0016,  0.0109, -0.0259,  0.0027, -0.0076, -0.0266, -0.0238],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1026, -0.0015,  0.0275, -0.0238,  0.0047, -0.0188, -0.0652,  0.0053,
         0.0068,  0.0033, -0.0176,  0.0045, -0.0072, -0.0219, -0.0244, -0.0043,
        -0.0009, -0.0303, -0.0203, -0.0416, -0.0134, -0.0123, -0.0450, -0.0023,
         0.0221,  0.0122, -0.0295, -0.0868, -0.0072, -0.0056,  0.0011,  0.0001,
        -0.0068, -0.1066,  0.0119, -0.0083, -0.0095, -0.0422, -0.0359, -0.0223,
        -0.0029,  0.0011,  0.0259, -0.0296,  0.0271,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0481, -0.2380, -0.1209, -0.1144, -0.0125, -0.0112, -0.0350, -0.0345,
        -0.0847, -0.0131,  0.0013, -0.0132,  0.0222, -0.0487, -0.0465, -0.0045,
        -0.0143,  0.0112, -0.0139, -0.0090, -0.0435, -0.0134,  0.0199,  0.0260,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0521,  0.0323,  0.0059,  0.0084,  0.0006,  0.0041, -0.0308, -0.0496,
         0.0281,  0.0030, -0.0076,  0.0023, -0.0069, -0.0441, -0.0035,  0.0314,
        -0.0223, -0.0280, -0.0266, -0.0484,  0.0068, -0.0153, -0.0849, -0.0896,
        -0.0003, -0.0513, -0.0682, -0.0117, -0.0429, -0.0755, -0.0336, -0.0313,
        -0.0529,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0206, -0.3784, -0.0043, -0.1147, -0.0039, -0.0586, -0.0058, -0.0725,
        -0.1324, -0.0369,  0.0083, -0.0062,  0.0199, -0.0675, -0.0303, -0.0093,
        -0.0032, -0.0261, -0.0011,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.1555e-02,  9.0460e-05, -2.4620e-02, -2.3550e-03, -1.3284e-01,
         1.5223e-02, -1.0174e-01, -6.6438e-02, -1.8741e-02, -2.4779e-03,
         3.6431e-03,  3.2307e-02, -1.2646e-02,  4.6541e-02,  7.7870e-03,
        -1.6457e-02, -1.2165e-02, -3.7106e-02, -2.7365e-02, -1.4931e-02,
        -6.7952e-02, -2.6290e-01, -7.7874e-03, -1.6897e-02, -7.4313e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0845, -0.2773, -0.0818, -0.0049, -0.0306, -0.1262, -0.1303,  0.0468,
        -0.0363, -0.0319,  0.0032,  0.0105, -0.0598, -0.0103, -0.0175, -0.0447,
        -0.0034,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0498,  0.0212, -0.0417, -0.0004,  0.0144, -0.0104,  0.0172, -0.0210,
         0.0014, -0.1252, -0.0264, -0.0270,  0.0063, -0.0277,  0.0187,  0.0071,
        -0.0735, -0.0858, -0.0018, -0.0573, -0.0043, -0.0889, -0.1417,  0.0114,
        -0.0330, -0.0090, -0.0149,  0.0015, -0.0086,  0.0106,  0.0022,  0.0393,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0572, -0.1112, -0.1435, -0.0117, -0.0273, -0.0017, -0.0580, -0.1011,
         0.0029,  0.0154,  0.0075,  0.0200, -0.0500, -0.0237,  0.0021,  0.0237,
        -0.0026,  0.0022, -0.0782, -0.0224, -0.0032, -0.0017, -0.0287, -0.0945,
        -0.0273, -0.0378, -0.0175, -0.0268,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0131, -0.0843, -0.0005,  0.0224, -0.0012, -0.0939, -0.1621, -0.0541,
        -0.0243,  0.0023, -0.0015, -0.1629,  0.0021, -0.0358, -0.0159, -0.0442,
         0.0048, -0.0749, -0.0117,  0.0324, -0.0216, -0.0016, -0.0100, -0.0059,
        -0.0388,  0.0081, -0.0161,  0.0158, -0.0377,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0254, -0.0159, -0.0023, -0.0395, -0.1452,  0.0269, -0.0055,  0.0075,
        -0.0034,  0.0162,  0.0506,  0.0009, -0.0615, -0.1372,  0.0740, -0.0329,
        -0.0112, -0.0778, -0.0170, -0.0392,  0.0431,  0.0069, -0.0484, -0.0049,
        -0.0150, -0.0139, -0.0241, -0.0078, -0.0076, -0.0136,  0.0110,  0.0053,
         0.0082,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0440, -0.3112, -0.0050, -0.0815, -0.0480, -0.0331, -0.0148, -0.0135,
        -0.0214, -0.0546, -0.0608, -0.0030,  0.0111,  0.0099, -0.0047, -0.0117,
        -0.0287, -0.0076, -0.0616, -0.0337, -0.0870, -0.0317,  0.0063, -0.0149,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0464, -0.1820, -0.1153,  0.0102, -0.0374,  0.0212,  0.0034, -0.0312,
         0.0296, -0.0666, -0.0147, -0.0700, -0.0614, -0.0153,  0.0092,  0.0005,
         0.0010, -0.0216, -0.0052, -0.1189, -0.0300,  0.0244,  0.0845,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0427, -0.2491, -0.0232, -0.0127, -0.0084, -0.0699,  0.0197,  0.0079,
        -0.0028, -0.0048,  0.0019, -0.0061, -0.0260,  0.0094,  0.0033, -0.0233,
        -0.1152, -0.0352, -0.0033,  0.0069,  0.0094,  0.0133,  0.0141, -0.0047,
        -0.0388, -0.0086, -0.0044, -0.0009, -0.0121, -0.0444, -0.0206, -0.0102,
        -0.0252, -0.0063, -0.0404, -0.0388, -0.0223, -0.0015,  0.0121,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.1045e-02, -5.0474e-02, -1.0189e-03, -1.7315e-02, -1.7084e-01,
        -2.8908e-04,  4.9259e-03,  9.9940e-03,  2.0790e-02, -4.4341e-03,
         6.3221e-03,  1.6282e-02,  1.2131e-02, -1.1618e-03, -1.4475e-02,
        -3.1519e-02, -2.6049e-03, -1.1779e-03,  9.9643e-05, -9.7096e-03,
        -2.3131e-02,  3.6293e-04,  8.5812e-03, -8.0126e-03, -5.7111e-02,
        -5.8809e-02, -1.0824e-02, -2.7126e-02, -1.2048e-02, -8.5376e-02,
        -4.4858e-02, -2.3458e-03,  3.5341e-03,  1.2847e-03,  1.0820e-02,
        -1.0228e-02, -2.1069e-02, -9.3766e-02,  5.9549e-03,  1.0470e-03,
        -8.5967e-03,  1.1106e-02,  2.7406e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0561, -0.1441, -0.0020,  0.0059,  0.0019,  0.0079, -0.0151,  0.0049,
        -0.0592, -0.0052,  0.0234,  0.0006, -0.0198, -0.0010,  0.0005, -0.0135,
        -0.0179, -0.0576, -0.0008, -0.0068,  0.0093, -0.0032, -0.0110,  0.0011,
         0.0085,  0.0015,  0.0104,  0.0062,  0.0080, -0.0003,  0.0127, -0.0212,
        -0.0538,  0.0080,  0.0035, -0.0065, -0.0212, -0.0851, -0.0776,  0.0070,
        -0.0271, -0.0348, -0.0127, -0.0064, -0.0079, -0.0066,  0.0005, -0.0023,
         0.0021,  0.0075,  0.0068,  0.0085, -0.0110,  0.0060,  0.0097, -0.0159,
         0.0334], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0778, -0.2785, -0.0435, -0.0033, -0.0581,  0.0667, -0.0698, -0.1221,
         0.0084, -0.0447, -0.1239, -0.0320,  0.0558, -0.0154,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1000, -0.1527,  0.0362,  0.0542,  0.0072, -0.0177, -0.0367, -0.0095,
        -0.0138,  0.0012, -0.0525, -0.0525,  0.0216, -0.0049, -0.0422,  0.0034,
        -0.0039, -0.0197,  0.0573, -0.0575, -0.1362, -0.0179, -0.0340, -0.0033,
         0.0055, -0.0028,  0.0121, -0.0010,  0.0299, -0.0129,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0119, -0.1734,  0.0064, -0.3163, -0.2248, -0.1000, -0.0413, -0.1259,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([-0.0253, -0.1925, -0.0285, -0.0214, -0.0305, -0.0725,  0.0042,  0.0072,
         0.0050, -0.0173,  0.0162, -0.0062, -0.0526, -0.0686, -0.0156, -0.0074,
        -0.0091, -0.0013, -0.0020,  0.0024, -0.0188,  0.0042, -0.0117,  0.0170,
        -0.0607, -0.0004, -0.0501, -0.0014, -0.0563, -0.1011,  0.0252,  0.0290,
        -0.0383,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4947e-02, -2.1821e-01, -5.1131e-02, -4.8112e-02,  8.0567e-03,
        -4.3029e-02, -1.0033e-02, -1.8997e-02, -1.2719e-03, -1.9200e-02,
         4.0606e-04, -2.3646e-03, -1.3225e-02,  2.9058e-03,  1.6766e-02,
        -2.5033e-03, -1.2060e-03,  1.1659e-02, -6.9446e-03,  5.3515e-03,
        -5.9790e-03, -2.7307e-03,  8.6302e-04,  4.1911e-03,  3.0353e-03,
        -9.3640e-03, -1.3442e-02, -4.0918e-02,  3.1576e-03, -2.1008e-03,
         1.2509e-03, -1.7896e-02, -3.2460e-02, -6.1588e-02, -2.2712e-03,
        -3.7350e-03, -1.1364e-02, -4.8979e-03, -3.9434e-03,  4.7969e-03,
        -2.3166e-03, -1.9206e-03,  1.0223e-02, -3.6059e-02, -1.0334e-02,
        -1.3034e-03, -3.8194e-03, -5.4102e-03, -3.9639e-03,  3.2429e-03,
         8.1551e-03, -1.3281e-02,  2.4938e-03, -3.7003e-03, -7.0860e-03,
        -2.0850e-03, -3.1131e-03, -2.9508e-03, -5.3745e-03, -3.5558e-02,
        -7.8838e-03, -1.3781e-03, -6.5834e-02, -1.2734e-03,  5.0785e-05,
        -4.3292e-03, -1.9225e-03, -6.8901e-04,  7.0821e-04,  3.3810e-03,
         2.3855e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0780, -0.0235, -0.0165, -0.0002, -0.0239, -0.0107, -0.1805, -0.0195,
         0.0067,  0.0095, -0.0049, -0.0104, -0.0008, -0.0046,  0.0054, -0.0013,
        -0.0039,  0.0089, -0.0089, -0.0940, -0.0094, -0.0102,  0.0119, -0.0379,
         0.0321, -0.0384, -0.0191,  0.0158,  0.0009, -0.0052, -0.0100, -0.0293,
        -0.0705, -0.0503, -0.0239, -0.0261, -0.0368, -0.0027,  0.0052, -0.0029,
        -0.0134,  0.0358,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0272, -0.2039, -0.0011,  0.0073,  0.0078,  0.0003,  0.0095, -0.0180,
        -0.0279, -0.1100, -0.0395, -0.0294, -0.0828, -0.0042,  0.0022,  0.0013,
        -0.0013, -0.0420, -0.0931, -0.0107,  0.0098, -0.0105,  0.0059,  0.0017,
        -0.0021, -0.0534, -0.0225, -0.0176, -0.0917, -0.0066,  0.0013, -0.0049,
         0.0008, -0.0030, -0.0187, -0.0060, -0.0242,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0705, -0.4961, -0.0082, -0.0442,  0.0051,  0.0963,  0.0342,  0.0410,
         0.0216, -0.0073, -0.0233,  0.0178, -0.0863,  0.0188,  0.0291,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0672, -0.0316,  0.0098, -0.0495, -0.1700, -0.1325, -0.0206, -0.0510,
        -0.1429, -0.0163,  0.0045, -0.0951, -0.0468, -0.0170,  0.0289,  0.0015,
        -0.0087, -0.0062, -0.0086,  0.0912,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0584,  0.0198,  0.0041, -0.0159, -0.1408, -0.0270, -0.0130, -0.0127,
        -0.0105, -0.0134, -0.0316, -0.0581, -0.0273, -0.0158, -0.0266, -0.0240,
         0.0100, -0.0290,  0.0182, -0.0237, -0.0037,  0.0077, -0.0043,  0.0103,
         0.0048, -0.0425, -0.0773, -0.0064,  0.0044, -0.0373, -0.0694, -0.0116,
         0.0025, -0.0176, -0.0284, -0.0435,  0.0073, -0.0088, -0.0019, -0.0011,
         0.0008, -0.0103,  0.0181,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0068, -0.0124, -0.0020, -0.0015, -0.0223, -0.0232, -0.0193, -0.0279,
        -0.0020, -0.0074, -0.0036,  0.0055, -0.0061, -0.0693, -0.1160,  0.0043,
         0.0018,  0.0075, -0.0047, -0.0282, -0.0925, -0.0159,  0.0325,  0.0026,
        -0.0044, -0.0086, -0.0120, -0.0030, -0.1403, -0.0154, -0.0124, -0.0589,
        -0.0160, -0.0483, -0.0832, -0.0391,  0.0115, -0.0315,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0370, -0.2486, -0.1104, -0.1496, -0.0270, -0.0589,  0.0566, -0.0414,
        -0.0069, -0.0455, -0.0224, -0.0018, -0.0519, -0.0153,  0.0240,  0.0052,
        -0.0069, -0.0367,  0.0043,  0.0181,  0.0291,  0.0024,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0047, -0.1083, -0.0100, -0.0064, -0.0087, -0.0088,  0.0218, -0.0013,
        -0.0402, -0.1474, -0.0022,  0.0098, -0.0203, -0.0031, -0.0353, -0.0452,
        -0.0535, -0.0328, -0.0077, -0.0520, -0.0029,  0.0141, -0.0367, -0.0319,
         0.0042, -0.0057, -0.0093,  0.0123, -0.0060, -0.0381, -0.0039,  0.0239,
        -0.0081, -0.0069, -0.0264, -0.0604,  0.0037, -0.0109,  0.0611, -0.0143,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0268, -0.0062,  0.0012,  0.0088,  0.0131,  0.0027,  0.0132, -0.0009,
        -0.0414, -0.0067, -0.0048, -0.0063, -0.0013, -0.0999, -0.0874, -0.0219,
         0.0237,  0.0037, -0.0745, -0.0042, -0.0047,  0.0048,  0.0055,  0.0344,
         0.0028, -0.1107, -0.0116, -0.0909, -0.1339, -0.0059,  0.0265, -0.1195,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.5519e-02,  3.1143e-01, -2.4875e-03, -3.9576e-02,  2.0458e-02,
        -1.0403e-02, -1.2739e-05,  2.7077e-02,  1.0732e-01,  6.8735e-03,
         8.1497e-03,  4.0215e-02,  8.3077e-03,  2.4474e-03,  8.0782e-03,
         8.3182e-03,  3.5038e-02,  1.5915e-02, -1.0764e-02, -1.1416e-03,
         8.5726e-03,  6.3983e-03,  2.6604e-02,  8.0138e-02,  1.8500e-02,
         7.8062e-03,  2.1459e-02,  1.1721e-02,  1.8363e-03,  1.4425e-02,
         5.4614e-03, -2.3375e-03,  4.7330e-03, -6.3585e-02,  3.6896e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0373, -0.0361, -0.0926, -0.0542,  0.0033, -0.0081, -0.1910, -0.1503,
        -0.0570, -0.1013, -0.0013, -0.0556,  0.0186, -0.0666, -0.0206, -0.0101,
        -0.0049, -0.0179,  0.0058,  0.0614, -0.0059,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0014, -0.4057,  0.0090, -0.0636, -0.0226, -0.0519, -0.0251, -0.0767,
        -0.1037,  0.0185,  0.0043,  0.0126,  0.0090, -0.0130, -0.0342,  0.0241,
        -0.0138,  0.0563, -0.0542,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1174, -0.1026, -0.1001,  0.0084, -0.0243, -0.0242,  0.0079, -0.0051,
        -0.0192, -0.1080, -0.1778, -0.0068, -0.0423, -0.1448, -0.0106, -0.0414,
        -0.0016, -0.0361,  0.0175, -0.0038,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.3155e-02, -3.6878e-01, -2.7558e-04, -1.1132e-01,  3.0837e-02,
         1.0966e-02, -5.5330e-02, -4.7887e-02, -1.4832e-02,  1.5987e-02,
        -1.0519e-02, -1.4756e-03, -1.1574e-02, -4.7719e-02, -4.7364e-03,
        -3.3484e-02, -7.1474e-02, -6.3863e-03, -2.8317e-02, -3.6625e-02,
         2.2321e-02,  2.4942e-02, -2.1058e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0253, -0.1353, -0.1080,  0.0216,  0.0025, -0.0629,  0.0107, -0.0274,
        -0.0667, -0.0508, -0.0228, -0.0814, -0.0154, -0.1695,  0.0129, -0.0025,
         0.0073, -0.0088, -0.0271, -0.0122, -0.0649, -0.0137, -0.0283,  0.0159,
        -0.0062,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.2355e-02, -3.5674e-01, -2.5318e-02, -5.3713e-02,  1.4347e-02,
         1.5268e-02,  1.2145e-02, -8.3377e-03,  1.9962e-02,  4.0582e-03,
        -5.1652e-02,  1.2508e-02,  5.9382e-04, -4.2422e-02,  8.8974e-03,
         1.7257e-02, -8.4473e-03, -7.6984e-04,  2.3897e-02, -1.8715e-02,
        -7.4455e-02, -2.9403e-02, -3.5412e-02, -1.5851e-02, -2.0774e-02,
        -1.5360e-02,  1.6479e-02, -3.7692e-03,  1.6563e-02, -1.1951e-02,
        -4.1709e-03, -2.8429e-04, -2.0100e-02, -3.1477e-03,  2.4877e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0179,  0.0042, -0.0032, -0.0282, -0.0128, -0.0056, -0.0897, -0.0096,
        -0.0925, -0.0620, -0.0362,  0.0049, -0.0036, -0.0670, -0.0931, -0.2188,
         0.0206, -0.0796,  0.0181,  0.0234,  0.0592, -0.0499,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1553,  0.0482, -0.0249,  0.0389,  0.0293,  0.0777, -0.0025, -0.0003,
        -0.0151,  0.1990, -0.0559,  0.0153, -0.0229,  0.0274, -0.0743,  0.0943,
         0.0477,  0.0131,  0.0272,  0.0046,  0.0260,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.2491e-02,  3.6422e-01,  5.6202e-02, -2.9678e-02, -4.6158e-03,
         1.5402e-04,  8.5144e-02,  1.3523e-01,  3.9221e-03, -1.1169e-02,
         9.4378e-03,  1.9866e-02, -1.9111e-02,  9.1799e-02, -1.7622e-02,
         3.5523e-02, -5.2591e-02, -1.0350e-03, -3.0184e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.5812e-02,  3.2783e-02, -1.5271e-02,  1.8158e-02,  3.2775e-03,
        -5.0353e-02,  4.9468e-03, -5.0077e-03,  1.2323e-02, -1.0491e-02,
        -7.4410e-03, -2.5287e-02,  7.4159e-04, -1.4354e-02,  2.4496e-02,
         4.4169e-05, -3.0718e-02, -2.1766e-02, -1.3122e-03, -1.3272e-02,
        -4.4803e-03, -3.8004e-02, -4.2642e-02,  1.0513e-02, -6.0628e-02,
        -8.9706e-03, -3.1569e-03, -5.9437e-02, -2.1955e-02, -1.1869e-01,
        -8.3643e-03, -4.4954e-02, -4.3325e-02, -7.7611e-02,  3.4609e-03,
        -4.6443e-03,  9.3641e-03, -3.0744e-03,  6.1835e-03,  4.0840e-03,
         2.2951e-03,  9.4742e-03,  5.0638e-03,  1.7269e-03,  2.9621e-03,
        -2.9492e-02,  1.7591e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0136,  0.3778,  0.1192,  0.0146,  0.0071,  0.0085, -0.0035,  0.0233,
         0.0373,  0.0069, -0.0161, -0.0016, -0.0106,  0.0045, -0.0264, -0.0048,
         0.0665,  0.0153, -0.0070,  0.0048, -0.0158,  0.0109,  0.0455,  0.0787,
        -0.0011,  0.0052,  0.0211,  0.0105,  0.0052, -0.0019, -0.0015,  0.0053,
        -0.0044, -0.0235,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0332,  0.4443, -0.0087,  0.0121,  0.0335,  0.0346, -0.0172, -0.0185,
         0.0054,  0.0146,  0.0018,  0.0024,  0.0414,  0.0255,  0.0009,  0.0014,
         0.0345, -0.0242, -0.0042,  0.0800, -0.0008,  0.0167,  0.0215, -0.0108,
         0.0058,  0.0046,  0.0097,  0.0119,  0.0087,  0.0167,  0.0036,  0.0114,
         0.0039, -0.0354,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0052, -0.3621,  0.0192,  0.0263, -0.0394, -0.0977,  0.0515,  0.0132,
        -0.0236, -0.1294, -0.0161,  0.0200, -0.0363, -0.1087, -0.0134, -0.0190,
         0.0050,  0.0138,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1147,  0.0227,  0.0110, -0.0316, -0.0045,  0.0071, -0.0097,  0.0065,
        -0.1483, -0.2250,  0.0424,  0.0099, -0.0315, -0.0245,  0.0134, -0.0083,
        -0.0134, -0.0178,  0.0169, -0.0921, -0.0023,  0.0213, -0.0149,  0.0016,
        -0.0041,  0.0188,  0.0858,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0538, -0.0022,  0.0020, -0.0095, -0.0142,  0.0108, -0.0050, -0.0301,
        -0.1783, -0.0183,  0.0064, -0.0406, -0.2568, -0.0570,  0.0026,  0.0028,
        -0.0105, -0.0315,  0.0262, -0.1307, -0.0154, -0.0051, -0.0260, -0.0199,
        -0.0345,  0.0098,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0590, -0.3061,  0.0241, -0.0668, -0.0911, -0.0032, -0.0005, -0.0266,
        -0.0483, -0.0406, -0.0627, -0.0472, -0.0127,  0.0121, -0.0167, -0.0608,
        -0.0153, -0.0197, -0.0143, -0.0164,  0.0558,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0607, -0.2646, -0.1201, -0.1537, -0.0259, -0.0070, -0.1602,  0.0056,
        -0.0157, -0.0264, -0.0128, -0.0452,  0.0510,  0.0171, -0.0338,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.1432e-02,  1.4284e-02, -1.1335e-04, -7.4938e-03,  1.3864e-02,
        -3.1246e-02,  1.2393e-02, -5.1167e-02, -1.6608e-02, -4.7795e-02,
        -9.7110e-03, -3.9257e-02, -3.6496e-01, -1.4659e-01, -4.6811e-02,
        -5.6631e-02,  7.3007e-02,  1.6641e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0397, -0.2637, -0.1524, -0.1477, -0.0071, -0.0333,  0.0047, -0.0278,
        -0.0056, -0.0655, -0.0846, -0.0410, -0.0054, -0.0389, -0.0158,  0.0364,
        -0.0108,  0.0149, -0.0048,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0178, -0.1348, -0.0950, -0.0376, -0.0031, -0.0029,  0.0495,  0.0044,
        -0.0003,  0.0021, -0.0031, -0.0009, -0.0052,  0.0088,  0.0122,  0.0041,
        -0.0297,  0.0044,  0.0212,  0.0337, -0.0913, -0.0149, -0.0463, -0.0038,
        -0.0667,  0.0102, -0.0352,  0.0032,  0.0034, -0.0262,  0.0027, -0.0097,
        -0.0515, -0.0051, -0.0286, -0.0018, -0.0493,  0.0116, -0.0101,  0.0019,
         0.0022, -0.0059, -0.0039, -0.0439], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0386,  0.2350,  0.0007,  0.0512,  0.0256, -0.0029,  0.0460,  0.0214,
         0.1230,  0.0045,  0.0280, -0.0206, -0.0083,  0.0024, -0.0137, -0.0073,
         0.0071,  0.0863, -0.0133,  0.0223,  0.0419, -0.0030,  0.0006, -0.0016,
        -0.0207,  0.0005,  0.0092,  0.0619, -0.0024,  0.0118,  0.0139,  0.0129,
         0.0082, -0.0229, -0.0303,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.3370e-02, -3.7722e-01, -8.9799e-03, -5.6146e-02, -2.9228e-02,
        -1.7829e-02, -1.1019e-01, -3.3948e-02, -1.3520e-02, -9.4870e-03,
        -2.2894e-04, -2.0960e-02, -6.2194e-02, -1.8084e-04, -2.8291e-03,
        -8.4223e-02, -1.2346e-02,  2.7726e-03,  2.1422e-02,  5.3293e-03,
         8.3047e-03, -6.8883e-03,  1.7620e-02,  3.4780e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0517, -0.2725,  0.0129, -0.0316, -0.0230, -0.0206,  0.0285, -0.0168,
        -0.0848, -0.0109, -0.0015, -0.0008,  0.0215,  0.0077, -0.0530, -0.0108,
         0.0146, -0.0136, -0.0589, -0.0097,  0.0023, -0.0501,  0.0004, -0.0253,
        -0.0086, -0.0330, -0.0322, -0.0166, -0.0121, -0.0126, -0.0260, -0.0027,
         0.0027,  0.0300,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0102,  0.0141,  0.0034, -0.0234,  0.0221,  0.0161, -0.0896,  0.0089,
         0.0019,  0.0235, -0.0138, -0.0483, -0.1515, -0.0039, -0.1093, -0.1529,
        -0.0107, -0.0028,  0.0062,  0.0138,  0.0213,  0.0062,  0.0182, -0.0670,
        -0.0443, -0.0052, -0.0631, -0.0408,  0.0009, -0.0064,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0004, -0.0869,  0.0649,  0.2308,  0.0386,  0.0522,  0.0243,  0.0782,
         0.0656,  0.2218, -0.0322, -0.0467,  0.0575,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1893,  0.0120, -0.0197,  0.0505, -0.0247, -0.0336,  0.0323, -0.0727,
        -0.1152,  0.0168,  0.0132, -0.1275, -0.0140, -0.0562, -0.0812,  0.0780,
        -0.0263,  0.0368,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0954, -0.3218, -0.3103, -0.0418, -0.0803,  0.0049, -0.1349, -0.0106,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.3967e-03, -1.2763e-01,  9.1161e-03, -9.6662e-03, -1.3833e-02,
        -2.0688e-02, -5.0039e-03, -7.8499e-03,  1.0508e-02,  3.2216e-03,
         9.0304e-03,  1.5885e-02, -1.2587e-02,  5.7729e-04,  1.1611e-03,
        -8.8235e-03,  2.5379e-03,  8.6558e-03,  6.9903e-03, -1.4401e-03,
         4.2437e-03,  1.8999e-02, -2.3970e-02, -4.1306e-02,  6.6860e-03,
        -4.9151e-03,  1.1626e-05, -1.8960e-02, -1.1006e-01, -4.9298e-02,
        -3.0603e-02, -3.2423e-02, -1.7383e-02, -1.0142e-01,  3.2127e-03,
        -7.6680e-02, -5.7778e-02, -8.3127e-03,  1.3087e-02, -2.4935e-03,
        -1.2443e-02, -1.9926e-02, -2.3855e-02,  2.5297e-03,  3.4267e-03,
         3.4752e-02,  4.6258e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0252,  0.0281, -0.0199,  0.0969,  0.1638, -0.1460, -0.0128, -0.1367,
        -0.2292,  0.1112, -0.0303,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0814, -0.2842, -0.0070, -0.0579, -0.0137,  0.0003, -0.0028, -0.1752,
        -0.0473, -0.0967, -0.0207, -0.0163, -0.1090,  0.0060,  0.0109,  0.0016,
        -0.0006, -0.0684,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0008, -0.0901, -0.3239, -0.0757,  0.0227,  0.0140, -0.1651,  0.0467,
        -0.0994,  0.0025,  0.0009, -0.0124, -0.0216,  0.0464,  0.0778,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0767,  0.3621,  0.0525,  0.0101, -0.0093,  0.1176,  0.1229,  0.0122,
         0.0325,  0.0485,  0.0960, -0.0141, -0.0233, -0.0177,  0.0044,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1206,  0.0041,  0.0090,  0.0345,  0.0062,  0.0046, -0.0229, -0.0089,
        -0.0410, -0.0447, -0.0156, -0.0914, -0.0874, -0.0231, -0.0092, -0.0255,
        -0.0823,  0.0311, -0.0259,  0.0024, -0.1066, -0.1631,  0.0078,  0.0047,
        -0.0205,  0.0068,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0143,  0.0408,  0.0372,  0.0242, -0.0680, -0.0148, -0.0062,  0.2076,
         0.3148,  0.0412, -0.0211, -0.0064,  0.0374, -0.0555,  0.1105,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0789, -0.2893, -0.0163, -0.0037, -0.0745, -0.0279, -0.0416,  0.0050,
        -0.0127, -0.0989,  0.0413,  0.0043, -0.0147, -0.0082,  0.0245, -0.0148,
        -0.0122, -0.0089, -0.0329,  0.0258, -0.0239,  0.0108,  0.0312,  0.0163,
        -0.0462,  0.0351,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0020,  0.0399,  0.0239,  0.0158, -0.0104, -0.0380,  0.0119,  0.0072,
        -0.0547, -0.0807, -0.1920,  0.0015, -0.0173,  0.0432, -0.1118, -0.0734,
         0.0175, -0.0396, -0.0233, -0.0593,  0.0008, -0.0150, -0.0292, -0.0514,
         0.0054, -0.0349,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-2.7343e-02, -2.5609e-01,  9.1992e-03, -1.1595e-02, -7.7239e-03,
         1.3806e-03, -3.4332e-02, -5.1755e-02, -2.7534e-02, -4.3983e-02,
        -1.9371e-02, -3.8633e-02, -5.4429e-03,  6.4630e-04, -4.3865e-02,
        -6.7690e-03, -5.0688e-03, -2.5658e-03, -1.5674e-02, -1.1788e-02,
         8.3444e-03, -5.7860e-03,  1.8470e-02, -6.4139e-02, -6.6690e-02,
        -1.7572e-02, -3.9689e-02, -7.5715e-03,  6.8639e-03,  7.7073e-03,
         3.5713e-05, -6.4525e-02, -1.0726e-02,  9.0360e-03, -3.4481e-02,
         1.7608e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0187, -0.0721, -0.0449, -0.0285, -0.0893, -0.0309, -0.0668, -0.0846,
         0.0014,  0.0188, -0.0231, -0.0059, -0.0131,  0.0045, -0.0024, -0.0131,
        -0.0171, -0.0005,  0.0014,  0.0258,  0.0006, -0.0439, -0.0130, -0.0039,
        -0.0434, -0.0739,  0.0116, -0.0344, -0.0339, -0.0111,  0.0149, -0.0524,
        -0.0641,  0.0188, -0.0050,  0.0119,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0190, -0.1090, -0.0314, -0.0071,  0.0017, -0.1065, -0.0330, -0.0117,
        -0.0334, -0.0067, -0.0445, -0.0737, -0.2459, -0.0441, -0.0489, -0.0356,
        -0.0666, -0.0104, -0.0533, -0.0176,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.9794e-02, -8.5498e-02,  1.1548e-02, -3.7760e-02, -1.6874e-02,
        -3.3464e-02,  3.4469e-04, -2.9179e-02, -1.2213e-02, -3.9166e-02,
        -9.4770e-03,  1.4191e-03, -1.3500e-02, -1.0155e-02, -1.0044e-01,
        -3.6218e-02, -6.7727e-03, -5.4624e-02, -2.8499e-02, -5.8453e-03,
        -1.0066e-03, -6.8721e-03,  3.6559e-04,  2.1732e-03, -2.0128e-02,
         2.2233e-02,  7.9212e-03, -6.3014e-03,  3.4546e-05, -4.5893e-02,
        -6.3823e-02, -2.5078e-02, -3.0566e-02, -6.8962e-02,  3.9286e-02,
         3.8807e-02,  8.2240e-03,  1.9532e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0105, -0.2566, -0.0923, -0.0849, -0.0194,  0.0025, -0.0158,  0.0182,
        -0.0090, -0.0121, -0.0145, -0.0731, -0.0028,  0.0201, -0.0111, -0.0484,
         0.0119, -0.0424, -0.0016, -0.0056,  0.0075,  0.0135,  0.0087,  0.0092,
        -0.0016,  0.0006, -0.0074, -0.0256, -0.0808, -0.0009, -0.0154,  0.0018,
        -0.0006,  0.0015, -0.0235,  0.0075, -0.0132, -0.0062,  0.0050,  0.0161,
        -0.0006], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.9832e-02, -4.5995e-01, -3.1439e-02, -2.3732e-02, -5.0246e-03,
        -3.6308e-02, -5.0149e-03, -1.9654e-02, -2.4800e-02, -4.7628e-03,
        -1.7478e-02, -6.8671e-02, -4.5277e-02, -1.9799e-02,  3.2366e-02,
        -5.9178e-02,  1.1354e-02, -6.3024e-02,  4.4228e-03, -5.4346e-03,
         3.4235e-04,  2.2140e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.5883e-02, -3.2863e-03,  4.5938e-03, -4.8382e-03,  1.8025e-02,
        -1.9368e-02, -8.1522e-03,  1.0568e-02, -9.2241e-03, -6.2403e-05,
         3.2625e-03, -9.4089e-02, -5.7367e-03, -2.6098e-02, -2.4752e-02,
        -8.5340e-02,  6.6584e-03, -7.2627e-02,  1.6765e-02, -6.7840e-02,
        -2.7269e-02,  3.1873e-02,  1.0177e-02, -4.5325e-02,  1.5354e-03,
         2.1391e-03, -1.6588e-03, -3.5409e-02, -1.3224e-01,  2.3635e-02,
        -3.3417e-02, -4.3700e-02, -2.8090e-02, -1.3785e-02,  5.2572e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1218,  0.0361,  0.2156,  0.0563,  0.1388,  0.0156,  0.0011, -0.0097,
        -0.0206,  0.0129,  0.0326, -0.0130, -0.0072, -0.0060, -0.0044,  0.0032,
        -0.0319,  0.0006,  0.0689, -0.0191,  0.0403, -0.0039, -0.0084,  0.0198,
         0.0073,  0.0215,  0.0558, -0.0080,  0.0195,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0118,  0.3571,  0.0696,  0.0671, -0.0028,  0.0700,  0.0020,  0.0260,
         0.0130, -0.0112,  0.0236,  0.0381,  0.0394,  0.0287, -0.0087,  0.0656,
         0.0338,  0.0153,  0.0304, -0.0079,  0.0179, -0.0017,  0.0120, -0.0382,
         0.0080,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0504, -0.0052,  0.0224,  0.0040, -0.1204,  0.0060, -0.0458, -0.1218,
        -0.1264, -0.0148,  0.0069,  0.0333,  0.0044, -0.0399, -0.0549, -0.2111,
         0.0264, -0.0461, -0.0081, -0.0346, -0.0115, -0.0039, -0.0018,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0391,  0.0030,  0.0123,  0.0326,  0.0219,  0.0170, -0.0075,  0.0065,
         0.0038,  0.0016,  0.0021,  0.0027, -0.0867, -0.1639,  0.0028, -0.0294,
        -0.0138, -0.0318, -0.0161, -0.0044, -0.0077, -0.0086,  0.0061,  0.0026,
         0.0046, -0.0010, -0.0296,  0.0070, -0.0910, -0.0194, -0.0698, -0.0311,
        -0.0529, -0.0356, -0.0461,  0.0012, -0.0330, -0.0064, -0.0474,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0535,  0.0276,  0.0605,  0.2208,  0.2458, -0.0022,  0.0108,  0.0615,
         0.2355, -0.0531, -0.0021,  0.0266,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0133, -0.1200, -0.3078, -0.0087, -0.0136, -0.0445, -0.0465,  0.0080,
        -0.0094, -0.0519, -0.0082,  0.0122, -0.0038,  0.0017, -0.0079,  0.0172,
         0.0075, -0.0339,  0.0012,  0.0038, -0.0049, -0.0168,  0.0126, -0.0007,
        -0.0025, -0.0025, -0.0075, -0.0077, -0.0093, -0.0221,  0.0064,  0.0434,
        -0.0314, -0.0012, -0.0099, -0.0033, -0.0384, -0.0039,  0.0087, -0.0039,
         0.0076,  0.0018,  0.0102,  0.0062, -0.0162,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0543, -0.4188, -0.0498, -0.0856,  0.0320,  0.0284, -0.0061, -0.0625,
        -0.1266, -0.0426,  0.0135,  0.0798,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0241, -0.4877, -0.0826, -0.1132, -0.0671,  0.1174, -0.0844, -0.0235,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.0871e-02,  2.0745e-01, -1.0708e-02, -3.4081e-02,  5.1126e-02,
         9.9999e-03,  2.3541e-02,  8.1297e-03,  1.8168e-02,  4.2743e-02,
         2.7016e-02,  1.2514e-02,  2.2337e-04,  1.6125e-03, -7.5221e-04,
         1.7524e-03, -1.1879e-02,  2.9145e-04, -2.5648e-03,  1.6867e-03,
        -3.0651e-03, -7.3040e-03,  3.9535e-03,  1.4214e-03,  4.9220e-03,
        -2.6851e-03, -1.5775e-03,  6.1183e-03,  2.3994e-02,  2.0591e-03,
         7.5869e-03,  3.1283e-03, -2.2357e-03, -6.8038e-03, -4.2480e-03,
        -2.8413e-03, -5.0301e-03,  3.2662e-03,  2.4212e-05, -1.2950e-03,
         5.5462e-03,  6.7390e-03,  2.0390e-04,  1.7612e-02,  1.5400e-02,
         9.7540e-03,  8.0467e-02, -2.7452e-02,  5.8547e-02,  4.6273e-02,
         1.6206e-02,  7.6764e-04,  2.8557e-02,  1.4312e-03,  1.0243e-02,
         2.7358e-04,  2.1831e-02,  9.4064e-03,  2.4340e-03,  1.4489e-02,
        -1.5704e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0382, -0.0262,  0.0033, -0.0325, -0.0819, -0.0137, -0.0244, -0.0117,
         0.0384,  0.0057, -0.0321, -0.0619, -0.0101, -0.0518, -0.0658,  0.0064,
        -0.0041, -0.0185, -0.0381, -0.0279, -0.0735, -0.0620, -0.0285, -0.0136,
        -0.0419, -0.0165, -0.0063, -0.0168, -0.0256, -0.0356, -0.0348, -0.0226,
        -0.0298,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0456, -0.0576, -0.2152, -0.0603, -0.0077, -0.0046, -0.0272,  0.0215,
         0.0190,  0.0027,  0.0115,  0.0052, -0.0274,  0.0059, -0.0028,  0.0399,
        -0.0211, -0.0556, -0.2259,  0.0074,  0.0076, -0.0459,  0.0455, -0.0371,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0213, -0.0916, -0.1269, -0.0578, -0.0223, -0.0993,  0.0104, -0.0348,
         0.0180, -0.0681, -0.0855, -0.1002, -0.0962, -0.0070, -0.0079,  0.0026,
         0.0129,  0.0198,  0.0124,  0.0159, -0.0255, -0.0157,  0.0025,  0.0457,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0463, -0.0161, -0.0560, -0.0605,  0.0005,  0.0124,  0.0204, -0.0298,
         0.0121,  0.0057,  0.0058, -0.0110, -0.0051,  0.0047, -0.0621, -0.0993,
        -0.1094, -0.0128, -0.0078, -0.0267, -0.0930,  0.0052, -0.0363, -0.0975,
        -0.0020, -0.0105, -0.0411, -0.0157,  0.0177, -0.0036, -0.0159, -0.0113,
         0.0223, -0.0233,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0480,  0.0188, -0.0092,  0.0094, -0.0819,  0.0063, -0.0044,  0.0055,
         0.0016,  0.0216,  0.0064,  0.0242,  0.0460, -0.1061,  0.0028,  0.0364,
        -0.1568,  0.0194, -0.0462, -0.0154, -0.1003, -0.1428,  0.0383, -0.0147,
        -0.0380,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0006,  0.0172, -0.0249,  0.0319,  0.0108, -0.0128, -0.1643, -0.0979,
        -0.1149, -0.3103, -0.0089, -0.0652, -0.0465,  0.0120,  0.0271, -0.0548,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0110, -0.1986,  0.0105, -0.0771, -0.0427, -0.0521, -0.2226, -0.0206,
         0.0230,  0.0176, -0.0065, -0.0202, -0.0465, -0.0094,  0.0120, -0.0280,
        -0.0375, -0.0679, -0.0291, -0.0150,  0.0177,  0.0246, -0.0098,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0287,  0.0412,  0.0555,  0.1691,  0.3545, -0.0516, -0.0879, -0.0135,
         0.0155, -0.0270,  0.0449, -0.0364, -0.0305,  0.0437,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.0031, -0.0733, -0.0523, -0.0972, -0.0158,  0.0027, -0.0481,  0.0271,
        -0.0520, -0.0370, -0.0912, -0.0570, -0.0302,  0.0128, -0.0049,  0.0185,
        -0.0071, -0.1169, -0.0206, -0.1197, -0.0134, -0.0122, -0.0508, -0.0361,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0529,  0.0214, -0.1474,  0.0415, -0.2556,  0.0394, -0.2183, -0.0564,
        -0.0030,  0.0552, -0.1089,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1104,  0.0436,  0.0597,  0.0067,  0.0184, -0.0354, -0.1505,  0.0049,
        -0.1128, -0.0479, -0.0272,  0.0151, -0.0124, -0.0478, -0.1001,  0.0146,
         0.0024, -0.1277, -0.0085, -0.0157, -0.0381,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0240, -0.0092,  0.0028, -0.0293, -0.0429, -0.0112, -0.0311, -0.3519,
        -0.1110, -0.2359, -0.0313,  0.0044, -0.0369, -0.0782,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.4851e-05,  3.3789e-02,  3.2623e-02,  3.0198e-02,  5.4162e-02,
         1.9856e-02,  3.9071e-02, -6.1596e-02, -4.2917e-02,  4.0651e-02,
         2.5857e-02, -2.0572e-02,  3.6644e-01, -3.3806e-03,  5.5362e-02,
         4.9247e-02,  1.4519e-02,  1.4052e-02,  2.7743e-02, -7.0717e-03,
         6.0822e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0661,  0.3535, -0.0261,  0.0683, -0.0709,  0.0951, -0.0101,  0.1248,
        -0.0031,  0.0532,  0.0267,  0.0774, -0.0246,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0530, -0.0873, -0.0165, -0.0089, -0.0068, -0.0111, -0.0330, -0.0466,
        -0.0069,  0.0008, -0.0011,  0.0282, -0.0020,  0.0133, -0.0004,  0.0055,
         0.0058,  0.0033,  0.0034,  0.0439,  0.0026,  0.0214,  0.0165, -0.0149,
        -0.1376,  0.0010, -0.0942, -0.0710, -0.0439, -0.0096, -0.0150,  0.0337,
        -0.0056, -0.0027, -0.0097, -0.0076, -0.0151, -0.0017,  0.0031, -0.0053,
        -0.0163,  0.0039, -0.0065,  0.0588,  0.0246], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.7940e-02, -5.3016e-02, -6.9933e-02, -3.1356e-03, -4.0081e-03,
        -1.1521e-02, -6.9267e-04, -1.5495e-02,  7.6184e-03, -1.5749e-02,
        -1.0937e-01,  3.3690e-02, -1.1933e-02,  5.4974e-03, -5.8158e-02,
         1.5511e-02, -1.2921e-01,  1.8102e-02, -1.6149e-02,  1.5335e-02,
        -7.6637e-03, -1.0230e-01, -1.4674e-02, -1.0050e-01,  1.1828e-02,
         1.7342e-02,  6.8834e-05, -1.0768e-02, -4.9368e-02, -4.3685e-03,
         2.2147e-02,  7.3158e-03, -1.6156e-02, -1.3434e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.6007e-02,  3.0737e-02, -2.5564e-02, -3.9311e-02, -3.6324e-02,
         1.0116e-05, -3.1222e-02, -3.3256e-03,  4.6925e-02, -4.7739e-03,
        -1.3766e-02, -1.9105e-03, -1.4630e-04,  1.8498e-03,  5.4819e-03,
         5.7450e-03, -1.2152e-03,  7.2176e-03,  7.5202e-03, -1.8798e-01,
        -2.4182e-01, -6.8582e-02, -3.8816e-03,  2.9151e-02,  2.8117e-02,
        -8.1417e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0047, -0.2310, -0.0156, -0.0877, -0.0032,  0.0209,  0.0450, -0.0231,
        -0.0044, -0.0446,  0.0064, -0.0045, -0.0118, -0.0137, -0.0501, -0.1111,
        -0.1356,  0.0141, -0.0379, -0.0205, -0.0483, -0.0186, -0.0474,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0474, -0.2139, -0.0579, -0.0316, -0.0007,  0.0033, -0.0015, -0.0197,
        -0.0520, -0.0074,  0.0059,  0.0192,  0.0037,  0.0108, -0.1600,  0.0020,
        -0.0136, -0.0260, -0.0559,  0.0009, -0.0018,  0.0014,  0.0122, -0.0077,
         0.0089, -0.0037, -0.0440, -0.0728, -0.0034, -0.0018, -0.0058,  0.0152,
        -0.0036,  0.0160, -0.0025, -0.0069, -0.0154, -0.0036,  0.0291, -0.0110,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.7235e-02,  2.6922e-04, -1.1014e-02, -2.6028e-02, -1.1567e-01,
        -1.2985e-02, -8.6985e-03, -7.8403e-02,  1.2308e-02, -8.7631e-03,
        -2.8801e-03, -7.5230e-03, -2.7389e-02, -1.6339e-02, -1.7198e-02,
        -6.1237e-02, -5.1972e-02, -2.9171e-02, -5.8107e-02, -3.2466e-02,
         8.8677e-03, -4.4638e-02, -3.0739e-02, -3.1371e-03, -1.4249e-02,
        -3.9688e-02, -1.3906e-02, -5.8744e-02,  2.2852e-04, -8.7384e-05,
        -1.0489e-01, -1.5021e-02, -3.9855e-02, -2.9143e-04,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0551, -0.0192, -0.0379, -0.0114, -0.0516,  0.0107,  0.0241,  0.0276,
        -0.0064,  0.0075, -0.0084, -0.0213,  0.0141, -0.0175, -0.1247,  0.0157,
        -0.0054,  0.0027, -0.0462,  0.0042, -0.0137, -0.0024,  0.0048,  0.0156,
        -0.0324, -0.0766, -0.0060, -0.0148, -0.0090, -0.0042, -0.0439, -0.0093,
        -0.0022, -0.0136, -0.0433, -0.0073, -0.0207, -0.0444, -0.0145, -0.0228,
        -0.0206, -0.0019, -0.0213, -0.0025,  0.0051, -0.0017,  0.0016,  0.0080,
        -0.0074, -0.0030,  0.0060, -0.0076], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0499,  0.0063, -0.0550,  0.0061, -0.1260,  0.0009,  0.0163,  0.0226,
         0.0045, -0.0058,  0.0039,  0.0130, -0.0221, -0.0905,  0.0035, -0.0453,
        -0.1347, -0.0510, -0.0622,  0.0165, -0.0361, -0.0246, -0.0355, -0.0751,
         0.0158, -0.0037, -0.0031, -0.0080, -0.0246,  0.0016,  0.0360,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0832, -0.1532,  0.0104,  0.0096,  0.0136,  0.0085, -0.0341,  0.0031,
        -0.0050,  0.0257, -0.0116, -0.0699,  0.0126, -0.0745, -0.0724, -0.0375,
         0.0182, -0.0631, -0.0956, -0.0827, -0.0215, -0.0176, -0.0330,  0.0399,
        -0.0034,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0715, -0.0118,  0.0534, -0.0557,  0.0156, -0.2444, -0.3398, -0.0456,
        -0.0710, -0.0370,  0.0052,  0.0490,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0061, -0.3301, -0.0024, -0.0462, -0.0770, -0.0197, -0.0051, -0.0012,
        -0.0059, -0.0005,  0.0309, -0.0176, -0.0642, -0.0820, -0.0558, -0.0323,
        -0.0032, -0.0185,  0.0073,  0.0119, -0.0249,  0.0093,  0.0211,  0.0162,
         0.0181, -0.0496,  0.0052, -0.0108,  0.0269,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0346, -0.3875, -0.0032,  0.0508, -0.0183, -0.1250, -0.2038,  0.0430,
        -0.0323,  0.0304,  0.0545,  0.0167,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0109, -0.4376,  0.0035, -0.1539, -0.0010, -0.1447, -0.1375, -0.0241,
        -0.0468,  0.0056,  0.0088, -0.0257,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0170,  0.0312, -0.0870,  0.0094, -0.0952,  0.0322, -0.0359, -0.0314,
        -0.1965, -0.0254, -0.0158, -0.1280, -0.0173,  0.0176, -0.0507, -0.1074,
         0.0003, -0.0009, -0.0088, -0.0921,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0182, -0.2559, -0.0227, -0.0495, -0.1409, -0.0016, -0.0190,  0.0253,
        -0.0303, -0.0030, -0.0096, -0.0345, -0.0216, -0.0094, -0.0059,  0.0042,
        -0.0047, -0.0222, -0.0260, -0.0027, -0.0269,  0.0055,  0.0132,  0.0156,
        -0.0121,  0.0135, -0.0120, -0.0257, -0.0301,  0.0090, -0.0007,  0.0035,
        -0.0443, -0.0428, -0.0086, -0.0071, -0.0223,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0374,  0.7397, -0.0541, -0.0652, -0.0093,  0.0356, -0.0097,  0.0316,
        -0.0174,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0850,  0.0283,  0.0592, -0.1286, -0.0221,  0.0078, -0.0809, -0.0941,
        -0.0023,  0.0050,  0.0147, -0.0415, -0.1209, -0.1101, -0.0002,  0.0225,
         0.0069, -0.0123, -0.0494, -0.0464, -0.0060,  0.0011, -0.0074,  0.0025,
         0.0062,  0.0128, -0.0260,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0547,  0.3077, -0.0076,  0.0507,  0.0125,  0.0435, -0.0023, -0.0054,
         0.0131, -0.0268,  0.0050,  0.0686,  0.0176, -0.0078,  0.0035, -0.0018,
         0.0477, -0.0019, -0.0124, -0.0064,  0.0003,  0.0351,  0.0532, -0.0115,
         0.0021, -0.0370,  0.0007,  0.0232,  0.0918, -0.0005,  0.0093,  0.0151,
        -0.0216, -0.0015,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0020, -0.0043,  0.0083, -0.0059, -0.0597, -0.0136,  0.0071,  0.0126,
         0.0106,  0.0378,  0.0037, -0.0819,  0.0130,  0.0004,  0.0119, -0.0625,
        -0.1051, -0.0007, -0.0124,  0.0011,  0.0085,  0.0046, -0.0437,  0.0031,
         0.0061, -0.0107, -0.0017, -0.0043, -0.0152, -0.0024,  0.0098, -0.0822,
        -0.0910, -0.0116, -0.0085, -0.0347, -0.0571, -0.0267, -0.0005, -0.0071,
        -0.0204, -0.0588, -0.0088,  0.0026, -0.0253,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1130, -0.0084, -0.0475, -0.0191, -0.0630, -0.1018, -0.1944,  0.0014,
         0.0043, -0.2208, -0.0197, -0.0378, -0.0866, -0.0273, -0.0272, -0.0277,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0663,  0.0211, -0.0017, -0.0151,  0.0108, -0.0021,  0.0294, -0.0176,
        -0.2085, -0.1646, -0.0673, -0.1113,  0.0333, -0.0394, -0.0597,  0.0061,
        -0.0296, -0.0012,  0.0058,  0.0039,  0.0011,  0.0146, -0.0097, -0.0019,
         0.0166, -0.0366, -0.0247,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0126, -0.0262, -0.0609, -0.0023, -0.0263, -0.0020, -0.0112, -0.0051,
        -0.0169,  0.0059,  0.0017, -0.0011, -0.0254,  0.0013, -0.0162, -0.0161,
        -0.0210,  0.0110,  0.0255, -0.0104,  0.0133, -0.0294, -0.0136, -0.1106,
         0.0076,  0.0075, -0.0586, -0.0639, -0.0039, -0.0306,  0.0010, -0.0310,
        -0.0567, -0.0002,  0.0013, -0.0304, -0.0083, -0.0427,  0.0056,  0.0058,
        -0.0118, -0.0442,  0.0010, -0.0034,  0.0373,  0.0813], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0729,  0.0073,  0.0056,  0.0092, -0.0167, -0.0264, -0.0311, -0.0811,
        -0.0012,  0.0297,  0.0013,  0.0184, -0.0024, -0.0216, -0.0137,  0.0085,
        -0.0035, -0.0321, -0.0576, -0.3098, -0.0197, -0.0208, -0.0125,  0.0086,
         0.0055, -0.0071, -0.0603,  0.0523,  0.0136, -0.0496,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1092,  0.1477,  0.0858, -0.0235,  0.0732,  0.0542, -0.0239,  0.1279,
        -0.2297, -0.0386, -0.0864,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0185, -0.4196, -0.0178, -0.0424, -0.0439, -0.0244, -0.0539, -0.0613,
         0.0318, -0.0214,  0.0196, -0.0012, -0.0496, -0.0177, -0.0053, -0.0829,
        -0.0225, -0.0280, -0.0019,  0.0361,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0765, -0.2747, -0.2009, -0.1401,  0.0149, -0.0675, -0.1005,  0.0099,
        -0.0009, -0.0734, -0.0408,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.5891e-02, -2.6963e-01, -1.0782e-01, -1.4376e-01,  1.1947e-02,
         5.2278e-02, -5.3401e-03, -4.9297e-02, -6.5180e-02,  2.2289e-02,
         6.3241e-03, -2.6691e-02, -2.6242e-02,  1.8661e-03, -1.1668e-04,
        -3.3994e-02,  1.8266e-02, -8.0753e-03, -1.0831e-02, -1.2033e-02,
         7.2126e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0154, -0.1974, -0.1539, -0.0235,  0.0082, -0.3354,  0.0701,  0.0057,
        -0.0559, -0.0085, -0.0178, -0.0150, -0.0136,  0.0051, -0.0304, -0.0171,
        -0.0087, -0.0183,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.7228e-03, -9.0112e-02, -1.6772e-01, -1.8515e-03, -1.1392e-02,
         2.4665e-02,  2.5560e-02,  6.6206e-03, -8.1901e-03, -1.7881e-02,
         1.5874e-02,  1.7561e-02, -9.1276e-03, -6.1231e-02, -1.0331e-01,
         7.2119e-05, -8.5242e-03, -3.5796e-02, -1.0662e-02, -1.7776e-02,
        -6.4296e-03, -3.1474e-02, -8.8759e-02, -7.9598e-03, -2.5028e-02,
        -5.6321e-04, -2.3237e-02,  3.8014e-03, -2.0992e-02, -4.0642e-02,
         1.5052e-02, -4.7003e-03, -3.3335e-03,  9.8566e-03, -2.3786e-02,
         3.8859e-03, -6.7163e-03, -1.7141e-02, -2.4001e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0805, -0.2871, -0.0096, -0.0819, -0.0022, -0.0004, -0.0079, -0.0094,
        -0.0096, -0.0017, -0.0315, -0.0613,  0.0010,  0.0045,  0.0188, -0.0048,
        -0.0406, -0.0052,  0.0079, -0.0138, -0.0037,  0.0073,  0.0087,  0.0006,
         0.0020, -0.0058, -0.0167, -0.0191, -0.0426, -0.0885, -0.0071, -0.0163,
        -0.0348,  0.0670,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0348,  0.1643,  0.1482,  0.2083, -0.1069,  0.0183,  0.1046,  0.0756,
         0.1390,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0062, -0.2077, -0.0223, -0.0359, -0.0059, -0.1053,  0.0018, -0.0164,
        -0.0053,  0.0019, -0.0239, -0.0106,  0.0095, -0.0222, -0.0283, -0.0615,
        -0.0141, -0.1293, -0.0116, -0.0053,  0.0072,  0.0031, -0.0022,  0.0114,
        -0.0324, -0.0046, -0.0246, -0.0103, -0.0067,  0.0008, -0.0118, -0.0291,
         0.0086, -0.0237,  0.0180,  0.0091, -0.0003, -0.0097, -0.0025,  0.0589,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.8665e-02, -6.3441e-02, -1.5505e-01, -2.0204e-02, -7.0589e-02,
        -2.7563e-02, -1.0447e-02, -1.8680e-02, -5.3523e-02, -4.0333e-02,
        -1.1717e-03,  1.6623e-03,  7.3381e-04,  1.1462e-03, -1.7477e-02,
         5.4892e-03,  1.1485e-02,  2.2048e-02, -3.1824e-02, -9.4567e-03,
        -1.0869e-02,  7.5328e-03,  8.2867e-03, -4.8047e-04,  5.8322e-03,
        -1.8751e-02, -1.7153e-02,  1.4046e-02, -5.4401e-02, -8.8593e-02,
        -1.9395e-03, -1.7150e-02, -9.9883e-04,  2.6765e-03,  2.8244e-03,
        -1.3519e-03, -3.4428e-02, -4.8245e-02, -3.2789e-03, -5.8571e-03,
         7.1006e-03, -6.6342e-03,  6.3322e-03,  9.4781e-03,  1.4385e-03,
         3.3536e-03, -1.1221e-02, -2.4048e-03,  5.0097e-05,  3.6562e-03,
        -9.0037e-03, -1.3643e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0027, -0.2715,  0.0182, -0.0350, -0.0423, -0.0049, -0.0974, -0.0144,
        -0.0279,  0.0006, -0.0284, -0.0263, -0.0610, -0.0586, -0.0403, -0.0937,
        -0.0421, -0.0205, -0.0895, -0.0144, -0.0029,  0.0073,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1148, -0.4213, -0.0497, -0.0151, -0.0151, -0.0010,  0.0097, -0.0451,
        -0.0218, -0.0017,  0.0048, -0.0543, -0.0089, -0.0253,  0.0205, -0.0394,
        -0.0143,  0.0196,  0.0392,  0.0123,  0.0661,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0829, -0.1278, -0.0510, -0.0243, -0.0014, -0.0022, -0.0232,  0.0058,
        -0.0216, -0.0036,  0.0013,  0.0057,  0.0047,  0.0110,  0.0079,  0.0025,
         0.0179,  0.0098, -0.0164, -0.0015,  0.0012, -0.0026, -0.0068,  0.0007,
        -0.0026,  0.0152,  0.0027, -0.0177, -0.0031, -0.0165, -0.0536, -0.0731,
        -0.0181, -0.0500, -0.0169,  0.0013,  0.0018,  0.0005,  0.0054, -0.0138,
        -0.1308, -0.0079,  0.0030,  0.0105, -0.0297,  0.0041, -0.0189, -0.0124,
        -0.0052,  0.0120, -0.0042, -0.0023,  0.0093,  0.0053, -0.0181],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0205,  0.0005,  0.0228, -0.0195, -0.0100, -0.0200, -0.0724, -0.0184,
        -0.0066, -0.0046, -0.0026,  0.0143,  0.0024, -0.0087, -0.0475, -0.0048,
        -0.0042, -0.0135,  0.0053, -0.0721,  0.0039, -0.0185, -0.0464, -0.0041,
        -0.0029,  0.0182, -0.0179, -0.0955,  0.0114,  0.0050,  0.0133, -0.0122,
        -0.0173, -0.1055,  0.0015, -0.0343, -0.0023, -0.0452, -0.0493, -0.0484,
        -0.0110, -0.0006, -0.0373, -0.0153, -0.0116,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1071,  0.1902,  0.0708,  0.0709, -0.0146,  0.0105,  0.0209, -0.0119,
         0.0412,  0.0091,  0.0008,  0.0477,  0.0169,  0.0869,  0.0706,  0.0357,
         0.0119,  0.0474,  0.0210,  0.0261,  0.0402, -0.0084,  0.0313,  0.0079,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0402,  0.0271, -0.0009, -0.0057, -0.0177, -0.0111, -0.0662, -0.1129,
         0.0162,  0.0121, -0.0098, -0.0214, -0.0063, -0.0475,  0.0127,  0.0150,
        -0.0589, -0.0543, -0.0280, -0.0704, -0.0146, -0.0135, -0.0497, -0.1087,
        -0.0074, -0.0353, -0.0153, -0.0179, -0.0307, -0.0401, -0.0202,  0.0013,
        -0.0110,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0179,  0.4517,  0.0136,  0.1122, -0.0016,  0.0107, -0.0116, -0.0140,
         0.1456,  0.0018, -0.0121,  0.0046,  0.0190,  0.0662, -0.0327,  0.0144,
         0.0225, -0.0400, -0.0079,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1460, -0.0072,  0.0356, -0.0210, -0.0596, -0.0296, -0.0674, -0.0675,
        -0.1147, -0.0074,  0.0134,  0.0006,  0.0085,  0.0027, -0.0123,  0.0052,
        -0.0121, -0.1152, -0.0285, -0.0440, -0.0499, -0.1254, -0.0041, -0.0209,
         0.0012,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0721, -0.1907, -0.0977, -0.0141, -0.0094, -0.1365, -0.2074,  0.0216,
        -0.0318, -0.0367,  0.0261, -0.0058, -0.0645, -0.0159, -0.0133,  0.0280,
         0.0285,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0770,  0.0075, -0.0192, -0.0109, -0.0049,  0.0008, -0.0119, -0.0115,
        -0.0016, -0.0515, -0.0257, -0.0188,  0.0018, -0.0010, -0.0120, -0.0146,
        -0.0637, -0.1465,  0.0029, -0.0929, -0.0036, -0.1500, -0.1232,  0.0066,
        -0.0185, -0.0061,  0.0015,  0.0039,  0.0168,  0.0104,  0.0372,  0.0454,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0553, -0.1284, -0.1584, -0.0070,  0.0126, -0.0029, -0.0844, -0.1261,
         0.0018, -0.0008,  0.0061, -0.0144, -0.0494, -0.0061,  0.0030,  0.0051,
        -0.0201,  0.0021, -0.0574, -0.0043, -0.0390,  0.0066, -0.0582, -0.0915,
         0.0241, -0.0110,  0.0055, -0.0184,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0697,  0.0522, -0.0015, -0.0109,  0.0278,  0.0658,  0.1050,  0.0353,
         0.0120,  0.0330,  0.0091,  0.1216,  0.0049,  0.0107,  0.0099,  0.0390,
         0.0201,  0.1289,  0.0241, -0.0189,  0.0052,  0.0081,  0.0144,  0.0109,
         0.0424, -0.0043, -0.0326, -0.0094, -0.0725,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0271,  0.0262, -0.0076,  0.0170,  0.1404,  0.0060, -0.0165,  0.0007,
         0.0047, -0.0007, -0.0338,  0.0074,  0.0385,  0.1288, -0.0194,  0.0340,
        -0.0280,  0.0802,  0.0170,  0.0220, -0.0172,  0.0034,  0.0480,  0.0069,
         0.0090,  0.0738,  0.0425,  0.0082,  0.0119,  0.0299, -0.0064, -0.0508,
         0.0359,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0220, -0.3351, -0.0115, -0.0518, -0.0205, -0.0579,  0.0040, -0.0082,
        -0.0201, -0.0700, -0.0508, -0.0149,  0.0029,  0.0099, -0.0102,  0.0118,
        -0.0215,  0.0190, -0.0970,  0.0259, -0.0436,  0.0146,  0.0034,  0.0735,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0710, -0.1762, -0.1142, -0.0061, -0.0307, -0.0228, -0.0123, -0.0448,
         0.0298, -0.0635, -0.0089, -0.0716, -0.0334, -0.0257,  0.0096,  0.0037,
        -0.0070, -0.0374, -0.0137, -0.1126, -0.0561, -0.0074, -0.0415,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0076, -0.1743,  0.0066,  0.0123, -0.0099, -0.0324,  0.0203,  0.0160,
         0.0012, -0.0086,  0.0070, -0.0147, -0.0138,  0.0139, -0.0079, -0.0202,
        -0.0829, -0.0078, -0.0088,  0.0057, -0.0095, -0.0252, -0.0151, -0.0155,
        -0.0621, -0.0109, -0.0042, -0.0110, -0.0190, -0.0957, -0.0744, -0.0258,
        -0.0107, -0.0071, -0.0170, -0.0340,  0.0004, -0.0378,  0.0527,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0636,  0.1093, -0.0074,  0.0131,  0.0989, -0.0026, -0.0163, -0.0142,
         0.0234, -0.0082, -0.0078,  0.0034, -0.0061, -0.0035, -0.0072,  0.0573,
        -0.0146,  0.0039, -0.0006, -0.0111,  0.0449,  0.0061, -0.0225,  0.0152,
         0.0819,  0.0648,  0.0298,  0.0335,  0.0134,  0.0571, -0.0050, -0.0032,
        -0.0088, -0.0018,  0.0050, -0.0084,  0.0098,  0.0451, -0.0181,  0.0004,
         0.0015, -0.0441,  0.0071,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0469, -0.2474,  0.0049,  0.0043, -0.0220,  0.0068, -0.0348, -0.0078,
        -0.0729, -0.0165,  0.0081,  0.0033, -0.0324, -0.0008,  0.0028,  0.0067,
        -0.0494, -0.0530,  0.0034, -0.0132,  0.0059, -0.0072, -0.0152, -0.0013,
         0.0055, -0.0010,  0.0025,  0.0038,  0.0027, -0.0016,  0.0012, -0.0164,
        -0.0693,  0.0014, -0.0154,  0.0052, -0.0022, -0.0302, -0.0286, -0.0183,
        -0.0238, -0.0113, -0.0009,  0.0197, -0.0021,  0.0008, -0.0046, -0.0031,
         0.0046,  0.0020,  0.0155,  0.0011,  0.0051,  0.0068, -0.0081,  0.0178,
        -0.0006], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.2827e-01, -4.7098e-01,  2.3079e-04,  1.0739e-02, -4.1616e-02,
        -2.5148e-02, -7.6499e-02, -9.7186e-02, -1.5758e-02, -1.5759e-02,
        -6.7817e-02, -1.7887e-02, -4.3958e-03, -2.7718e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.8426e-02, -2.2077e-01,  4.7537e-03,  5.0084e-03,  2.9162e-02,
        -2.1211e-03,  9.3007e-03, -4.0977e-03, -3.1817e-02, -3.5322e-02,
        -7.4284e-02, -6.5550e-02,  3.0864e-02,  9.3451e-03, -4.4945e-02,
        -2.5748e-05, -1.3547e-02,  3.4330e-03, -7.9320e-03, -3.2143e-02,
        -1.4591e-01,  4.0429e-03, -4.6357e-02, -1.4779e-02,  2.7310e-02,
        -1.4928e-02, -4.2381e-03,  5.3094e-03, -9.0500e-03, -3.5235e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0897, -0.2247,  0.0925, -0.3631, -0.1354, -0.0090, -0.0247, -0.0610,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0169, -0.0713, -0.0209, -0.0295, -0.0440, -0.1164,  0.0081,  0.0138,
         0.0053, -0.0162, -0.0120, -0.0099, -0.0609, -0.0897, -0.0151, -0.0054,
        -0.0101, -0.0058,  0.0051,  0.0034,  0.0157,  0.0004,  0.0160,  0.0185,
        -0.0865,  0.0042, -0.0351,  0.0063, -0.0708, -0.1054, -0.0269,  0.0312,
        -0.0229,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.3024e-02,  1.9916e-01,  8.9566e-02,  3.3720e-02,  8.1363e-04,
         2.8644e-02,  2.0855e-02,  8.6995e-03,  4.6621e-03,  1.4651e-02,
        -1.7513e-03,  3.1689e-03,  1.7533e-03, -3.3896e-03, -2.4018e-03,
        -3.1027e-03, -5.8323e-03, -9.0009e-03,  3.8194e-03, -7.6007e-03,
         2.2035e-03, -4.2987e-03,  2.2203e-03, -3.2058e-03, -2.3778e-03,
         5.2416e-03, -6.4688e-04,  3.8453e-02, -6.0279e-03, -2.1698e-03,
         6.1587e-03,  1.8549e-02,  3.3555e-02,  4.4145e-02, -1.0701e-02,
         5.2411e-03,  1.4689e-02,  6.0929e-03,  1.9246e-04, -1.6836e-03,
        -1.3044e-03,  4.3908e-03, -5.5034e-03,  1.7652e-02, -1.3928e-02,
         3.1374e-03,  2.2536e-03,  9.6170e-03,  7.5241e-03,  5.9145e-03,
        -2.2956e-03,  2.0277e-02,  8.1234e-03,  1.3972e-02,  8.2825e-03,
        -4.3740e-03,  3.1378e-03,  1.1334e-04,  6.9951e-03,  1.3216e-02,
         1.4390e-02,  1.5789e-02,  4.9743e-02,  1.1869e-02, -6.3338e-03,
         1.0022e-02,  4.4622e-03, -1.0977e-02,  1.5951e-03,  2.0600e-02,
         3.8733e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0323, -0.0070, -0.0232, -0.0131, -0.0411, -0.0114, -0.1178, -0.0202,
        -0.0035,  0.0170, -0.0079,  0.0076,  0.0121,  0.0070, -0.0015,  0.0018,
        -0.0036, -0.0010, -0.0231, -0.1128, -0.0512, -0.0017, -0.0148, -0.0452,
         0.0013, -0.0420, -0.0265,  0.0045,  0.0037,  0.0001, -0.0070, -0.0193,
        -0.0459, -0.0693, -0.0176,  0.0097, -0.1050, -0.0119, -0.0034,  0.0069,
        -0.0055, -0.0424,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0110,  0.2618, -0.0323, -0.0010, -0.0038, -0.0078, -0.0210, -0.0200,
         0.0127,  0.0436,  0.0208,  0.0299,  0.0627,  0.0074, -0.0047,  0.0011,
         0.0097,  0.0521,  0.0791,  0.0032,  0.0145,  0.0038,  0.0066,  0.0042,
        -0.0072,  0.0637,  0.0297,  0.0229,  0.0819,  0.0113,  0.0113,  0.0036,
        -0.0058, -0.0007,  0.0044, -0.0123,  0.0304,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1121, -0.3922, -0.0258, -0.0843, -0.0283,  0.0282,  0.0341, -0.0237,
        -0.0348, -0.0022, -0.0203, -0.0020, -0.0460, -0.0394,  0.1267,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0369, -0.0843,  0.0412, -0.0384, -0.0901, -0.1903, -0.0203, -0.0623,
        -0.1569,  0.0013,  0.0246, -0.0372, -0.0505,  0.0076,  0.0816, -0.0186,
        -0.0234, -0.0048,  0.0241, -0.0055,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0523,  0.0229, -0.0166,  0.0164, -0.1232, -0.0212,  0.0016,  0.0054,
         0.0045, -0.0023, -0.0443, -0.0525, -0.0034,  0.0062, -0.0288, -0.0397,
         0.0022, -0.0247,  0.0167, -0.0241, -0.0089,  0.0007, -0.0102, -0.0038,
         0.0112, -0.0471, -0.0755, -0.0050,  0.0060, -0.0352, -0.0822, -0.0099,
        -0.0142, -0.0047, -0.0384, -0.0613, -0.0071, -0.0267, -0.0070,  0.0038,
         0.0202,  0.0104,  0.0019,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0813, -0.0008,  0.0011, -0.0060, -0.0174, -0.0121, -0.0375, -0.1059,
        -0.0209,  0.0145,  0.0212,  0.0015, -0.0115, -0.0321, -0.1004, -0.0084,
         0.0108,  0.0080, -0.0085, -0.0202, -0.0635, -0.0144,  0.0132,  0.0357,
        -0.0157,  0.0049, -0.0458, -0.0097, -0.0534,  0.0116, -0.0051, -0.0391,
        -0.0140, -0.0336, -0.0418, -0.0101,  0.0210, -0.0473,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0186, -0.3794, -0.0967, -0.1070,  0.0334, -0.0881, -0.0024,  0.0188,
        -0.0089, -0.0609, -0.0009, -0.0108, -0.0272, -0.0060,  0.0168, -0.0105,
        -0.0027, -0.0474, -0.0201,  0.0204,  0.0096, -0.0136,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.3373e-03, -9.4799e-02, -1.1756e-02, -5.3804e-03,  1.2249e-02,
        -4.4562e-03,  3.4341e-02,  2.0833e-02,  1.6132e-02, -8.4699e-02,
        -1.4515e-02,  7.7295e-04, -2.2610e-02, -3.0191e-04, -3.3425e-02,
        -2.5680e-02, -8.0252e-02, -2.0726e-02, -5.5283e-03, -1.0946e-01,
        -1.7965e-02, -9.2521e-06, -6.1382e-02, -2.6618e-02,  1.5492e-03,
        -1.4065e-04, -1.6194e-02, -5.7325e-03, -1.0146e-02, -4.1079e-02,
        -9.9560e-03,  1.1824e-02, -1.9046e-02,  2.3895e-03, -7.6214e-03,
        -1.0386e-01,  5.4014e-04, -2.3133e-02,  3.2475e-02, -7.0801e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0556, -0.0088, -0.0127,  0.0118, -0.0070, -0.0051,  0.0030,  0.0117,
        -0.0797,  0.0016, -0.0074, -0.0068, -0.0174, -0.1371, -0.1015, -0.0227,
        -0.0056, -0.0127, -0.0388, -0.0017,  0.0244,  0.0232,  0.0166, -0.0033,
        -0.0130, -0.0546, -0.0272, -0.0727, -0.1185, -0.0316,  0.0366, -0.0296,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0153, -0.3406, -0.0524,  0.0102, -0.0253,  0.0271, -0.0097, -0.0471,
        -0.0970, -0.0005,  0.0057, -0.0297,  0.0226, -0.0087,  0.0199, -0.0196,
        -0.0562, -0.0074,  0.0023,  0.0059, -0.0029, -0.0103,  0.0058, -0.0486,
         0.0011,  0.0020, -0.0266, -0.0103, -0.0026, -0.0218,  0.0009, -0.0064,
        -0.0111, -0.0228, -0.0236,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.1210, -0.0489, -0.1212, -0.1054,  0.0076, -0.0052, -0.1124, -0.1145,
        -0.0432, -0.1094,  0.0523, -0.0240, -0.0039, -0.0709, -0.0033, -0.0129,
        -0.0264,  0.0005, -0.0114, -0.0021,  0.0035,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0482, -0.4295, -0.0227, -0.0781, -0.0931, -0.0008, -0.0202, -0.0556,
        -0.0548,  0.0087,  0.0045,  0.0196, -0.0128, -0.0020, -0.0476, -0.0070,
        -0.0744, -0.0119,  0.0086,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0178,  0.0828,  0.1180,  0.0055,  0.0881,  0.0139, -0.0120,  0.0222,
        -0.0064,  0.1026,  0.1461,  0.0020,  0.0733,  0.1604,  0.0171,  0.0599,
        -0.0157,  0.0262, -0.0178, -0.0119,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.4181e-02, -2.5130e-01, -2.0278e-02, -1.3338e-01,  5.1567e-03,
         9.0414e-04, -1.2943e-01, -6.5903e-02, -4.8605e-03,  2.9599e-03,
         5.6991e-03, -2.0668e-02,  6.0961e-03, -4.5842e-02, -2.3608e-02,
        -3.3046e-02, -1.0231e-01, -8.7896e-05, -2.3173e-02, -5.1592e-02,
         8.9235e-03,  1.3158e-02, -7.4311e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0258, -0.2131, -0.0851, -0.0898,  0.0018, -0.0526, -0.0009, -0.0458,
        -0.0646, -0.1080, -0.0146, -0.0492, -0.0028, -0.0726, -0.0021, -0.0121,
        -0.0033, -0.0129, -0.0222, -0.0022, -0.0574, -0.0016, -0.0237, -0.0242,
         0.0113,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.3529e-02,  2.9480e-01,  1.8910e-03,  3.4152e-02,  1.8661e-02,
         1.2039e-02,  1.0731e-02,  3.0192e-02, -1.2730e-02,  7.7662e-03,
         3.8638e-02,  2.7590e-04,  2.3308e-03,  1.8683e-03, -1.2394e-02,
        -7.0501e-04,  6.9096e-03,  7.3565e-04, -1.6344e-02,  2.0656e-02,
         8.6785e-02,  6.5130e-02,  3.3115e-02,  2.0814e-02,  2.5687e-02,
         2.2942e-02,  1.9519e-02,  1.7013e-02,  3.1755e-02, -4.2369e-04,
         7.1177e-03, -1.1929e-02,  5.5063e-03,  2.3267e-02, -4.1647e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0372, -0.0057, -0.0083, -0.0732, -0.0121,  0.0041, -0.0810, -0.0235,
        -0.1643, -0.0097,  0.0186, -0.0256, -0.0297, -0.0751, -0.0546, -0.1459,
        -0.0291, -0.0777,  0.0119,  0.0497, -0.0159,  0.0468,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1194,  0.0970, -0.0406,  0.0494, -0.0003,  0.0581,  0.0156,  0.0013,
         0.0434,  0.1689,  0.0066,  0.0339,  0.0161, -0.0334,  0.0797,  0.1474,
         0.0453, -0.0151, -0.0062,  0.0021,  0.0203,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0101, -0.5038,  0.0157, -0.0161, -0.0124, -0.0073, -0.0760, -0.1126,
         0.0162,  0.0130, -0.0307, -0.0128,  0.0199, -0.0464, -0.0089, -0.0385,
         0.0209,  0.0202, -0.0185,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.1399e-03, -1.4222e-02, -3.1598e-02, -8.0674e-03,  1.0355e-04,
        -6.2225e-02, -1.8957e-03, -4.8454e-03,  3.2276e-03, -5.7643e-03,
        -1.0352e-02, -5.0380e-03, -1.4304e-02, -2.7948e-02,  9.1178e-03,
         4.9083e-04, -3.2414e-02, -5.5398e-02, -3.2093e-03, -1.9334e-02,
        -1.1466e-03, -1.1803e-02, -5.0736e-02,  1.0375e-02, -5.0796e-02,
        -8.8669e-03, -1.0235e-02, -3.6895e-02, -2.2467e-02, -1.2377e-01,
        -2.3558e-03, -3.3843e-02, -7.4844e-02, -8.6313e-02,  6.0111e-03,
        -9.9540e-03, -1.1498e-02,  5.7285e-03, -9.1024e-03, -2.7237e-02,
         3.7239e-03,  1.0093e-02,  6.7007e-03, -5.3961e-03, -1.8578e-03,
         2.9303e-02,  3.3255e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0414, -0.2473, -0.2011, -0.0063, -0.0038, -0.0239,  0.0069, -0.0118,
        -0.0350, -0.0044,  0.0108, -0.0175,  0.0011,  0.0039,  0.0152,  0.0172,
        -0.0525, -0.0033,  0.0046, -0.0175,  0.0055, -0.0118, -0.0606, -0.0862,
        -0.0144,  0.0197, -0.0134, -0.0004, -0.0140, -0.0037,  0.0146,  0.0122,
        -0.0005, -0.0178,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0756,  0.3064,  0.0077,  0.0011,  0.0055,  0.0542, -0.0129, -0.0143,
        -0.0023,  0.0133, -0.0082,  0.0021,  0.0388,  0.0842, -0.0073,  0.0120,
         0.0135,  0.0240,  0.0748,  0.1029, -0.0037,  0.0058,  0.0114, -0.0095,
         0.0044, -0.0074, -0.0101,  0.0066,  0.0085,  0.0194, -0.0053,  0.0010,
         0.0247, -0.0211,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0474,  0.3982, -0.0298, -0.0132,  0.0432,  0.0859,  0.0040,  0.0031,
        -0.0026,  0.0839,  0.0132,  0.0056,  0.0367,  0.1022,  0.0219,  0.0369,
        -0.0116,  0.0606,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.5557e-02,  1.3926e-02,  1.0440e-02,  1.9828e-02, -5.5481e-03,
        -1.0598e-02, -7.7798e-03,  8.9730e-03,  1.4658e-01,  2.0429e-01,
        -1.9947e-02,  1.1816e-03,  3.3879e-02,  2.4195e-02, -1.3064e-04,
         4.2525e-02,  2.0845e-02, -2.1620e-03,  1.8336e-02,  2.1660e-01,
        -9.8500e-03, -1.4112e-03,  2.7894e-02, -4.3419e-04,  1.1658e-02,
        -6.1151e-03,  4.9323e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0563, -0.0044,  0.0213,  0.0110, -0.0285,  0.0034,  0.0122, -0.0673,
        -0.1868,  0.0356, -0.0011,  0.0213, -0.1609, -0.0210, -0.0067, -0.0026,
        -0.0161, -0.0144, -0.0442, -0.1472,  0.0034,  0.0093, -0.0367,  0.0196,
         0.0251, -0.0438,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0094, -0.2388, -0.0174, -0.0731, -0.1266,  0.0021, -0.0182, -0.0202,
        -0.0714, -0.0539, -0.0722, -0.0467,  0.0071, -0.0180, -0.0216, -0.0660,
        -0.0263, -0.0093, -0.0258,  0.0084,  0.0677,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0570, -0.1810, -0.1564, -0.0996, -0.0698,  0.0042, -0.2359,  0.0026,
        -0.0101, -0.0043,  0.0002, -0.0427,  0.0511,  0.0017, -0.0834,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0558, -0.0254, -0.0029,  0.0155, -0.0538, -0.0204, -0.0315, -0.0576,
         0.0267, -0.0306,  0.0116, -0.0257, -0.3366, -0.1603,  0.0178,  0.0406,
        -0.0169, -0.0703,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0121, -0.3043, -0.1355, -0.1259,  0.0114, -0.0642,  0.0144, -0.0156,
        -0.0060, -0.0429, -0.0901, -0.0368,  0.0037, -0.0632, -0.0061,  0.0117,
         0.0041, -0.0188,  0.0331,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0308, -0.0922, -0.0603,  0.0040, -0.0222, -0.0220,  0.0299,  0.0083,
        -0.0110, -0.0016,  0.0039,  0.0005, -0.0219,  0.0056, -0.0095, -0.0043,
        -0.0911, -0.0055, -0.0079,  0.0079, -0.0792,  0.0119, -0.0351,  0.0097,
        -0.0724, -0.0060, -0.0291,  0.0020, -0.0176, -0.0262, -0.0095,  0.0348,
        -0.0478, -0.0062, -0.0247, -0.0067, -0.0474,  0.0221, -0.0039, -0.0055,
        -0.0015,  0.0021,  0.0540,  0.0046], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0403, -0.3131, -0.0317, -0.0568, -0.0509, -0.0162, -0.0186, -0.0069,
        -0.0539, -0.0142, -0.0241,  0.0163, -0.0036, -0.0071,  0.0087,  0.0111,
        -0.0019, -0.1106, -0.0064, -0.0363, -0.0171,  0.0074, -0.0072,  0.0027,
         0.0236,  0.0012,  0.0029, -0.0449,  0.0157,  0.0024, -0.0143, -0.0172,
         0.0073,  0.0056, -0.0017,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0503,  0.3692,  0.0086,  0.0713,  0.0183,  0.0032,  0.0848,  0.0597,
        -0.0214, -0.0114,  0.0051,  0.0307,  0.0779, -0.0147, -0.0228,  0.0395,
         0.0048,  0.0025,  0.0009,  0.0034, -0.0168, -0.0057, -0.0386, -0.0383,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.0409e-02, -2.9076e-01, -7.5848e-03, -2.3218e-02, -1.1270e-02,
         8.5218e-03, -4.4820e-03, -2.5533e-02, -9.8115e-02,  9.9858e-03,
        -1.6337e-02,  1.5235e-04,  1.3222e-03, -3.1869e-02, -3.9554e-02,
         1.9343e-03,  6.0086e-03, -7.7514e-03, -3.6054e-02,  1.5724e-02,
        -9.3414e-03, -4.2041e-02,  5.9032e-03, -1.8444e-02, -8.1955e-03,
        -3.7838e-02, -1.2411e-02, -1.5876e-02, -7.5855e-03, -2.2361e-02,
        -4.6672e-02, -4.1537e-03, -1.8701e-04, -6.2405e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.1322e-02,  1.0105e-02, -1.2755e-02,  4.0950e-03,  1.3142e-02,
         1.2747e-02, -6.9400e-02,  1.4519e-02,  1.3830e-02,  4.6729e-03,
         1.3213e-02, -3.9498e-03, -1.3829e-01, -2.2784e-02, -9.0631e-02,
        -2.0421e-01,  1.0696e-02,  8.9677e-03, -4.4126e-05,  2.0776e-02,
         1.9345e-02, -2.2704e-02, -7.7244e-03, -8.6952e-02, -5.7700e-02,
        -2.3611e-02, -4.9286e-02,  1.5329e-03, -3.7081e-04, -3.0623e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0555,  0.0374, -0.0342, -0.1799, -0.0504,  0.0153, -0.0330, -0.0630,
        -0.0966, -0.2433,  0.0683,  0.0913, -0.0318,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0294,  0.0286,  0.0900,  0.0055, -0.0091, -0.0194, -0.0620, -0.0937,
        -0.2021, -0.0077,  0.0221, -0.1086,  0.0087, -0.0597, -0.1985, -0.0025,
        -0.0139, -0.0386,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1701, -0.1841, -0.2133, -0.0948, -0.0717,  0.0882,  0.0443, -0.1335,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0324, -0.1427,  0.0172, -0.0148,  0.0048, -0.0109, -0.0160, -0.0129,
         0.0046,  0.0042,  0.0042,  0.0105, -0.0204, -0.0006, -0.0050, -0.0143,
         0.0002, -0.0053,  0.0067,  0.0007,  0.0132, -0.0096, -0.0324, -0.0184,
         0.0064, -0.0119, -0.0064, -0.0126, -0.0731, -0.0218,  0.0027, -0.0579,
        -0.0154, -0.1511, -0.0061, -0.0632, -0.0295, -0.0076,  0.0104, -0.0038,
        -0.0162, -0.0107, -0.0128,  0.0039,  0.0247, -0.0115,  0.0384],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1049,  0.2522, -0.1531,  0.2773,  0.0166,  0.0636, -0.0182,  0.0011,
        -0.0312,  0.0783, -0.0034,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0560, -0.2382, -0.0349, -0.1400, -0.0208,  0.0163, -0.0446, -0.1739,
        -0.0158, -0.0916, -0.0075,  0.0128, -0.0646,  0.0012,  0.0225, -0.0140,
        -0.0196,  0.0258,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0056, -0.1372, -0.3401,  0.0188,  0.0644,  0.0497, -0.0582,  0.0113,
        -0.1293, -0.0405, -0.0447, -0.0278, -0.0348, -0.0291,  0.0085,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0029, -0.3500, -0.0325, -0.0610,  0.0129, -0.1034, -0.1308,  0.0461,
        -0.0331, -0.0435, -0.0721,  0.0210, -0.0387, -0.0180,  0.0340,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1293,  0.0124,  0.0036,  0.0025,  0.0029,  0.0099, -0.0024,  0.0023,
        -0.0217, -0.0390,  0.0028, -0.1387, -0.1337, -0.0147,  0.0091, -0.0089,
        -0.0764,  0.0772, -0.0123, -0.0098, -0.0942, -0.1329,  0.0062,  0.0149,
        -0.0046,  0.0375,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0925,  0.0359, -0.0338, -0.0024,  0.0489,  0.0751, -0.0050, -0.2163,
        -0.3910, -0.0318,  0.0173,  0.0041, -0.0243,  0.0078, -0.0137,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0068, -0.2265, -0.0213, -0.0735, -0.0510, -0.0246, -0.0592, -0.0223,
        -0.0222, -0.1824,  0.0218,  0.0059, -0.0321, -0.0158,  0.0250, -0.0040,
        -0.0124, -0.0055, -0.0373, -0.0335, -0.0360, -0.0081,  0.0064,  0.0161,
         0.0192,  0.0313,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1427, -0.0046,  0.0086,  0.0219,  0.0035,  0.0004, -0.0037, -0.0134,
        -0.0124, -0.0086, -0.0395, -0.0058, -0.0092, -0.0912, -0.0223, -0.0234,
        -0.0104, -0.0335,  0.0108, -0.0231,  0.1170, -0.0097,  0.0370,  0.0966,
         0.1733, -0.0774,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0555, -0.2851,  0.0382,  0.0065,  0.0028,  0.0090, -0.0501, -0.0394,
        -0.0120, -0.0541, -0.0271, -0.0337, -0.0640,  0.0091, -0.0356,  0.0045,
        -0.0032, -0.0031, -0.0107,  0.0040,  0.0149, -0.0154,  0.0117, -0.0302,
        -0.0718, -0.0031, -0.0182, -0.0011,  0.0028, -0.0012,  0.0034, -0.0382,
         0.0146,  0.0157,  0.0010,  0.0088,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0230, -0.1154, -0.0578,  0.0123, -0.0604,  0.0034, -0.0571, -0.0820,
         0.0019, -0.0017, -0.0215,  0.0028, -0.0020,  0.0096, -0.0106, -0.0164,
         0.0060, -0.0041,  0.0268,  0.0146,  0.0003, -0.1128, -0.0191,  0.0053,
        -0.0564, -0.0033,  0.0076, -0.0271, -0.0780,  0.0168, -0.0088, -0.0485,
        -0.0439,  0.0219, -0.0073,  0.0135,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0821, -0.0457, -0.0643, -0.0477, -0.0396, -0.2034, -0.0370, -0.0222,
        -0.0193, -0.0025, -0.0349, -0.0947, -0.1292, -0.0071,  0.0256,  0.0474,
        -0.0648, -0.0071, -0.0131,  0.0123,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0768, -0.0594, -0.0035, -0.0520, -0.0011, -0.0002,  0.0229, -0.0012,
         0.0040, -0.0536, -0.0056, -0.0067, -0.0067, -0.0213, -0.0944, -0.0431,
        -0.0208, -0.0497, -0.0203, -0.0097,  0.0162, -0.0114,  0.0070,  0.0200,
         0.0050, -0.0038, -0.0106, -0.0178, -0.0082, -0.0359, -0.1095,  0.0123,
         0.0038, -0.0771,  0.0164, -0.0296,  0.0553,  0.0074,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.8645e-02, -1.9107e-01, -1.1331e-01, -9.3960e-02, -4.1060e-03,
         1.2753e-03,  4.3769e-04,  1.3010e-02, -2.5616e-03,  3.3795e-03,
        -5.0843e-02, -4.6955e-02, -9.9292e-04,  8.3772e-03, -1.1496e-02,
        -4.8616e-02,  1.0284e-02, -5.2001e-02,  3.1396e-03,  4.9053e-03,
         1.1800e-02,  1.3727e-02,  1.7175e-02,  1.3106e-02,  1.7616e-02,
        -7.6953e-03, -4.2634e-03, -2.6625e-02, -8.1935e-02, -4.4908e-03,
        -1.8479e-04, -4.9594e-03,  7.4583e-04, -4.7098e-03, -2.0229e-02,
        -2.5646e-03, -8.2105e-03,  5.4640e-03,  4.5935e-03,  1.6039e-02,
        -3.4505e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.5095e-02, -4.2531e-01, -1.9166e-02, -1.1564e-02, -2.0550e-02,
        -5.7030e-02, -1.9579e-02, -1.2051e-02, -1.2907e-02,  4.0439e-03,
         1.1055e-04, -4.1605e-02, -5.3401e-02, -2.6777e-03,  1.1026e-02,
        -6.2661e-02, -1.0309e-02, -7.8411e-02,  5.0040e-03,  9.5377e-04,
        -2.0554e-02,  6.5995e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1550e-02,  2.9891e-03, -1.0595e-03,  8.8877e-03,  3.7028e-02,
        -4.3174e-02, -2.2865e-03,  1.1441e-02, -2.3150e-02,  6.7057e-03,
        -3.6657e-03, -6.9347e-02, -7.1479e-05, -2.7172e-02, -2.4044e-02,
        -8.0067e-02,  1.8351e-02, -3.7854e-02, -6.4836e-03, -7.0106e-02,
        -1.0089e-01,  3.1853e-02,  1.3376e-02, -6.3352e-02, -1.0188e-02,
         1.9294e-02, -7.4761e-03, -5.7769e-02, -8.5295e-02,  3.2712e-03,
        -3.8253e-02, -4.5982e-02, -2.0104e-02, -1.5462e-02, -1.9996e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0636, -0.0441, -0.1752, -0.0464, -0.0633, -0.0166,  0.0169,  0.0052,
        -0.0029, -0.0068,  0.0028,  0.0190,  0.0021,  0.0134,  0.0144, -0.0119,
         0.0687, -0.0260, -0.0938, -0.0254, -0.0633, -0.0068, -0.0100, -0.0695,
        -0.0075,  0.0147, -0.0840,  0.0209, -0.0048,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0749, -0.4191, -0.0881, -0.0435, -0.0502, -0.0530, -0.0020,  0.0108,
        -0.0055, -0.0205, -0.0199, -0.0345, -0.0134, -0.0083, -0.0022, -0.0415,
        -0.0218, -0.0020, -0.0056,  0.0152, -0.0074, -0.0027, -0.0028,  0.0144,
         0.0411,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0623,  0.0093,  0.0213,  0.0303, -0.1585, -0.0168, -0.0431, -0.1256,
        -0.1512, -0.0294,  0.0085,  0.0184,  0.0042,  0.0058, -0.1020, -0.1136,
        -0.0072, -0.0356,  0.0027,  0.0058, -0.0169,  0.0271,  0.0044,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.8679e-03, -2.0129e-02, -2.0698e-03, -8.4654e-03,  1.8562e-02,
         6.4099e-02,  1.0974e-02,  7.5536e-03,  1.0102e-02, -3.0431e-03,
        -1.9720e-03, -1.3273e-02, -5.3074e-02, -1.0217e-01,  1.4626e-02,
        -2.6226e-02, -1.8344e-02, -6.0363e-02, -2.5062e-02, -1.6084e-02,
        -2.9276e-02, -1.4776e-02, -2.0861e-02, -1.5130e-04, -1.9118e-02,
         5.6338e-03, -1.8669e-01,  1.8141e-03, -1.0514e-01, -1.8093e-02,
        -4.2093e-02, -6.4306e-03, -1.7474e-02, -6.3303e-03, -2.9181e-02,
         4.8301e-03, -2.4803e-03,  9.1279e-03, -4.3675e-04,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0178, -0.0871,  0.0007, -0.1285, -0.3263, -0.0226,  0.0758, -0.1073,
        -0.1382,  0.0065, -0.0062, -0.0829,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 1.5170e-03, -9.9377e-02, -3.2323e-01, -3.6288e-03, -1.4031e-02,
        -3.4237e-02, -3.9119e-02,  3.2174e-02, -1.6796e-02, -3.4333e-02,
        -2.9650e-03,  1.3814e-02, -3.6124e-03, -3.7375e-04, -2.1000e-02,
         1.4559e-02, -2.8075e-02, -4.5443e-02, -2.2690e-03, -1.1001e-02,
        -2.7704e-02, -1.9146e-02,  4.1348e-03, -4.5489e-03,  1.6360e-03,
         2.3603e-03,  1.8895e-03, -1.0782e-03, -8.4746e-03, -1.8349e-02,
         1.4189e-02, -5.0849e-03, -3.6577e-02,  1.2202e-04,  6.6797e-03,
        -8.8072e-03, -5.4566e-02, -3.1403e-03,  6.8560e-03, -3.3684e-05,
         8.1064e-03,  4.1017e-03,  2.7355e-03, -5.4653e-04,  1.7577e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5951e-02,  5.7744e-01,  3.7901e-02,  1.0305e-01, -2.6746e-02,
        -6.2724e-04, -1.7700e-02,  3.5306e-02,  1.4446e-01, -3.2151e-02,
        -8.1937e-03,  4.7204e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0901,  0.6921,  0.0097,  0.0985, -0.0114,  0.0403,  0.0223, -0.0356,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0574, -0.1905, -0.0214, -0.0161, -0.0539,  0.0111, -0.0085, -0.0113,
        -0.0146, -0.0674, -0.0074, -0.0109, -0.0123, -0.0014, -0.0018,  0.0069,
         0.0081, -0.0072,  0.0214, -0.0017,  0.0037,  0.0105, -0.0047,  0.0060,
        -0.0108, -0.0019,  0.0045, -0.0100, -0.0111,  0.0021, -0.0040, -0.0064,
         0.0004,  0.0040, -0.0093, -0.0012,  0.0005, -0.0021, -0.0022,  0.0029,
        -0.0053, -0.0043, -0.0010, -0.0151,  0.0144,  0.0485, -0.0640, -0.0207,
        -0.0339, -0.0449, -0.0251, -0.0042, -0.0219,  0.0018,  0.0015, -0.0068,
        -0.0198, -0.0024, -0.0027,  0.0044,  0.0280], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0408, -0.0007,  0.0052, -0.0215, -0.0486,  0.0278, -0.0295, -0.0052,
        -0.0151, -0.0117, -0.0109, -0.0704,  0.0004, -0.0479, -0.1079,  0.0016,
        -0.0068,  0.0117, -0.0539, -0.0311, -0.0806, -0.0663, -0.0255, -0.0048,
        -0.0497, -0.0179, -0.0068, -0.0223, -0.0274, -0.0342,  0.0115,  0.0164,
        -0.0879,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0871, -0.0524, -0.1766, -0.0817,  0.0024,  0.0079, -0.0708,  0.0065,
         0.0111, -0.0037,  0.0136, -0.0054, -0.0234,  0.0033, -0.0137,  0.0397,
         0.0219, -0.0780, -0.1810,  0.0163,  0.0263, -0.0319, -0.0109, -0.0342,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0305, -0.1096, -0.1719, -0.0532, -0.0382, -0.1039, -0.0249, -0.0487,
        -0.0076, -0.0401, -0.0831, -0.0381, -0.0921,  0.0075, -0.0148, -0.0081,
         0.0096,  0.0031,  0.0201,  0.0270, -0.0261, -0.0031,  0.0148, -0.0239,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0132,  0.0363, -0.0338, -0.0939,  0.0116, -0.0033,  0.0118, -0.0527,
         0.0026,  0.0106,  0.0032, -0.0084, -0.0106, -0.0122, -0.0887, -0.0815,
        -0.0683, -0.0060, -0.0217, -0.0154, -0.0958, -0.0081, -0.0364, -0.0283,
         0.0056, -0.0050, -0.0564, -0.0073,  0.0096, -0.0011, -0.0217, -0.0111,
         0.0619,  0.0658,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1050,  0.0277,  0.0032, -0.0036, -0.0816,  0.0140, -0.0174, -0.0061,
        -0.0277, -0.0095,  0.0056,  0.0162,  0.0120, -0.1350, -0.0472,  0.0453,
        -0.1285, -0.0052, -0.0839,  0.0170, -0.0504, -0.1115, -0.0148, -0.0034,
        -0.0281,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0695,  0.0250, -0.0401, -0.0060, -0.0121, -0.0531, -0.1447, -0.0615,
        -0.0958, -0.3201,  0.0364, -0.0342, -0.0119, -0.0035, -0.0004, -0.0857,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0518, -0.3442, -0.0215, -0.0909, -0.0298, -0.0308, -0.2050, -0.0079,
         0.0159, -0.0011,  0.0110, -0.0098, -0.0205,  0.0093, -0.0080,  0.0054,
        -0.0295, -0.0429, -0.0214, -0.0092,  0.0081, -0.0077, -0.0181,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1229, -0.0953, -0.0474, -0.2520, -0.2734, -0.0693, -0.0185, -0.0209,
        -0.0409, -0.0040, -0.0167,  0.0177,  0.0195, -0.0015,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0038, -0.0877, -0.0502, -0.0992, -0.0120,  0.0118, -0.0692,  0.0256,
        -0.0336, -0.0284, -0.1275, -0.1117, -0.0115, -0.0147, -0.0094,  0.0186,
        -0.0040, -0.1202,  0.0094, -0.1002, -0.0137,  0.0135, -0.0131, -0.0110,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1388,  0.0607, -0.1098, -0.0348, -0.2315, -0.0277, -0.2617, -0.0063,
         0.0060,  0.0310, -0.0918,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0259, -0.0038, -0.0138, -0.0083,  0.0008, -0.1032, -0.2109, -0.0011,
        -0.0920, -0.0161, -0.0340,  0.0120, -0.0168,  0.0334, -0.0771, -0.0222,
        -0.0564, -0.1004, -0.0521, -0.0575, -0.0622,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.0341e-01, -3.9981e-03,  5.0688e-02,  4.9973e-02, -3.7132e-02,
         2.8553e-02,  5.8256e-02, -3.6667e-01, -2.1985e-02, -1.2753e-01,
        -1.4254e-02, -2.9231e-04,  1.2117e-02, -1.2514e-01,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0377,  0.0646,  0.0097, -0.0019, -0.0202,  0.0060,  0.0009, -0.0040,
         0.0151,  0.0149, -0.0444, -0.1511, -0.3804,  0.0161, -0.0140, -0.0476,
        -0.0559, -0.0037, -0.0373, -0.0213, -0.0533,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.4816e-02,  3.7358e-01, -3.1003e-02,  6.8263e-02, -7.3562e-02,
         7.2151e-02,  2.5895e-02,  1.7630e-01, -3.2542e-04,  2.5156e-02,
         5.7269e-02, -4.0548e-02,  3.1128e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.7056e-02, -1.4768e-01, -1.7720e-02, -1.6676e-02, -1.2331e-02,
        -2.2680e-02, -5.6069e-02, -1.0854e-01, -5.0968e-02,  7.1577e-03,
         8.2658e-03,  5.8039e-03, -5.7558e-03,  6.4821e-03,  3.5260e-03,
        -6.7919e-03,  1.0008e-03,  4.5005e-05, -3.6896e-03,  1.7770e-02,
         1.8327e-03,  1.8557e-02,  6.7713e-03, -1.7492e-02, -1.3047e-01,
        -1.8396e-03, -7.3849e-02, -7.5307e-02, -3.2191e-02, -4.1870e-04,
        -9.7730e-03, -3.5705e-03,  2.3577e-03, -4.1242e-03,  6.2069e-03,
         4.2636e-03, -2.2173e-02, -5.5280e-03,  3.1257e-03, -8.1697e-03,
        -1.7164e-02, -6.6525e-03,  4.5560e-03,  7.3423e-03,  2.0258e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0430, -0.0684, -0.0363, -0.0047, -0.0023,  0.0041, -0.0073, -0.0724,
        -0.0196, -0.0264, -0.1022, -0.0381, -0.0070,  0.0094, -0.0499, -0.0192,
        -0.1300,  0.0042, -0.0083,  0.0124, -0.0076, -0.0548,  0.0040, -0.0512,
         0.0140,  0.0133, -0.0187, -0.0065, -0.0777, -0.0257, -0.0136, -0.0183,
         0.0024,  0.0269,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1118,  0.0073,  0.0026, -0.0056, -0.0008, -0.0108,  0.0101, -0.0355,
         0.0440, -0.0061, -0.0006, -0.0051,  0.0114, -0.0239, -0.0264, -0.0177,
        -0.0048,  0.0098, -0.0146, -0.2715, -0.2401, -0.0081,  0.0336,  0.0143,
        -0.0191, -0.0646,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0195, -0.4121, -0.0113, -0.1167,  0.0570, -0.0446,  0.0151, -0.0489,
        -0.0114,  0.0024,  0.0035,  0.0106,  0.0133, -0.0110, -0.0129, -0.0378,
        -0.0341,  0.0164, -0.0334, -0.0024, -0.0159,  0.0135,  0.0560,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0149, -0.1505, -0.0112, -0.0260, -0.0013, -0.0034, -0.0212, -0.0537,
        -0.1271, -0.0159, -0.0053, -0.0171,  0.0082, -0.0017, -0.1341,  0.0396,
         0.0006, -0.0410, -0.0682, -0.0171, -0.0045, -0.0073, -0.0054, -0.0078,
        -0.0045, -0.0006, -0.0579, -0.0864, -0.0029,  0.0038, -0.0078, -0.0091,
         0.0021,  0.0072, -0.0006,  0.0035, -0.0214,  0.0068,  0.0017, -0.0006,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1032, -0.0521, -0.0039, -0.0113, -0.0952, -0.0134, -0.0480, -0.0632,
         0.0241, -0.0068,  0.0013,  0.0031, -0.0264, -0.0071, -0.0243, -0.0879,
        -0.0461,  0.0098, -0.0039, -0.0473,  0.0040, -0.0617, -0.0384, -0.0079,
        -0.0321, -0.0013, -0.0044, -0.0115, -0.0066, -0.0319, -0.0706,  0.0078,
        -0.0195, -0.0239,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0292, -0.0065,  0.0044, -0.0036, -0.0457,  0.0054,  0.0087,  0.0047,
         0.0023,  0.0029, -0.0105, -0.0257,  0.0143, -0.0126, -0.1627,  0.0320,
        -0.0214, -0.0148, -0.0527, -0.0031, -0.0125,  0.0021,  0.0021,  0.0154,
        -0.0286, -0.0876,  0.0198,  0.0164, -0.0048, -0.0031,  0.0082, -0.0012,
         0.0167, -0.0254, -0.0386, -0.0098, -0.0331, -0.0453,  0.0011, -0.0240,
        -0.0431,  0.0003, -0.0430, -0.0072,  0.0095,  0.0030,  0.0029,  0.0063,
        -0.0112, -0.0051, -0.0091,  0.0003], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0506, -0.0060, -0.0106, -0.0214, -0.1204, -0.0176, -0.0445,  0.0143,
        -0.0134,  0.0128,  0.0157, -0.0028, -0.0277, -0.1123, -0.0312, -0.0246,
        -0.0959,  0.0014, -0.0480,  0.0013, -0.0569, -0.0424, -0.0318, -0.1314,
        -0.0048, -0.0036,  0.0060, -0.0106, -0.0269, -0.0038,  0.0091,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0262,  0.1948, -0.0183,  0.0217, -0.0268, -0.0032,  0.0269,  0.0243,
         0.0101, -0.0495,  0.0180,  0.0994,  0.0003, -0.0079,  0.0656,  0.0467,
         0.0009,  0.0636,  0.0930,  0.0736,  0.0292,  0.0083,  0.0181, -0.0359,
        -0.0375,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1240,  0.0074, -0.0739, -0.0139, -0.0456, -0.2310, -0.3878, -0.0598,
        -0.0195,  0.0162,  0.0088,  0.0121,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0014, -0.3635, -0.0243, -0.0422, -0.0944, -0.0322,  0.0067, -0.0100,
        -0.0049, -0.0128, -0.0102, -0.0228, -0.0580, -0.0422, -0.0074, -0.0657,
        -0.0093, -0.0140, -0.0064,  0.0006, -0.0250, -0.0037, -0.0274,  0.0111,
         0.0322, -0.0551, -0.0018, -0.0063, -0.0086,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0050, -0.4696, -0.0403, -0.0117, -0.0105, -0.1170, -0.1750,  0.0848,
         0.0150,  0.0032,  0.0406,  0.0275,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0649, -0.3527, -0.0370, -0.0939,  0.0091, -0.0803, -0.1955, -0.0128,
        -0.0596, -0.0297,  0.0223, -0.0422,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0868,  0.0671, -0.0804,  0.0248, -0.1004, -0.0185,  0.0096, -0.0397,
        -0.1365,  0.0054, -0.0109, -0.1405, -0.0117, -0.0237, -0.0487, -0.1061,
         0.0039,  0.0166,  0.0354, -0.0331,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.0782e-02, -2.9058e-01, -1.7640e-02, -6.9335e-02, -1.3454e-01,
         1.3949e-03, -2.8403e-02, -2.2623e-02, -1.8927e-02, -6.7165e-03,
         3.4081e-03, -3.1524e-02, -1.9846e-02, -8.9781e-03,  1.0263e-02,
        -2.3355e-04, -2.4485e-02,  1.1388e-02,  1.7426e-03, -4.5618e-03,
        -1.6597e-02,  1.8680e-02,  1.8857e-02, -1.1724e-02,  4.1934e-03,
        -8.2359e-03,  7.1093e-03, -3.7309e-02, -4.4771e-02,  5.4358e-03,
        -1.0018e-02, -3.2188e-03, -3.1745e-02, -3.0228e-02, -1.3511e-03,
         1.0220e-02, -2.9339e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1522, -0.5971, -0.0597,  0.0706,  0.0205,  0.0139, -0.0595,  0.0049,
         0.0216,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0216,  0.0353,  0.0214, -0.1958, -0.0110, -0.0172, -0.0667, -0.0866,
         0.0042, -0.0164, -0.0217, -0.0175, -0.0813, -0.1379,  0.0011,  0.0031,
         0.0051,  0.0162, -0.0588, -0.0973,  0.0077,  0.0025, -0.0087,  0.0010,
         0.0049, -0.0156, -0.0432,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0809,  0.2962, -0.0040,  0.0484,  0.0275,  0.0466, -0.0054, -0.0037,
        -0.0010, -0.0024, -0.0176,  0.0357,  0.0023, -0.0045, -0.0020, -0.0087,
         0.0279, -0.0054,  0.0014, -0.0068, -0.0056,  0.0380,  0.0394, -0.0067,
        -0.0041, -0.0094,  0.0032,  0.0158,  0.1281,  0.0025, -0.0059,  0.0232,
        -0.0503,  0.0390,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0582,  0.0039,  0.0047,  0.0068, -0.0289,  0.0037,  0.0054,  0.0059,
         0.0091,  0.0162, -0.0091, -0.0343, -0.0111, -0.0097, -0.0005, -0.0449,
        -0.1293, -0.0216, -0.0318,  0.0014,  0.0207,  0.0121, -0.0387, -0.0232,
        -0.0017, -0.0140, -0.0110, -0.0065,  0.0046,  0.0095, -0.0376, -0.0613,
        -0.0598, -0.0127, -0.0192, -0.0364, -0.0462, -0.0031,  0.0140,  0.0091,
        -0.0187, -0.0560,  0.0111,  0.0111, -0.0250,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1050,  0.0010, -0.0189, -0.0104, -0.0150, -0.1287, -0.2007, -0.0063,
         0.0439, -0.1528, -0.0225, -0.0626, -0.1323, -0.0549, -0.0079, -0.0372,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0148, -0.0129,  0.0244,  0.0146, -0.0230, -0.0187, -0.0236,  0.0324,
         0.2286,  0.1869, -0.0059,  0.1232, -0.0150,  0.0515,  0.0637, -0.0283,
         0.0086, -0.0114, -0.0143,  0.0024, -0.0006, -0.0245,  0.0071, -0.0099,
        -0.0318,  0.0048, -0.0169,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0363,  0.0087, -0.0258,  0.0049, -0.0204, -0.0004, -0.0184,  0.0005,
        -0.0149, -0.0257,  0.0017, -0.0067, -0.0418, -0.0029, -0.0398, -0.0282,
        -0.0446,  0.0021, -0.0014, -0.0233,  0.0031, -0.0227, -0.0219, -0.0705,
         0.0025, -0.0051, -0.0302, -0.0763, -0.0135, -0.0559,  0.0031, -0.0636,
        -0.0564, -0.0080,  0.0003, -0.0087, -0.0002, -0.0418,  0.0230,  0.0214,
        -0.0210, -0.0233, -0.0020,  0.0088,  0.0205, -0.0474], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0706,  0.0071,  0.0199,  0.0110, -0.0022,  0.0072, -0.0050, -0.0627,
         0.0103, -0.0106,  0.0030,  0.0488,  0.0165, -0.0051, -0.0260, -0.0027,
         0.0029, -0.0187, -0.1140, -0.2620, -0.0293, -0.0321, -0.0088, -0.0176,
        -0.0145, -0.0100, -0.1021, -0.0217, -0.0477, -0.0100,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0645, -0.0667, -0.1091, -0.0460,  0.0302,  0.0080,  0.0417,  0.3361,
        -0.0591,  0.1053,  0.1332,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1352, -0.3389, -0.0496, -0.0968, -0.0475, -0.0203, -0.0065, -0.1002,
         0.0244, -0.0183, -0.0011, -0.0052,  0.0194,  0.0021, -0.0089, -0.0543,
        -0.0137, -0.0209, -0.0218, -0.0150,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0243, -0.3385, -0.2028, -0.0769,  0.0207, -0.0770, -0.0921, -0.0152,
        -0.0278,  0.0610,  0.0637,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0277, -0.2591, -0.1271, -0.1330, -0.0336,  0.0077, -0.0389, -0.0735,
        -0.1028,  0.0162, -0.0057, -0.0038, -0.0164, -0.0122,  0.0005, -0.0356,
         0.0030, -0.0110, -0.0238, -0.0046,  0.0637,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0140, -0.3796, -0.1577, -0.0283,  0.0038, -0.2621,  0.0400,  0.0021,
         0.0040,  0.0055,  0.0195, -0.0115,  0.0028,  0.0015, -0.0061,  0.0128,
         0.0449, -0.0039,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.0710e-02, -1.2296e-01, -1.7139e-01,  1.1680e-02, -1.6330e-02,
         9.7964e-03, -1.7186e-02,  4.8270e-03, -3.2043e-03, -2.7645e-05,
         1.1844e-02,  1.6705e-02, -1.2621e-02, -7.4041e-02, -4.5483e-02,
         4.1624e-03, -2.9715e-02, -8.8276e-02, -1.1425e-02, -1.5319e-02,
        -2.3903e-02, -3.6031e-02, -6.4029e-02,  1.0634e-03, -2.0686e-02,
         3.3770e-03, -3.0888e-02, -7.8083e-03, -1.3341e-02, -4.0670e-02,
        -3.0767e-03,  2.8151e-03, -1.4480e-02,  4.6448e-03, -1.6874e-02,
        -3.8731e-03,  3.1342e-03, -5.7486e-03,  2.5849e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.1582e-02, -3.7056e-01, -1.0076e-02, -6.1220e-02, -1.0481e-02,
        -1.4854e-02, -2.3653e-02, -1.1029e-04, -4.0261e-02, -1.2414e-02,
        -2.3140e-02, -1.1255e-01, -2.4952e-02, -9.9662e-04, -1.9994e-02,
        -1.6480e-03, -1.0178e-02, -5.0577e-04,  4.6454e-04, -7.7401e-03,
         1.2895e-02,  8.1184e-03,  7.5445e-03, -1.3838e-03,  6.3812e-03,
        -4.2968e-03, -1.2209e-02,  1.0641e-02, -4.5616e-02, -8.2413e-02,
         1.3603e-03, -1.3714e-03,  2.4157e-02,  1.4236e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.1100, -0.0212, -0.2285, -0.3115,  0.0466, -0.0717,  0.0415, -0.0567,
        -0.1123,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.3166e-03, -1.8049e-01, -3.9258e-02, -4.1596e-02, -1.8223e-02,
        -1.4010e-01,  7.1418e-03, -1.8291e-02,  4.5571e-03,  1.2297e-02,
        -1.5495e-02,  2.0073e-05, -9.1858e-03, -1.6319e-02, -1.3475e-02,
        -7.5863e-02, -1.8428e-02, -1.1999e-01, -1.0420e-03, -1.0249e-02,
        -2.9267e-03,  1.6099e-03,  2.9156e-03, -1.6817e-03, -2.6427e-02,
         7.2313e-03, -5.3325e-03, -2.2008e-02,  1.2366e-03, -6.8867e-03,
        -2.6073e-02, -5.1351e-02,  7.8716e-04, -3.9459e-02,  1.7550e-03,
        -5.8399e-03, -4.6923e-03,  1.8165e-02,  6.1028e-03,  1.6183e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0004, -0.0533, -0.1360, -0.0110, -0.0489, -0.0220, -0.0108, -0.0144,
        -0.0817, -0.0809, -0.0041, -0.0046,  0.0018, -0.0022,  0.0032,  0.0066,
         0.0094, -0.0322, -0.0502, -0.0005, -0.0288, -0.0075,  0.0024, -0.0044,
         0.0029, -0.0186,  0.0117,  0.0155, -0.0405, -0.0898,  0.0100,  0.0136,
         0.0046,  0.0041,  0.0077,  0.0196, -0.0141, -0.0549, -0.0118,  0.0085,
        -0.0048, -0.0013,  0.0054, -0.0055, -0.0048,  0.0052, -0.0012,  0.0032,
         0.0165, -0.0026, -0.0014, -0.0032,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0328,  0.1689,  0.0288,  0.0296,  0.0146,  0.0282,  0.0708,  0.0098,
         0.0298,  0.0358, -0.0081,  0.0498,  0.1115,  0.0245,  0.0372,  0.0672,
         0.0499,  0.0405,  0.0800,  0.0129,  0.0109,  0.0583,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0223,  0.5096,  0.0045, -0.0578, -0.0036, -0.0241,  0.0233,  0.0768,
         0.0198, -0.0050, -0.0012,  0.0372,  0.0106,  0.0162,  0.0116,  0.0747,
        -0.0014, -0.0192,  0.0133, -0.0615,  0.0062,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.8695e-02, -2.3205e-01, -2.2290e-02, -1.8582e-02,  1.0659e-02,
         1.2214e-02, -4.5555e-03,  1.8488e-02, -1.7100e-02,  2.5055e-03,
         6.1819e-03, -1.2368e-03,  3.8215e-03,  9.4563e-03, -4.6060e-03,
         1.3522e-03, -3.9435e-03, -7.6221e-04, -1.7414e-02,  1.0684e-02,
         5.3780e-03, -4.8024e-03, -5.6069e-03,  8.3399e-03,  2.1920e-03,
         5.6332e-03, -2.1031e-02, -3.1073e-02,  3.1580e-03, -2.9106e-02,
        -9.3042e-02, -6.1788e-02, -7.6509e-03, -2.0901e-02, -1.7108e-02,
         2.8983e-03,  1.3970e-03,  1.8798e-03, -9.3293e-04, -1.8390e-03,
        -8.7165e-02,  2.5474e-03,  5.5719e-03, -1.6651e-02, -4.6399e-02,
         2.5085e-03, -2.4729e-02, -1.5465e-02,  6.7986e-03,  1.8509e-03,
        -1.5079e-02,  1.4317e-04,  4.7266e-03, -7.7106e-03,  2.0296e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0213, -0.0141,  0.0406,  0.0020,  0.0092, -0.0197, -0.0963,  0.0006,
         0.0083,  0.0061, -0.0125,  0.0109,  0.0166, -0.0173, -0.0542, -0.0178,
         0.0059, -0.0286, -0.0097, -0.0569, -0.0090, -0.0218, -0.0359,  0.0037,
        -0.0033,  0.0066,  0.0009, -0.0620,  0.0044,  0.0078,  0.0339, -0.0104,
        -0.0176, -0.1004, -0.0214, -0.0652, -0.0045, -0.0320, -0.0435, -0.0225,
        -0.0094, -0.0090, -0.0055,  0.0033,  0.0175,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0539, -0.2613, -0.0786, -0.0719,  0.0453, -0.0082, -0.0141, -0.0172,
        -0.0568,  0.0184,  0.0020, -0.0263,  0.0064, -0.0798, -0.0211,  0.0362,
        -0.0068, -0.0505, -0.0129, -0.0266, -0.0457,  0.0294, -0.0188,  0.0119,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0343,  0.0192,  0.0113,  0.0168, -0.0161,  0.0060, -0.0297, -0.0352,
         0.0158,  0.0083,  0.0131, -0.0029,  0.0093, -0.0217, -0.0036,  0.0300,
        -0.0380, -0.0317, -0.0368, -0.0568, -0.0152, -0.0008, -0.0458, -0.0750,
         0.0446, -0.0520, -0.0966, -0.0323, -0.0414, -0.0905, -0.0201, -0.0147,
        -0.0342,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.3563e-02, -4.7826e-01, -3.2546e-04, -8.3908e-02,  4.9237e-03,
         9.3979e-03,  7.1718e-03,  1.4347e-02, -1.0855e-01, -1.6471e-02,
         3.5068e-03,  4.6069e-03,  3.8411e-02, -7.5564e-02,  2.0074e-02,
        -7.0352e-03, -1.7888e-02,  2.3380e-03,  2.3665e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0579, -0.0089,  0.0235, -0.0583, -0.1309,  0.0154, -0.0600, -0.1078,
        -0.1342, -0.0117,  0.0139,  0.0112, -0.0095,  0.0386, -0.0033,  0.0046,
        -0.0177, -0.0528, -0.0133,  0.0134, -0.0421, -0.1519, -0.0120,  0.0025,
         0.0043,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0168, -0.2456, -0.0647, -0.0178, -0.0322, -0.1087, -0.2608, -0.0351,
        -0.0258, -0.0136, -0.0014, -0.0223, -0.0821, -0.0109, -0.0195,  0.0060,
        -0.0368,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.0887, -0.0158,  0.0199,  0.0037, -0.0019,  0.0321,  0.0004,  0.0220,
        -0.0063,  0.1278,  0.0415,  0.0237, -0.0067,  0.0071, -0.0125, -0.0002,
         0.0355,  0.1038,  0.0044,  0.0407, -0.0098,  0.1022,  0.1467,  0.0118,
         0.0079, -0.0198,  0.0022,  0.0039, -0.0136,  0.0103, -0.0226, -0.0547,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.1119e-02,  9.7918e-02,  1.2867e-01,  2.1414e-02, -2.7967e-02,
         2.4979e-03,  4.4889e-02,  1.1078e-01,  1.0320e-02, -4.1372e-03,
        -2.4807e-02, -3.1121e-03, -2.3221e-02, -1.3250e-03,  1.6950e-02,
        -1.5804e-02,  2.5297e-02,  6.6113e-03,  6.5895e-02,  4.5452e-02,
         3.4468e-02,  2.2958e-02,  3.0424e-02,  9.2475e-02,  4.5121e-02,
         3.9990e-02, -4.7129e-06,  1.6371e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0253, -0.1229,  0.0055, -0.0091,  0.0136, -0.0928, -0.1043, -0.0450,
        -0.0025,  0.0006,  0.0205, -0.0863, -0.0142, -0.0083,  0.0060, -0.0905,
         0.0105, -0.1227, -0.0241,  0.0036, -0.0128,  0.0059, -0.0176, -0.0119,
        -0.0385,  0.0275,  0.0425, -0.0005, -0.0342,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0273,  0.1416,  0.0405,  0.0489,  0.1106,  0.0199,  0.0229, -0.0123,
        -0.0078, -0.0157, -0.0120,  0.0039,  0.0244,  0.0998, -0.0096,  0.0187,
        -0.0015,  0.0919,  0.0134,  0.0258,  0.0128, -0.0190,  0.0490,  0.0148,
         0.0322,  0.0342,  0.0349,  0.0053,  0.0027, -0.0003,  0.0083, -0.0239,
         0.0143,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0099, -0.3588,  0.0245, -0.0499, -0.0244, -0.0797,  0.0027, -0.0496,
        -0.0210, -0.0219, -0.0300, -0.0103, -0.0078,  0.0148,  0.0013,  0.0068,
        -0.0302,  0.0012,  0.0108,  0.0221, -0.0770,  0.0116,  0.0520,  0.0817,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0235, -0.1648, -0.1126, -0.0149, -0.0538, -0.0290, -0.0027, -0.0308,
         0.0157, -0.1246,  0.0035, -0.0600, -0.0335, -0.0224, -0.0006,  0.0092,
         0.0131, -0.0284, -0.0088, -0.1510, -0.0203,  0.0479,  0.0288,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0337, -0.1879,  0.0235, -0.0160, -0.0011, -0.0516,  0.0144,  0.0150,
         0.0006, -0.0099,  0.0083, -0.0083, -0.0139,  0.0009, -0.0003,  0.0063,
        -0.0677, -0.0378, -0.0222, -0.0005, -0.0047, -0.0215,  0.0045, -0.0162,
        -0.0567, -0.0150,  0.0027, -0.0050, -0.0142, -0.0615, -0.0852,  0.0039,
        -0.0301, -0.0445, -0.0283, -0.0530, -0.0264, -0.0038, -0.0028,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.8060e-02,  1.0917e-01,  5.7099e-03,  6.9479e-03,  1.0715e-01,
         1.3151e-03, -4.0272e-03,  1.3254e-03,  5.0036e-03, -8.9897e-03,
         2.3505e-03, -2.4216e-03, -6.8671e-03,  1.4300e-02,  8.1542e-03,
         5.9604e-02,  1.5634e-02, -1.4606e-02,  9.4248e-05,  1.4936e-02,
         4.5823e-02, -7.9773e-03, -2.6174e-03,  1.7111e-02,  6.3390e-02,
         9.1809e-02,  2.3794e-02,  3.5927e-02,  7.9509e-03,  7.9807e-02,
         1.4932e-03,  1.4358e-02,  8.4274e-03,  1.4630e-03, -9.2366e-03,
        -1.4773e-02,  2.3029e-02,  7.2682e-02, -1.2024e-02,  2.3885e-03,
         4.1623e-03,  2.1007e-03, -3.0990e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0641, -0.1938,  0.0025,  0.0078,  0.0071,  0.0165, -0.0119, -0.0103,
        -0.0690, -0.0164,  0.0092, -0.0014, -0.0219, -0.0026,  0.0039, -0.0047,
        -0.0452, -0.0725, -0.0046, -0.0120,  0.0026, -0.0004, -0.0006, -0.0116,
         0.0038, -0.0063, -0.0106, -0.0013,  0.0041,  0.0056,  0.0052, -0.0287,
        -0.0581, -0.0063, -0.0118, -0.0156, -0.0212, -0.0263, -0.0334, -0.0048,
        -0.0268, -0.0280, -0.0081,  0.0025, -0.0143, -0.0236, -0.0007, -0.0028,
        -0.0017, -0.0096,  0.0148, -0.0023, -0.0146, -0.0018, -0.0050, -0.0025,
         0.0051], device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.5555e-02, -4.5943e-01, -3.3760e-02,  3.9232e-03, -4.0569e-02,
        -3.0036e-03, -8.5872e-02, -8.4412e-02, -2.7238e-02, -3.1441e-02,
        -4.9681e-02,  1.8203e-02,  6.6591e-02,  3.1638e-04,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0990, -0.2575, -0.0062,  0.0192,  0.0093, -0.0135, -0.0025,  0.0046,
        -0.0266, -0.0052, -0.0437, -0.1098,  0.0211, -0.0132, -0.0458,  0.0082,
        -0.0168, -0.0395,  0.0265, -0.0290, -0.0827,  0.0007, -0.0468, -0.0067,
         0.0172, -0.0111,  0.0026,  0.0037, -0.0248,  0.0065,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0425,  0.3419, -0.1056,  0.2728,  0.1019, -0.0172,  0.0953,  0.0228,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0460, -0.1196,  0.0352, -0.0142, -0.0436, -0.0896, -0.0147,  0.0006,
        -0.0094, -0.0096,  0.0114, -0.0264, -0.0612, -0.0893, -0.0112, -0.0033,
        -0.0087, -0.0005,  0.0082, -0.0009,  0.0137,  0.0147,  0.0048, -0.0481,
        -0.0376,  0.0052, -0.0570, -0.0152, -0.0412, -0.0822,  0.0028, -0.0571,
        -0.0168,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.2797e-02, -1.8632e-01, -7.0031e-02, -3.1778e-02, -2.2458e-03,
        -3.0988e-02, -3.6220e-03, -1.0510e-02,  2.1082e-02, -1.8329e-02,
         1.4248e-03, -3.0962e-04, -8.5708e-03,  5.2107e-03,  6.7609e-03,
         2.6883e-03,  3.3547e-03,  1.1755e-02, -1.1795e-02,  4.4756e-03,
         6.2825e-05,  6.5497e-04, -2.8557e-03,  6.7197e-03, -2.8843e-03,
        -7.5386e-03, -1.1370e-02, -3.8272e-02, -1.6801e-04, -3.9683e-03,
        -6.2706e-03, -1.1339e-02, -4.2550e-02, -5.9697e-02, -2.4045e-02,
        -5.7231e-03, -2.2819e-02, -2.7279e-03, -1.5508e-03,  1.1073e-02,
         4.0195e-03,  4.7530e-03, -3.7418e-03, -5.1423e-02, -4.1162e-03,
        -6.2209e-03, -5.5081e-03, -5.2112e-03, -7.4223e-03, -8.9296e-03,
        -3.8985e-03, -2.2993e-02, -2.9362e-03, -4.3749e-03,  3.7601e-03,
         6.9426e-03,  2.5418e-03,  4.3705e-03,  6.3683e-04, -7.0184e-03,
        -2.5645e-03, -1.0956e-02, -5.0894e-02, -1.5911e-03,  1.3386e-03,
         5.0409e-04,  2.7407e-04,  8.1779e-03,  1.0256e-02, -4.9925e-03,
        -3.1291e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.9478e-02,  5.5483e-03, -9.5211e-03,  1.4873e-02, -3.2769e-02,
        -1.7806e-02, -2.0971e-01, -1.1285e-02,  1.3003e-02,  1.6389e-02,
         1.1637e-03, -1.0790e-02,  9.0582e-03, -2.3811e-03, -6.4123e-03,
        -4.4213e-03, -4.3214e-03,  2.0324e-02, -7.7078e-03, -9.2625e-02,
        -6.9931e-05,  5.6082e-04,  2.1198e-02, -5.3862e-02,  2.8223e-03,
        -5.4698e-02, -2.0172e-02, -1.4540e-02,  7.0125e-03, -2.1357e-02,
        -1.1573e-02, -1.3785e-02, -5.2612e-02, -3.5999e-02, -2.3287e-02,
        -3.4305e-03, -6.8409e-02, -2.8216e-02, -9.5414e-03, -5.9611e-03,
         6.6706e-03, -2.4633e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0139,  0.3005, -0.0125,  0.0123,  0.0008, -0.0059, -0.0103, -0.0079,
         0.0211,  0.0717,  0.0125,  0.0147,  0.0587, -0.0053, -0.0106, -0.0054,
        -0.0073,  0.0524,  0.0748,  0.0042,  0.0131,  0.0149,  0.0024, -0.0013,
        -0.0024,  0.0638,  0.0246,  0.0117,  0.0718, -0.0137, -0.0083,  0.0032,
        -0.0005, -0.0056,  0.0118,  0.0157, -0.0322,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0165, -0.5232, -0.1248, -0.0244, -0.0070,  0.0511, -0.0305, -0.0249,
        -0.0293,  0.0032, -0.0071,  0.0051, -0.0790, -0.0458,  0.0282,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0884,  0.0058, -0.0512, -0.0270,  0.1000,  0.1495,  0.0468,  0.0908,
         0.1813,  0.0283,  0.0068, -0.0119,  0.0679, -0.0009, -0.0590, -0.0275,
         0.0056,  0.0166, -0.0263,  0.0083,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0152, -0.0026,  0.0011,  0.0010, -0.0703, -0.0158,  0.0045,  0.0042,
        -0.0011, -0.0026, -0.0440, -0.0819, -0.0077, -0.0060, -0.0219, -0.0204,
         0.0078, -0.0155,  0.0390, -0.0405, -0.0156, -0.0027, -0.0016, -0.0012,
         0.0039, -0.0616, -0.1258,  0.0022,  0.0037, -0.0579, -0.0887, -0.0108,
         0.0044, -0.0041, -0.0314, -0.0627,  0.0160, -0.0170,  0.0010, -0.0098,
         0.0090,  0.0235, -0.0425,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0055,  0.0019, -0.0198,  0.0058, -0.0498, -0.0267, -0.0420, -0.1558,
        -0.0254,  0.0019, -0.0048,  0.0043, -0.0196, -0.0428, -0.0766, -0.0139,
         0.0024,  0.0043,  0.0060, -0.0007, -0.0427, -0.0106,  0.0062,  0.0170,
        -0.0054, -0.0167,  0.0252,  0.0188, -0.0770, -0.0153, -0.0037, -0.0487,
        -0.0053, -0.0458, -0.0839, -0.0267,  0.0224, -0.0185,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0672, -0.2308, -0.0835, -0.1334,  0.0080, -0.0936, -0.0041,  0.0119,
        -0.0268, -0.0862, -0.0037,  0.0061, -0.0366,  0.0055,  0.0108,  0.0045,
        -0.0263, -0.0657,  0.0320,  0.0265,  0.0335, -0.0033,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0112, -0.1003,  0.0331,  0.0090,  0.0097,  0.0025, -0.0157,  0.0234,
         0.0041, -0.1097, -0.0066,  0.0106, -0.0008, -0.0013, -0.0430, -0.0447,
        -0.0580, -0.0008, -0.0109, -0.0906,  0.0075, -0.0160, -0.0663, -0.0245,
        -0.0213, -0.0045, -0.0185, -0.0018, -0.0002, -0.0761, -0.0082, -0.0322,
        -0.0094,  0.0060, -0.0309, -0.0690,  0.0031, -0.0095, -0.0036,  0.0054,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0024,  0.0012,  0.0013, -0.0142, -0.0023, -0.0425, -0.0079,  0.0016,
        -0.0927, -0.0123, -0.0008,  0.0010, -0.0394, -0.2186, -0.0815, -0.0207,
         0.0020,  0.0248, -0.0658,  0.0018, -0.0038,  0.0147,  0.0025, -0.0103,
        -0.0281, -0.1170, -0.0100, -0.0437, -0.0637, -0.0113, -0.0108, -0.0494,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.0474e-02, -3.3251e-01, -1.7079e-02, -1.7171e-03, -5.2147e-03,
         8.6589e-04, -1.5503e-02, -5.7460e-02, -9.7123e-02, -4.2728e-03,
        -1.8763e-03, -4.1556e-02, -1.8641e-02, -1.8116e-02, -9.5247e-03,
        -2.2218e-02, -5.3797e-02, -2.0682e-02,  1.3072e-02, -1.1519e-02,
        -3.1712e-03,  6.1812e-03, -1.7632e-02, -7.2356e-02, -2.6296e-03,
         4.7017e-03, -2.5034e-02, -3.4350e-03,  5.5594e-03, -1.7279e-02,
         8.6562e-03,  1.4239e-04,  1.6554e-03,  1.8432e-02,  9.9176e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.1068,  0.0008, -0.0672, -0.1355, -0.0446, -0.0335, -0.1887, -0.1531,
         0.0083, -0.0860, -0.0116, -0.0361, -0.0086, -0.0298, -0.0128, -0.0134,
        -0.0158, -0.0023,  0.0103, -0.0202,  0.0146,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0348, -0.3751,  0.0094, -0.0685, -0.0603, -0.0576, -0.0115, -0.0447,
        -0.1185,  0.0308,  0.0039, -0.0063,  0.0211,  0.0050, -0.0230,  0.0171,
        -0.0199, -0.0457,  0.0470,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0209, -0.1618, -0.1350, -0.0394, -0.0493, -0.0024,  0.0245,  0.0619,
        -0.0230, -0.0974, -0.0815,  0.0048, -0.0328, -0.0779, -0.0140, -0.0435,
         0.0178, -0.0080,  0.0429, -0.0614,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.4113e-02, -4.6439e-01, -1.6696e-02, -3.8632e-02,  2.6259e-02,
         4.2410e-03,  5.1891e-03, -3.8247e-02, -3.8409e-03, -1.2472e-04,
         6.8985e-03, -7.4161e-03, -4.7509e-03, -5.5985e-02,  4.2023e-03,
        -3.4547e-02, -8.0370e-02, -1.3510e-02, -1.0146e-02, -7.3523e-02,
         8.1874e-03,  6.4726e-02,  1.4007e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1036, -0.0959, -0.0647,  0.0130, -0.0157, -0.0584, -0.0126, -0.0448,
        -0.0757, -0.1038, -0.0219, -0.0733, -0.0283, -0.0600,  0.0030,  0.0120,
        -0.0094,  0.0036, -0.0324,  0.0079, -0.0503, -0.0007, -0.0098, -0.0794,
         0.0198,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0591,  0.3836,  0.0619,  0.0118, -0.0222, -0.0024, -0.0046, -0.0040,
        -0.0093, -0.0047,  0.0456,  0.0102, -0.0039,  0.0109,  0.0056,  0.0141,
         0.0056,  0.0004,  0.0090, -0.0052,  0.0330,  0.0385,  0.0199,  0.0077,
         0.0265, -0.0088,  0.0090,  0.0128,  0.0009,  0.0396,  0.0351,  0.0089,
         0.0307, -0.0394,  0.0148,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0616, -0.0090,  0.0022, -0.0750, -0.0016, -0.0206, -0.0907, -0.0443,
        -0.1309, -0.0296, -0.0249, -0.0044,  0.0291, -0.1242, -0.0880, -0.1224,
        -0.0170, -0.0901, -0.0074, -0.0039, -0.0132,  0.0098,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1553e-01,  4.1585e-02,  3.2405e-02,  2.5682e-03, -6.6559e-03,
         6.0953e-02, -2.6589e-03,  3.0231e-05, -3.4642e-02,  1.4128e-01,
        -8.8720e-02, -1.6459e-02, -2.6174e-02, -3.1792e-02, -5.4822e-03,
         1.6881e-01,  7.1679e-02, -4.3712e-02, -3.2415e-02,  1.1674e-02,
        -6.4780e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0242, -0.4308, -0.0330, -0.0216, -0.0075, -0.0134, -0.0587, -0.1166,
         0.0107,  0.0053, -0.0030,  0.0077,  0.0195, -0.0441, -0.0064, -0.1018,
         0.0210,  0.0259, -0.0489,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.4624e-02,  8.5421e-03, -3.9647e-02,  1.1603e-03,  1.1329e-02,
        -2.6463e-02, -2.3233e-04, -3.6211e-03,  1.3593e-02, -3.7715e-03,
        -6.5683e-03, -1.4162e-02, -4.0176e-03, -3.9511e-02, -9.6358e-03,
        -6.1487e-03, -4.1097e-02, -7.5130e-02, -1.1720e-02, -3.6867e-02,
        -1.0953e-02, -5.4686e-03, -1.8472e-02,  1.2806e-02, -3.9994e-02,
        -1.6744e-02, -9.7519e-03, -2.9955e-02, -1.3746e-02, -1.1847e-01,
        -1.4491e-02, -9.4497e-03, -4.3374e-02, -1.3552e-01,  3.6578e-03,
        -1.1652e-04, -2.0468e-03, -4.5579e-03, -2.2693e-03, -1.8812e-02,
         9.7269e-04, -3.1689e-03,  7.6071e-03, -2.4339e-03, -8.9493e-04,
         3.9211e-02, -7.2211e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0467, -0.1829, -0.1254, -0.0033,  0.0044,  0.0062, -0.0085, -0.0177,
        -0.0371,  0.0040,  0.0111,  0.0117,  0.0076, -0.0293,  0.0375,  0.0093,
        -0.0630, -0.0135,  0.0169, -0.0068,  0.0117, -0.0104, -0.1141, -0.1215,
        -0.0034, -0.0037, -0.0155,  0.0070,  0.0031, -0.0012,  0.0106,  0.0031,
        -0.0442, -0.0075,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.3421e-04, -4.0282e-01, -2.2016e-02,  2.2215e-05, -3.7166e-02,
        -2.7555e-02,  1.6103e-02,  2.4906e-04, -1.0318e-02, -3.5458e-02,
         1.4125e-02,  2.3877e-03, -5.3647e-02, -1.0394e-01, -1.5684e-02,
         6.0018e-03, -7.4088e-03, -1.2038e-02, -6.6869e-03, -7.2678e-02,
        -5.5917e-03,  9.0857e-03, -3.5338e-03,  4.2917e-03, -6.9477e-03,
         1.6080e-03,  1.8719e-03, -1.0989e-02, -7.6076e-03, -3.2369e-02,
         4.9311e-03,  1.5134e-02, -2.0193e-02,  2.9411e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0681, -0.3101,  0.0089, -0.0127, -0.0326, -0.0893,  0.0346,  0.0140,
        -0.0251, -0.0915, -0.0097, -0.0152, -0.0341, -0.1504, -0.0025,  0.0028,
        -0.0168, -0.0816,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0273,  0.0191,  0.0026,  0.0174, -0.0169,  0.0125,  0.0096, -0.0344,
        -0.1324, -0.2731,  0.0012, -0.0142, -0.0262,  0.0080,  0.0133,  0.0030,
        -0.0065,  0.0085, -0.0231, -0.2307,  0.0270, -0.0339, -0.0061,  0.0129,
         0.0022,  0.0049,  0.0331,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0357,  0.0104, -0.0010,  0.0026, -0.0004, -0.0106,  0.0726,  0.0356,
        -0.1139,  0.0175,  0.0405, -0.0190, -0.2430, -0.0146, -0.0043, -0.0008,
        -0.0489, -0.0218, -0.0049, -0.1259, -0.0162,  0.0186, -0.0414, -0.0095,
         0.0315, -0.0589,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0660, -0.2249,  0.0249, -0.0799, -0.1247, -0.0042, -0.0028, -0.0152,
        -0.0528, -0.0238, -0.0963, -0.0492,  0.0146, -0.0082, -0.0178, -0.0991,
        -0.0237, -0.0113, -0.0098, -0.0084,  0.0424,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.9659e-02, -2.8281e-01, -1.1386e-01, -1.9733e-01, -9.0218e-03,
        -6.7388e-03, -1.8319e-01, -1.2974e-02, -3.2777e-02, -7.9262e-03,
         1.8676e-02, -4.1334e-02,  1.7984e-02,  9.6127e-05, -5.5619e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0304,  0.0276, -0.0207, -0.0297,  0.0242,  0.0440, -0.0149,  0.0205,
        -0.0258,  0.0437,  0.0156,  0.0311,  0.3943,  0.1374, -0.0283,  0.0577,
         0.0317,  0.0224,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0641, -0.2578, -0.0812, -0.1588,  0.0222, -0.0347, -0.0028, -0.0299,
        -0.0175, -0.0439, -0.0864, -0.0440, -0.0203, -0.0689, -0.0036, -0.0034,
        -0.0290, -0.0031,  0.0283,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0372, -0.1226, -0.0936, -0.0142, -0.0221, -0.0182,  0.0210, -0.0074,
        -0.0127, -0.0036, -0.0022,  0.0027, -0.0220,  0.0001,  0.0106,  0.0095,
        -0.0617, -0.0030,  0.0061,  0.0026, -0.0974, -0.0242, -0.0656, -0.0181,
        -0.0510, -0.0016, -0.0386, -0.0146, -0.0118, -0.0292,  0.0003,  0.0071,
        -0.0585, -0.0015, -0.0412, -0.0112, -0.0170,  0.0086, -0.0005,  0.0078,
         0.0064,  0.0042, -0.0037, -0.0067], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0847,  0.2725,  0.0278,  0.0795,  0.0542,  0.0008,  0.0142, -0.0085,
         0.0531,  0.0030,  0.0179,  0.0133,  0.0215,  0.0069,  0.0050, -0.0006,
         0.0091,  0.0483, -0.0207,  0.0072,  0.0238, -0.0069, -0.0259,  0.0053,
        -0.0179,  0.0092,  0.0032,  0.0628, -0.0077, -0.0060,  0.0028,  0.0174,
         0.0041,  0.0351,  0.0231,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0488, -0.3791, -0.0365, -0.0540, -0.0287, -0.0054, -0.1278, -0.0777,
         0.0046, -0.0049,  0.0050, -0.0290, -0.0834,  0.0067,  0.0059, -0.0241,
        -0.0049, -0.0045, -0.0055,  0.0049,  0.0013, -0.0021,  0.0316,  0.0236,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0122, -0.2780,  0.0114, -0.0694, -0.0215, -0.0018,  0.0020, -0.0302,
        -0.1001, -0.0034, -0.0303,  0.0030,  0.0032, -0.0425, -0.0477, -0.0003,
        -0.0171, -0.0091, -0.0576, -0.0131,  0.0116, -0.0390,  0.0066, -0.0226,
        -0.0154, -0.0344, -0.0247, -0.0256, -0.0043, -0.0105, -0.0352, -0.0098,
        -0.0031, -0.0035,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0124,  0.0119, -0.0260, -0.0417, -0.0007,  0.0136, -0.1309, -0.0389,
         0.0239,  0.0073, -0.0047, -0.0045, -0.1043, -0.0278, -0.1125, -0.1764,
         0.0007, -0.0016, -0.0079, -0.0073,  0.0063, -0.0071,  0.0193, -0.0756,
        -0.0397, -0.0179, -0.0568, -0.0032, -0.0056, -0.0135,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0958, -0.0527,  0.0583,  0.1636,  0.0274, -0.0163,  0.0215,  0.0286,
         0.0690,  0.2745, -0.0128, -0.0524,  0.1271,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0687,  0.0270,  0.0356, -0.0363, -0.0022, -0.0386, -0.0004, -0.1400,
        -0.1803, -0.0021,  0.0033, -0.0800, -0.0212, -0.0738, -0.1775, -0.0137,
        -0.0585, -0.0407,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0866, -0.2319, -0.2901, -0.1037, -0.1159,  0.0412, -0.0633,  0.0674,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.0708e-02, -1.7550e-01, -7.0831e-03, -1.9420e-02, -1.0885e-02,
        -1.6821e-02, -4.7481e-03, -4.4124e-03,  3.4665e-03,  5.5999e-03,
         5.1475e-03,  1.7502e-02, -1.1197e-02, -3.9689e-03, -3.0320e-03,
        -5.3863e-03,  4.5929e-03,  1.8360e-03,  3.7809e-07,  3.4070e-03,
         1.1859e-02, -1.7525e-02, -3.9594e-02, -1.4961e-02,  3.6417e-04,
        -1.6541e-02, -1.4215e-02, -1.6896e-02, -7.2862e-02, -1.4544e-02,
         8.5981e-03, -7.8681e-02, -1.8138e-02, -1.1337e-01, -2.8350e-03,
        -3.4089e-02, -5.5850e-02, -3.5436e-03,  1.1631e-02,  2.7140e-03,
        -9.5549e-03, -8.9080e-03, -1.8424e-02,  1.1657e-02,  1.3491e-02,
         2.4666e-03,  3.1978e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0593, -0.2633, -0.2026, -0.2659, -0.0096,  0.0291,  0.0584, -0.0170,
        -0.0392,  0.0385,  0.0170,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1061, -0.1614, -0.0502, -0.0693, -0.0012, -0.0167, -0.0454, -0.1463,
         0.0011, -0.1117, -0.0294, -0.0267, -0.1132, -0.0040, -0.0576,  0.0126,
         0.0170,  0.0301,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0441, -0.1285, -0.3023,  0.0454, -0.0170,  0.0610, -0.0961, -0.0091,
        -0.0657, -0.0354,  0.0130, -0.0828, -0.0231,  0.0204, -0.0562,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0280, -0.4120, -0.0214, -0.0304, -0.0101, -0.0981, -0.1315,  0.0632,
        -0.0192, -0.0525, -0.0630,  0.0307,  0.0297,  0.0014, -0.0087,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.8919e-02,  7.1806e-03,  1.6852e-02,  4.3155e-02,  1.2461e-02,
        -1.0639e-04,  3.9073e-03,  1.9156e-02, -6.1861e-02, -2.9638e-02,
        -2.0184e-03, -7.8906e-02, -1.4386e-01, -1.2235e-02, -2.4642e-04,
        -1.6875e-02, -3.3352e-02,  4.6860e-02, -2.3217e-02,  1.5895e-03,
        -9.4222e-02, -1.3321e-01,  3.2560e-02, -4.2427e-02, -3.9347e-02,
        -7.5836e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0085,  0.0102, -0.0416,  0.0020, -0.0653, -0.0302,  0.0452, -0.1560,
        -0.4153, -0.0375,  0.0423, -0.0395, -0.0032,  0.0997,  0.0033,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0174, -0.2329,  0.0024, -0.0612, -0.0741, -0.0076, -0.0737, -0.0113,
        -0.0190, -0.1714,  0.0295,  0.0231, -0.0548, -0.0170, -0.0195,  0.0004,
        -0.0135,  0.0054, -0.0530,  0.0092, -0.0362,  0.0167,  0.0104,  0.0112,
         0.0217, -0.0075,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.8001e-02, -2.8024e-02,  2.4768e-02,  3.1846e-02, -2.2128e-03,
        -2.1464e-02,  1.3470e-04,  4.4839e-02, -3.9071e-02, -4.0806e-02,
        -2.1810e-01, -3.1179e-02, -1.3364e-03,  5.7489e-03, -8.2470e-02,
        -7.8571e-02,  2.2968e-02, -4.5153e-02, -1.0664e-02, -4.9610e-02,
        -1.2220e-02, -1.4904e-02, -4.1394e-03, -6.9621e-02,  3.5287e-02,
        -1.6866e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0112, -0.2214,  0.0125,  0.0086, -0.0130,  0.0062, -0.0321, -0.0469,
        -0.0085, -0.0400, -0.0053, -0.0326, -0.0615,  0.0018, -0.0250, -0.0056,
        -0.0121, -0.0284, -0.0163, -0.0315,  0.0069, -0.0352,  0.0073, -0.1034,
        -0.0857, -0.0037, -0.0188, -0.0076,  0.0165,  0.0016, -0.0009, -0.0297,
         0.0080,  0.0179,  0.0006,  0.0359,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0161, -0.1026, -0.0465,  0.0015, -0.0745,  0.0128, -0.0547, -0.0700,
         0.0018,  0.0286, -0.0302,  0.0040, -0.0165,  0.0040, -0.0010, -0.0128,
         0.0022, -0.0088,  0.0144,  0.0019, -0.0005, -0.0740, -0.0085, -0.0164,
        -0.0589, -0.0731, -0.0026, -0.0498, -0.0482,  0.0096, -0.0054, -0.0359,
        -0.0425,  0.0215, -0.0188,  0.0294,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0012, -0.0671,  0.0232, -0.0112,  0.0126, -0.1890, -0.0700, -0.0078,
        -0.0009,  0.0009, -0.0318, -0.0431, -0.2678, -0.0363,  0.0028,  0.0077,
        -0.0107, -0.0173, -0.0235,  0.1752,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0084, -0.1870, -0.0301, -0.0195,  0.0118,  0.0141, -0.0021, -0.0672,
        -0.0053, -0.0389, -0.0067, -0.0065,  0.0044, -0.0023, -0.0942, -0.0202,
        -0.0053, -0.0115,  0.0047, -0.0034, -0.0113, -0.0055,  0.0027,  0.0242,
        -0.0374,  0.0010, -0.0031,  0.0034,  0.0198, -0.0195, -0.0352, -0.0203,
         0.0088, -0.1253,  0.0208,  0.0583, -0.0187, -0.0407,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0141, -0.2177, -0.1004, -0.0919, -0.0134,  0.0080, -0.0033, -0.0219,
         0.0052, -0.0096, -0.0455, -0.0616, -0.0022,  0.0097, -0.0005, -0.0457,
         0.0051, -0.0733, -0.0017, -0.0077,  0.0092,  0.0117,  0.0174, -0.0046,
         0.0109, -0.0006, -0.0084, -0.0190, -0.0614, -0.0073, -0.0146, -0.0141,
        -0.0032, -0.0022, -0.0181,  0.0056, -0.0060,  0.0046,  0.0114,  0.0176,
        -0.0136], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0724, -0.3851,  0.0019,  0.0139,  0.0145, -0.0134, -0.0021, -0.0239,
        -0.0072, -0.0022, -0.0137, -0.0511, -0.0671, -0.0130,  0.0011, -0.0910,
         0.0022, -0.0459,  0.0031, -0.0168, -0.1400,  0.0183,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0065,  0.0085,  0.0128,  0.0100,  0.0269, -0.0227, -0.0029,  0.0078,
         0.0056,  0.0086, -0.0007, -0.0542,  0.0011, -0.0261, -0.0217, -0.0627,
         0.0020, -0.0412, -0.0201, -0.0705, -0.1346,  0.0083,  0.0179, -0.0678,
        -0.0070,  0.0103, -0.0305, -0.0512, -0.0727,  0.0229, -0.0543, -0.0561,
        -0.0092,  0.0004,  0.0441,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0182, -0.0559, -0.1886, -0.0189, -0.0841, -0.0050,  0.0106,  0.0338,
         0.0616, -0.0140, -0.0421,  0.0009, -0.0449,  0.0058,  0.0008,  0.0093,
         0.0602, -0.0175, -0.0403,  0.0078, -0.0438,  0.0420, -0.0137, -0.0607,
        -0.0035, -0.0106, -0.0591, -0.0369, -0.0094,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0336, -0.3980, -0.0720, -0.0197, -0.0094, -0.0851, -0.0013, -0.0181,
        -0.0089, -0.0244, -0.0296, -0.0439,  0.0104,  0.0145,  0.0007, -0.0417,
        -0.0415, -0.0221, -0.0377, -0.0025, -0.0233, -0.0161, -0.0125, -0.0008,
        -0.0322,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0499, -0.0355, -0.0114, -0.0381, -0.1728, -0.0076, -0.0092, -0.0943,
        -0.1439, -0.0494, -0.0044,  0.0087, -0.0094,  0.0025, -0.0733, -0.1608,
         0.0115, -0.0381, -0.0162,  0.0025, -0.0308, -0.0147,  0.0151,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.8997e-02, -1.8945e-03, -1.9608e-03, -5.0358e-03,  1.5581e-02,
         1.4104e-02,  1.5068e-02, -8.9187e-03, -5.8361e-04, -3.7844e-03,
         3.0647e-03, -3.6665e-03, -5.0694e-02, -1.2591e-01,  1.0539e-03,
        -1.4735e-02, -9.7741e-05, -5.2028e-02, -5.1251e-03,  2.9062e-04,
         2.5591e-02, -1.6850e-02, -6.1091e-03, -8.8972e-03, -1.2058e-02,
        -4.6846e-03, -1.5754e-01, -3.3991e-02, -9.8124e-02, -2.4886e-02,
        -6.5023e-02,  6.1236e-04, -1.4965e-02, -2.5396e-02, -2.7196e-02,
         9.4404e-03, -6.1825e-03,  2.8358e-02, -2.1503e-02,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1340,  0.0463,  0.0466, -0.0670, -0.2830, -0.0468, -0.0138, -0.1226,
        -0.0876,  0.0440, -0.0716, -0.0367,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 9.9642e-02, -3.1801e-02, -3.3875e-01,  9.9163e-03, -1.8492e-02,
        -2.6863e-02, -4.9986e-02,  2.2223e-02, -1.5259e-03, -3.0667e-02,
        -3.8330e-03,  2.0800e-02, -5.3839e-03,  8.7962e-03, -3.5280e-02,
         7.3627e-03, -1.1706e-02, -4.0189e-02, -9.7786e-05, -8.8567e-03,
        -1.5199e-02, -1.8018e-02,  4.4171e-03,  6.2475e-03, -7.1997e-03,
         3.2246e-03,  5.4701e-04, -1.1652e-02, -6.4070e-03, -2.4603e-02,
        -1.2491e-02,  1.1269e-02, -3.6139e-02,  2.9079e-03, -3.3670e-04,
         9.0036e-04, -2.3968e-02, -2.4616e-03, -4.1115e-04,  3.1055e-04,
         1.3220e-02,  8.7802e-03,  4.4663e-03,  4.0671e-03, -8.5901e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0088,  0.4248, -0.0105,  0.0973, -0.1583, -0.0056,  0.0186,  0.0368,
         0.1320, -0.0423, -0.0017, -0.0633,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1353, -0.5825, -0.0668, -0.1126, -0.0472, -0.0309, -0.0087,  0.0159,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.8486e-02,  1.9029e-01,  4.8389e-02,  1.5222e-03,  6.9211e-02,
        -2.5838e-03,  7.9266e-03,  6.3185e-03,  1.5595e-02,  5.6466e-02,
         2.3114e-03,  1.0355e-02,  3.7035e-03, -2.5263e-03, -7.4588e-03,
        -7.0415e-03, -2.4321e-02,  6.0640e-03, -3.6070e-03, -2.1938e-03,
        -1.5994e-03, -3.3865e-03,  5.6217e-03, -1.0020e-03,  7.3304e-03,
         3.6659e-03, -6.8036e-03,  4.3992e-03,  1.9743e-02, -7.0545e-03,
        -1.7101e-04,  1.1562e-02, -9.2404e-04,  1.0928e-02,  8.6400e-03,
        -2.3822e-03, -3.6483e-03,  5.1311e-03, -4.7896e-03,  1.8654e-03,
         5.1653e-03, -1.3249e-03, -7.1301e-03,  9.0016e-03,  6.4495e-03,
        -3.8509e-02,  4.8299e-02, -1.0876e-02,  5.7785e-02,  1.0421e-01,
         2.7737e-02, -6.0147e-03,  2.1399e-02,  3.3833e-03,  2.2505e-03,
        -7.8709e-04,  1.8934e-02, -1.3849e-04,  6.4987e-03,  3.6396e-03,
        -2.3440e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0375, -0.0206, -0.0051, -0.0407, -0.1595,  0.0308, -0.0491, -0.0046,
         0.0005,  0.0017, -0.0051, -0.0365,  0.0071, -0.0345, -0.0569, -0.0059,
        -0.0010, -0.0057, -0.0446, -0.0334, -0.0805, -0.0951, -0.0134,  0.0111,
        -0.0514, -0.0592, -0.0094, -0.0206, -0.0356, -0.0151, -0.0020,  0.0196,
         0.0061,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.4738e-02, -4.6527e-02, -1.2277e-01, -9.7871e-02, -4.8112e-02,
        -1.9733e-02, -1.0446e-01,  2.8957e-03,  4.4383e-03,  2.2586e-04,
         1.6427e-02,  6.4706e-03, -3.6556e-02, -8.7875e-03, -1.3549e-02,
         3.6843e-02, -1.3543e-02, -6.2726e-02, -2.2743e-01,  8.2627e-03,
        -2.3269e-02,  1.9548e-02,  1.5944e-02, -4.8865e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0028, -0.1683, -0.1240, -0.0270, -0.0376, -0.1278, -0.0125, -0.0472,
        -0.0056, -0.0691, -0.0749, -0.0537, -0.0940, -0.0251,  0.0078,  0.0030,
         0.0103,  0.0146,  0.0081,  0.0053, -0.0224,  0.0036,  0.0163,  0.0390,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0324,  0.0334, -0.0434, -0.0459,  0.0133,  0.0146, -0.0235, -0.1169,
        -0.0042, -0.0146,  0.0143, -0.0010, -0.0006,  0.0004, -0.0379, -0.0692,
        -0.0695,  0.0171, -0.0182, -0.0077, -0.0790, -0.0006, -0.0473, -0.0864,
        -0.0005,  0.0053, -0.1045, -0.0051,  0.0010, -0.0190, -0.0292, -0.0198,
        -0.0015,  0.0226,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1125, -0.0030,  0.0106,  0.0164, -0.0959, -0.0255,  0.0015,  0.0023,
        -0.0259,  0.0160,  0.0020,  0.0370, -0.0246, -0.1994, -0.0206, -0.0112,
        -0.0767,  0.0364, -0.0495, -0.0448, -0.0298, -0.0655, -0.0135,  0.0241,
        -0.0553,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1465,  0.0029, -0.0317, -0.0177, -0.0354, -0.0147, -0.1688, -0.0618,
        -0.0752, -0.2342,  0.0312, -0.0055, -0.0194, -0.0331,  0.0516, -0.0703,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.1421e-02, -2.7382e-01, -2.0632e-02, -9.4559e-02, -4.4418e-02,
        -3.6941e-02, -1.2380e-01, -2.6397e-02,  7.5799e-03, -8.4843e-03,
         2.9848e-03, -1.1315e-02, -4.2862e-02, -8.2271e-03,  1.0446e-02,
         9.1183e-03, -4.5806e-02, -5.1851e-02, -3.5244e-02,  1.6619e-04,
         2.2921e-02, -8.9891e-02,  1.1171e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0559,  0.0227,  0.0541,  0.2453,  0.2987,  0.0388, -0.0455,  0.0140,
         0.0323, -0.0435, -0.0058, -0.0516, -0.0500,  0.0418,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0508, -0.0933, -0.0522, -0.1090, -0.0171,  0.0155, -0.0302,  0.0155,
        -0.0357, -0.0174, -0.0841, -0.0871, -0.0317, -0.0082, -0.0246,  0.0203,
        -0.0020, -0.0953,  0.0187, -0.0768, -0.0027, -0.0092, -0.0265,  0.0760,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0860,  0.0196, -0.2211, -0.0360, -0.2535,  0.0180, -0.2478,  0.0360,
        -0.0407, -0.0010,  0.0403,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0779, -0.0095,  0.0697, -0.0009,  0.0170, -0.0639, -0.2174, -0.0089,
        -0.1115, -0.0122, -0.0537,  0.0267, -0.0466, -0.0029, -0.0809, -0.0130,
        -0.0135, -0.0898,  0.0346, -0.0477,  0.0017,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0108,  0.0164,  0.0200, -0.0136,  0.0165,  0.0112, -0.0269, -0.3092,
        -0.1238, -0.2297, -0.0933,  0.0375,  0.0233, -0.0679,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0165,  0.0164,  0.0298, -0.0252, -0.0078, -0.0177, -0.0169, -0.0377,
        -0.0090, -0.0171,  0.1046,  0.0734,  0.3790,  0.0274,  0.0070,  0.0202,
         0.0226,  0.0366,  0.0011, -0.0436,  0.0904,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0962,  0.2927,  0.0163,  0.0933,  0.0212,  0.0318,  0.0005,  0.0657,
         0.0557,  0.0575,  0.1640, -0.0657,  0.0392,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0314, -0.0858, -0.0087, -0.0119, -0.0032, -0.0169, -0.0453, -0.0887,
        -0.0168, -0.0048, -0.0107,  0.0136, -0.0058,  0.0066,  0.0005, -0.0118,
        -0.0059, -0.0101,  0.0039, -0.0099,  0.0042,  0.0243,  0.0079, -0.0103,
        -0.1015,  0.0258, -0.1126, -0.1123, -0.0488,  0.0100, -0.0055, -0.0058,
         0.0013, -0.0129,  0.0152,  0.0027, -0.0346,  0.0010,  0.0020, -0.0104,
        -0.0092, -0.0082,  0.0022, -0.0171,  0.0220], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0079, -0.0341, -0.0198, -0.0114,  0.0011,  0.0229,  0.0219, -0.0030,
        -0.0038, -0.0035, -0.0784, -0.0151, -0.0247, -0.0074, -0.0422,  0.0054,
        -0.0939,  0.0223, -0.0091,  0.0097,  0.0392, -0.0755, -0.0101, -0.0512,
        -0.0114,  0.0117, -0.0507, -0.0515, -0.0766, -0.0004, -0.0120, -0.0319,
        -0.1282, -0.0120,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1064,  0.0138,  0.0130, -0.0288, -0.0164, -0.0097, -0.0011,  0.0021,
         0.0066, -0.0051, -0.0198, -0.0017,  0.0039,  0.0035,  0.0016,  0.0626,
         0.0247,  0.0245, -0.0075, -0.2052, -0.3247, -0.0108,  0.0034,  0.0037,
         0.0231, -0.0764,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0304, -0.3115,  0.0310, -0.1225,  0.0319, -0.0339,  0.0039, -0.0388,
         0.0367, -0.0064,  0.0015,  0.0046, -0.0035,  0.0076, -0.0051, -0.0875,
        -0.1107,  0.0297, -0.0217, -0.0060, -0.0548, -0.0182, -0.0021,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0165,  0.0893,  0.0050,  0.0207, -0.0067, -0.0165,  0.0065,  0.0572,
         0.0831,  0.0223, -0.0092,  0.0018,  0.0058,  0.0064,  0.1705, -0.0204,
         0.0050,  0.0427,  0.0706,  0.0221,  0.0121,  0.0058,  0.0002, -0.0122,
        -0.0055,  0.0154,  0.0544,  0.0949,  0.0214,  0.0135,  0.0059,  0.0095,
        -0.0011, -0.0134,  0.0029,  0.0069,  0.0165,  0.0002,  0.0093,  0.0206,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0362, -0.0187, -0.0178, -0.0322, -0.0825, -0.0248, -0.0825, -0.0781,
         0.0011, -0.0260, -0.0103, -0.0095, -0.0407, -0.0084, -0.0247, -0.0539,
        -0.0451, -0.0049, -0.0055, -0.0347, -0.0104, -0.0664, -0.0285, -0.0056,
        -0.0137, -0.0109, -0.0068, -0.0031, -0.0160, -0.0251, -0.0532,  0.0019,
        -0.0882, -0.0323,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0101, -0.0538, -0.0452, -0.0103, -0.0506,  0.0244,  0.0073, -0.0069,
         0.0040,  0.0040, -0.0032, -0.0117,  0.0102,  0.0080, -0.1300, -0.0027,
        -0.0174,  0.0048, -0.0761, -0.0107,  0.0153,  0.0171, -0.0069,  0.0202,
        -0.0168, -0.0762,  0.0259,  0.0034, -0.0040, -0.0069, -0.0053,  0.0130,
         0.0015, -0.0094, -0.0282,  0.0119, -0.0128, -0.0273, -0.0186, -0.0312,
        -0.0366, -0.0008, -0.0445, -0.0086, -0.0081,  0.0041,  0.0092, -0.0002,
        -0.0129, -0.0140, -0.0012, -0.0164], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0145,  0.0103,  0.0215, -0.0035, -0.1655, -0.0492,  0.0142,  0.0178,
         0.0099, -0.0098, -0.0021,  0.0165, -0.0227, -0.0922, -0.0247, -0.0624,
        -0.0850, -0.0700, -0.0551, -0.0041, -0.0233, -0.0332, -0.0242, -0.0590,
         0.0063,  0.0013, -0.0204, -0.0042,  0.0322,  0.0101, -0.0348,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0641, -0.1590,  0.0294, -0.0468,  0.0111,  0.0119, -0.0076, -0.0042,
        -0.0217,  0.0065,  0.0051, -0.0860, -0.0373, -0.0027, -0.0520, -0.0665,
         0.0122, -0.0617, -0.0708, -0.1015, -0.0088, -0.0289, -0.0229,  0.0207,
        -0.0608,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2165, -0.0034,  0.0113,  0.0130, -0.0582, -0.1644, -0.3577, -0.0461,
        -0.0044, -0.0239, -0.0130,  0.0882,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0270, -0.3271, -0.0126, -0.0257, -0.0605, -0.0122,  0.0090, -0.0015,
         0.0039, -0.0337,  0.0154, -0.0091, -0.0628, -0.0888, -0.0275, -0.0382,
         0.0068, -0.0044,  0.0085,  0.0006, -0.0317,  0.0301, -0.0021, -0.0289,
        -0.0449, -0.0622, -0.0105,  0.0008,  0.0134,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0831, -0.3199, -0.0486, -0.0019, -0.0389, -0.0625, -0.3546,  0.0064,
         0.0297,  0.0188,  0.0101,  0.0255,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0081, -0.2161,  0.0096, -0.0856,  0.0867, -0.1684, -0.2907,  0.0281,
        -0.0714,  0.0186, -0.0049,  0.0117,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0670,  0.0537, -0.1436, -0.0029, -0.0909, -0.0833, -0.0008,  0.0026,
        -0.1094, -0.0431, -0.0131, -0.0587, -0.0040,  0.0102, -0.0394, -0.1160,
         0.0137,  0.0286, -0.0663, -0.0526,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.7642e-02, -2.3271e-01, -9.4492e-03, -5.8777e-02, -1.3727e-01,
        -4.2594e-06, -1.7508e-02, -2.0171e-02, -3.4028e-02, -4.5285e-04,
        -1.1120e-02, -4.0801e-02, -1.7569e-02, -1.2639e-02, -1.3296e-02,
         1.7741e-03, -1.3897e-02,  3.0996e-03, -2.9620e-02, -1.3518e-02,
        -1.3560e-02,  5.4264e-03,  1.5834e-03, -1.0966e-03,  6.4715e-03,
         6.5661e-03,  8.3124e-03, -4.4403e-02, -4.1543e-02,  5.8625e-03,
         3.1415e-03, -1.3204e-03, -5.2383e-02, -8.0652e-02, -3.4664e-03,
         2.3775e-02,  5.0984e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0815, -0.7744,  0.0247,  0.0028, -0.0579,  0.0029, -0.0103, -0.0337,
        -0.0119,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0002, -0.0009,  0.0664, -0.1251, -0.0419, -0.0453, -0.0569, -0.0742,
         0.0107, -0.0159, -0.0316, -0.0407, -0.0845, -0.0940,  0.0243, -0.0034,
         0.0017,  0.0500, -0.0733, -0.0907, -0.0049,  0.0025, -0.0067, -0.0072,
         0.0189,  0.0055,  0.0224,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.8222e-02, -2.9759e-01,  2.0901e-02, -7.1357e-02, -7.4369e-03,
        -2.1932e-02, -5.9746e-04, -9.4965e-03,  8.8749e-04, -1.5279e-02,
        -9.4273e-05, -7.6068e-02, -2.6952e-02, -4.0595e-03, -6.5573e-03,
        -4.4103e-03, -5.0890e-02, -6.9454e-03,  1.3612e-02, -8.2255e-03,
        -1.6736e-02, -3.2128e-02, -5.5126e-02, -1.6707e-02, -9.8036e-03,
        -4.0503e-03, -7.1819e-04, -2.6790e-02, -1.3133e-01,  1.0165e-03,
         1.7897e-03,  4.7188e-03, -1.0685e-02,  6.8949e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 3.3181e-02,  4.1977e-04,  2.8061e-03, -2.1385e-02, -1.9057e-02,
        -1.4885e-02,  4.6571e-03, -2.0616e-03,  9.4045e-03,  4.0809e-02,
        -6.7987e-03, -9.1855e-02,  7.7085e-03,  2.6149e-03, -8.5565e-03,
        -8.0477e-02, -1.2150e-01, -1.6199e-02, -3.1301e-02,  1.8714e-02,
         1.4586e-02, -1.9373e-03, -4.1863e-02, -3.6025e-03,  6.7458e-03,
        -9.6564e-03,  3.1244e-03, -7.4987e-03,  6.3977e-05, -1.9812e-03,
        -1.1241e-02, -5.1778e-02, -6.7972e-02, -2.0897e-03, -1.8893e-02,
        -2.7748e-02, -7.9483e-02,  7.5065e-03,  1.2345e-02, -7.9023e-03,
        -2.0276e-02,  1.2947e-03,  1.2980e-02,  2.0613e-02,  3.2421e-02,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.5711e-02,  5.3768e-03, -7.4936e-03, -1.1119e-02, -6.5476e-02,
        -1.8556e-01, -3.0446e-01, -1.2317e-02,  1.3362e-02, -1.5362e-01,
        -6.3466e-02, -1.5072e-04, -1.0221e-01, -3.5596e-03,  9.0283e-04,
        -5.2243e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0484, -0.0288,  0.0237,  0.0140,  0.0085, -0.0082,  0.0388,  0.0316,
         0.2004,  0.1432,  0.0752,  0.1237, -0.0171,  0.0629,  0.0222, -0.0064,
         0.0113, -0.0052, -0.0249,  0.0039,  0.0074, -0.0302, -0.0040, -0.0060,
        -0.0114,  0.0151, -0.0278,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0156, -0.0233, -0.0730, -0.0023, -0.0211,  0.0075, -0.0209, -0.0129,
        -0.0193,  0.0083,  0.0075,  0.0043, -0.0226, -0.0012, -0.0136,  0.0054,
        -0.0210,  0.0136,  0.0229, -0.0201,  0.0144, -0.0286, -0.0113, -0.0874,
        -0.0044,  0.0029, -0.0505, -0.0595,  0.0344, -0.0418,  0.0152, -0.0390,
        -0.0544, -0.0130,  0.0018, -0.0068, -0.0204, -0.0442, -0.0036,  0.0109,
        -0.0153, -0.0463, -0.0014, -0.0095,  0.0163,  0.0303], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0775,  0.0165,  0.0120,  0.0065, -0.0085, -0.0270, -0.0072, -0.1273,
         0.0284, -0.0314,  0.0028,  0.0073,  0.0102, -0.0082,  0.0009, -0.0140,
         0.0016, -0.0146, -0.1357, -0.2190, -0.0237, -0.0091, -0.0071, -0.0258,
         0.0239, -0.0259, -0.1024,  0.0024,  0.0048,  0.0184,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1094, -0.1825, -0.0872, -0.0358, -0.0401,  0.0210,  0.0102,  0.1664,
        -0.0325, -0.0502,  0.2648,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.7035e-02, -4.6692e-01,  3.8673e-04, -6.9926e-03, -6.1818e-02,
        -2.9825e-02, -6.1104e-02, -1.1654e-01,  8.1507e-03,  1.8806e-02,
         1.2811e-02,  1.4252e-02, -1.5549e-02, -2.0784e-02, -1.6801e-02,
        -6.7838e-02, -1.8746e-02, -4.2190e-02, -5.4869e-04,  2.9026e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0094,  0.3348,  0.1542,  0.0988, -0.0634,  0.0443,  0.0978, -0.0205,
         0.0420,  0.0322, -0.1027,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0047,  0.3014,  0.1466,  0.1442,  0.0312,  0.0246,  0.0069,  0.0451,
         0.0791,  0.0162, -0.0032,  0.0182, -0.0065,  0.0274, -0.0110,  0.0635,
         0.0022,  0.0083, -0.0026,  0.0026, -0.0546,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.2100e-02, -2.8058e-01, -8.4902e-02,  2.0721e-02,  5.6056e-03,
        -3.1927e-01,  6.8764e-02,  1.3887e-02, -3.2429e-02, -1.2001e-02,
         5.3214e-03, -9.8577e-03, -1.0070e-02, -9.8945e-03, -5.2589e-02,
         2.9485e-04,  3.2113e-02,  9.5963e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1202, -0.0238, -0.1071, -0.0104, -0.0022,  0.0112,  0.0092,  0.0042,
        -0.0008, -0.0085, -0.0015,  0.0053, -0.0139, -0.0757, -0.0727,  0.0064,
        -0.0169, -0.0634, -0.0238, -0.0288, -0.0173, -0.0463, -0.0817, -0.0038,
        -0.0216, -0.0087, -0.0441, -0.0079, -0.0146, -0.0559,  0.0116, -0.0003,
        -0.0017,  0.0064, -0.0098,  0.0049, -0.0019,  0.0257, -0.0297,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.4174e-02, -4.1664e-01, -1.8902e-02, -8.7487e-02, -5.3015e-03,
        -6.4975e-04,  1.4680e-02,  1.8918e-03, -3.2037e-02, -1.6100e-02,
        -6.3982e-03, -5.0851e-02,  2.0975e-04, -7.4575e-05,  7.7939e-03,
         1.4689e-03, -4.2987e-02, -3.8238e-03,  9.4647e-03, -9.6964e-03,
        -2.5771e-03, -1.0768e-03,  8.1337e-03,  2.0678e-03,  1.3087e-03,
        -4.8036e-03, -1.2839e-02, -7.1347e-03, -3.7351e-02, -4.4793e-02,
        -2.8363e-04, -1.1980e-02,  1.5630e-03, -5.3460e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.0432, -0.1214, -0.2465, -0.3226,  0.0313, -0.1034,  0.0293,  0.0115,
        -0.0908,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1287e-02, -2.2679e-01, -1.1971e-02, -5.9199e-02, -9.1808e-03,
        -8.7068e-02, -1.2376e-02, -4.3634e-02, -1.1353e-02,  2.5905e-03,
        -2.3787e-02, -4.6128e-03, -7.9087e-03, -2.3876e-02, -1.3739e-02,
        -7.3310e-02,  3.4663e-03, -8.4279e-02, -7.2627e-04, -6.8628e-05,
        -1.5912e-02,  1.8922e-03,  1.6069e-03, -5.1896e-03, -2.5287e-02,
         4.4529e-03, -1.9992e-02, -1.1295e-02,  3.9176e-03, -2.5357e-03,
        -2.3026e-02, -2.7477e-02,  9.2525e-03, -5.4625e-02,  6.7696e-03,
        -6.5023e-03, -1.7618e-02,  8.2267e-03, -1.0562e-02,  3.2635e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0109, -0.0817, -0.1643, -0.0080, -0.0337,  0.0045,  0.0032, -0.0036,
        -0.0189, -0.0492,  0.0003,  0.0005,  0.0004, -0.0011, -0.0090,  0.0066,
        -0.0098, -0.0226, -0.0529, -0.0039, -0.0116,  0.0063,  0.0233,  0.0051,
         0.0133, -0.0155, -0.0589, -0.0041, -0.0593, -0.1007, -0.0088,  0.0108,
        -0.0063,  0.0104,  0.0071, -0.0092, -0.0312, -0.0652, -0.0045,  0.0038,
        -0.0096,  0.0086, -0.0011,  0.0021,  0.0041,  0.0087, -0.0017, -0.0007,
        -0.0031, -0.0036,  0.0122,  0.0039,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0171,  0.3057,  0.0282,  0.0293,  0.0172,  0.0068,  0.0555,  0.0041,
         0.0061, -0.0009, -0.0136,  0.0105,  0.1052,  0.0187,  0.0580,  0.0602,
         0.0796,  0.0466,  0.0760,  0.0028, -0.0335, -0.0241,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5449e-02,  5.7883e-01, -1.9950e-02, -1.2169e-02,  4.2187e-04,
        -7.8370e-03,  5.4594e-02,  4.8034e-02,  2.8210e-02, -1.7514e-03,
         9.5885e-03,  4.1097e-02,  1.1015e-03,  1.4184e-03, -1.5924e-02,
         5.1547e-02,  2.4654e-02,  1.5477e-03, -1.3069e-02, -4.8202e-02,
         2.4607e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.7752e-02, -2.2649e-01, -4.0236e-02,  7.3487e-04, -3.9886e-04,
         1.6450e-02, -2.6761e-03,  1.2035e-02, -2.0518e-02, -3.7035e-03,
         3.5940e-03, -9.1741e-03,  6.8966e-04,  1.8846e-03,  1.0796e-02,
         1.9966e-03,  5.8319e-03,  8.8795e-03, -2.9873e-02,  2.7489e-03,
         1.2883e-03,  2.1073e-04, -1.7795e-03, -2.3611e-04, -6.2046e-03,
        -4.2612e-03, -8.4694e-03, -3.4129e-02,  4.8498e-03, -3.0788e-02,
        -5.0984e-02, -4.6366e-02,  1.3795e-03, -2.1994e-02, -3.4983e-02,
        -7.2821e-03,  1.4465e-03,  7.0869e-03,  1.4118e-02, -8.3791e-03,
        -9.2755e-02,  2.5435e-03, -3.8278e-03,  9.9598e-03, -4.0643e-02,
         1.2438e-02, -3.0508e-02, -1.1696e-02,  1.1510e-02,  9.0846e-03,
        -1.3038e-02, -1.4266e-03,  1.3138e-02,  1.0892e-02,  3.8393e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0261, -0.0067,  0.0381, -0.0188, -0.0051, -0.0290, -0.0368,  0.0317,
        -0.0068,  0.0095,  0.0012, -0.0101, -0.0160, -0.0233, -0.0395, -0.0225,
        -0.0215, -0.0344, -0.0195, -0.0553, -0.0031, -0.0172, -0.0425, -0.0028,
         0.0178,  0.0143, -0.0079, -0.0752,  0.0078, -0.0018,  0.0153, -0.0057,
        -0.0187, -0.0789, -0.0141, -0.0226,  0.0153, -0.0413, -0.0408, -0.0281,
        -0.0024, -0.0014, -0.0082, -0.0022, -0.0626,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0625, -0.1848, -0.1043, -0.0702,  0.0304, -0.0054, -0.0180,  0.0037,
        -0.0490,  0.0060,  0.0393, -0.0280,  0.0103, -0.0505, -0.0711,  0.0372,
        -0.0128, -0.0474,  0.0397, -0.0248, -0.0691,  0.0023, -0.0137,  0.0197,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0081,  0.0500, -0.0016,  0.0109,  0.0033, -0.0176, -0.0588, -0.0717,
        -0.0003,  0.0155, -0.0036, -0.0070,  0.0188, -0.0313, -0.0121,  0.0303,
        -0.0502, -0.0461, -0.0248, -0.0416, -0.0154, -0.0132, -0.0512, -0.0362,
        -0.0047, -0.0712, -0.1078, -0.0257, -0.0469, -0.0732, -0.0170,  0.0077,
        -0.0260,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0364, -0.4832, -0.0062, -0.0944,  0.0030, -0.0270,  0.0113, -0.0306,
        -0.1396, -0.0026, -0.0018, -0.0082,  0.0328, -0.0598,  0.0191, -0.0098,
        -0.0206, -0.0030,  0.0107,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0446,  0.0060,  0.0259, -0.0280, -0.1173, -0.0256, -0.1141, -0.0939,
        -0.1547, -0.0161,  0.0202,  0.0004, -0.0184,  0.0047,  0.0021, -0.0198,
        -0.0120, -0.0216,  0.0024,  0.0324, -0.0542, -0.1183, -0.0224,  0.0179,
        -0.0270,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0179, -0.1659, -0.1772, -0.0090, -0.0260, -0.1017, -0.2476,  0.0183,
        -0.0101, -0.0345, -0.0096,  0.0147, -0.0237, -0.0094, -0.0030,  0.0776,
         0.0538,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0628,  0.0017, -0.0129, -0.0129,  0.0185, -0.0097,  0.0088, -0.0146,
        -0.0173, -0.0765, -0.0548, -0.0211,  0.0048, -0.0157,  0.0069, -0.0050,
        -0.0519, -0.1232,  0.0158, -0.0327, -0.0100, -0.0878, -0.1789, -0.0118,
        -0.0189,  0.0007, -0.0057,  0.0225,  0.0306, -0.0038,  0.0180,  0.0438,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0169, -0.0858, -0.2036,  0.0092, -0.0223, -0.0025, -0.0483, -0.1086,
        -0.0092, -0.0050,  0.0046,  0.0041,  0.0896, -0.0022, -0.0391,  0.0108,
        -0.0178,  0.0092, -0.0377, -0.0135, -0.0146,  0.0005, -0.0512, -0.1404,
        -0.0200, -0.0021,  0.0152, -0.0160,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0605, -0.0583, -0.0106,  0.0227, -0.0184, -0.0575, -0.1886, -0.0251,
         0.0117,  0.0142,  0.0029, -0.0834, -0.0399, -0.0414, -0.0062, -0.0683,
        -0.0086, -0.0471,  0.0074, -0.0116, -0.0269,  0.0094,  0.0077, -0.0089,
        -0.0559,  0.0150, -0.0026, -0.0518,  0.0374,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0356, -0.0464, -0.0309, -0.0702, -0.1192, -0.0082, -0.0295, -0.0084,
         0.0002,  0.0299,  0.0058, -0.0059, -0.0530, -0.1620,  0.0445, -0.0472,
        -0.0048, -0.0648, -0.0060, -0.0269,  0.0075, -0.0053, -0.0276, -0.0095,
        -0.0288, -0.0353, -0.0407, -0.0042, -0.0129,  0.0037, -0.0047,  0.0007,
        -0.0195,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.1839e-02, -3.1192e-01, -6.2699e-03, -5.0329e-02, -3.2263e-02,
        -7.3504e-02, -4.5475e-03, -1.9260e-02,  5.2654e-03, -6.2303e-02,
        -3.9068e-02, -6.9222e-04,  1.1733e-02, -5.0158e-03,  2.0036e-03,
        -1.9411e-04, -5.4545e-02,  3.8598e-03, -1.0360e-01,  2.0469e-03,
        -9.0193e-02,  6.1988e-04,  2.2774e-02,  2.6144e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0162, -0.1507, -0.1541,  0.0046, -0.0868, -0.0238, -0.0303, -0.0864,
         0.0135, -0.0375, -0.0034, -0.0343, -0.0404, -0.0070, -0.0061, -0.0137,
         0.0177, -0.0453,  0.0055, -0.0736, -0.0736,  0.0430,  0.0328,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.4347e-02, -1.5147e-01,  5.8870e-03,  2.4411e-03, -2.7971e-02,
        -4.6299e-02,  1.6998e-03,  2.1794e-02, -2.9686e-03, -5.7263e-03,
         1.0788e-02, -6.5555e-03, -3.8136e-03, -8.6082e-03,  9.7975e-03,
        -2.9747e-02, -6.1914e-02, -3.0024e-02, -1.4141e-02,  1.0663e-02,
         2.1977e-03, -2.7516e-02, -1.9471e-02,  9.3660e-03, -4.6399e-02,
        -2.9714e-02,  2.3658e-03,  1.7999e-03, -2.0894e-02, -9.2229e-02,
        -7.0471e-02,  9.2788e-05, -5.0923e-02, -2.0266e-02, -3.4739e-02,
        -4.3088e-02, -5.8171e-04, -4.7285e-02, -1.3942e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0082, -0.0484,  0.0089, -0.0358, -0.1473, -0.0159,  0.0068, -0.0063,
         0.0114,  0.0143,  0.0051, -0.0047, -0.0078, -0.0054, -0.0208, -0.0849,
        -0.0188, -0.0045,  0.0055, -0.0090, -0.0474, -0.0219, -0.0143, -0.0279,
        -0.1065, -0.0824, -0.0311, -0.0088, -0.0091, -0.0396, -0.0126, -0.0109,
         0.0137, -0.0144, -0.0059, -0.0004, -0.0039, -0.0089,  0.0297,  0.0066,
        -0.0115, -0.0108,  0.0121,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.5868e-03, -2.5097e-01, -1.0223e-02, -4.2909e-03,  3.0186e-02,
         8.9504e-04, -8.9703e-03,  9.2650e-03, -5.8701e-02,  1.6385e-03,
         1.9534e-02,  1.1517e-02, -2.0425e-03,  1.4652e-03, -1.2994e-03,
         8.9137e-03, -1.5869e-02, -5.7517e-02, -2.3616e-03, -1.2667e-02,
         9.4786e-03,  2.2444e-02, -2.7266e-02,  5.1099e-03, -1.9112e-02,
        -1.1536e-02, -2.2301e-03,  3.1733e-03,  5.3183e-03,  2.1795e-03,
         3.9034e-03, -3.5238e-02, -7.0761e-02,  3.2494e-03, -2.1913e-02,
         7.6829e-03,  3.4808e-02, -2.4315e-02, -3.5448e-02, -6.2112e-03,
        -2.5413e-02, -4.2691e-02, -5.8863e-03, -7.8600e-04, -4.4818e-04,
         5.8732e-03, -7.7158e-04,  2.7961e-03,  7.2285e-04, -1.0131e-03,
         9.3866e-03, -2.6918e-03,  7.4723e-03,  6.0877e-03, -5.0341e-03,
        -1.4542e-02,  9.2774e-05], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0949, -0.4744, -0.0146, -0.0216, -0.0641,  0.0155, -0.0648, -0.1181,
        -0.0053,  0.0050, -0.0689,  0.0226,  0.0127,  0.0176,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0123, -0.2512, -0.0351, -0.0294,  0.0014, -0.0134, -0.0027, -0.0055,
        -0.0176, -0.0046, -0.0549, -0.0730, -0.0141, -0.0077, -0.0559, -0.0300,
        -0.0161, -0.0338,  0.0131, -0.0571, -0.0874, -0.0306, -0.0522, -0.0159,
         0.0109,  0.0047,  0.0017,  0.0099, -0.0304,  0.0275,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1539, -0.1529,  0.0532, -0.2072, -0.2167,  0.0934,  0.0470, -0.0757,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0350, -0.1635, -0.0357, -0.0212, -0.0595, -0.0943,  0.0038,  0.0151,
        -0.0062, -0.0159,  0.0062, -0.0148, -0.0484, -0.0895, -0.0131,  0.0012,
        -0.0105, -0.0063, -0.0101,  0.0007,  0.0195, -0.0006, -0.0070,  0.0063,
        -0.0675,  0.0037, -0.0340, -0.0057, -0.0642, -0.0820, -0.0054,  0.0013,
        -0.0518,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.2583e-02, -1.5552e-01, -1.0583e-01, -1.9395e-02, -5.8915e-03,
        -1.1264e-02, -5.1551e-03, -1.2486e-02,  1.4059e-02, -1.5718e-02,
         7.2001e-03, -4.0575e-03, -1.4197e-02,  2.8962e-03, -5.3620e-03,
         3.0203e-03,  9.3003e-04,  7.8974e-03, -1.5761e-02,  6.8086e-03,
         5.3866e-03,  4.1217e-03, -3.8808e-03, -8.9149e-04,  9.7799e-03,
        -1.1864e-02, -1.5218e-02, -5.5827e-02, -1.1244e-03, -6.2533e-03,
        -1.2799e-02, -1.2360e-02, -5.4080e-02, -8.7098e-02,  5.1928e-03,
         4.0246e-04, -2.3496e-02, -1.5409e-02, -6.4080e-03,  4.0240e-03,
         5.4154e-03, -4.5163e-03, -1.5633e-03, -3.6838e-02,  3.4054e-03,
         6.9139e-04, -2.9081e-03,  1.7002e-03, -6.2142e-04,  7.8547e-03,
         4.2904e-05, -1.8048e-02, -4.8533e-03, -6.2514e-03, -1.9268e-03,
         5.5676e-03,  1.0848e-03,  3.1906e-04,  1.0276e-04, -6.6746e-03,
        -7.4319e-04, -6.6000e-03, -4.5608e-02, -1.3426e-04,  1.8408e-04,
         2.9709e-03, -1.8343e-03,  7.1369e-03,  4.3363e-03, -1.4475e-02,
        -3.9373e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.2996e-02,  4.9673e-03,  1.9764e-03,  2.2439e-03, -3.0686e-02,
        -3.4565e-04, -6.7640e-02, -1.5252e-02,  6.1266e-03,  1.9678e-02,
         2.5661e-02,  8.9576e-03,  1.5310e-02,  3.6239e-03,  2.0077e-02,
         1.0380e-02, -6.9218e-03,  9.6121e-03, -1.5194e-02, -8.1422e-02,
         6.7231e-05,  1.3713e-03,  2.3535e-02, -7.0689e-02,  1.2998e-02,
        -3.9897e-02, -2.3420e-02,  2.1204e-03,  6.4760e-03,  1.1306e-03,
        -2.6154e-02, -2.9095e-02, -1.4371e-01, -5.6130e-02, -1.7418e-02,
        -1.0298e-02, -5.9080e-02, -4.1895e-02, -2.3379e-03,  7.4658e-03,
         2.6903e-02, -1.8740e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0501, -0.2405,  0.0010, -0.0023, -0.0011, -0.0113,  0.0113, -0.0046,
        -0.0078, -0.0503, -0.0316, -0.0323, -0.0513, -0.0074,  0.0081,  0.0024,
        -0.0095, -0.0503, -0.0799,  0.0018, -0.0028, -0.0121, -0.0043,  0.0011,
        -0.0187, -0.0912, -0.0082, -0.0126, -0.0675, -0.0061, -0.0022, -0.0077,
         0.0134,  0.0064, -0.0133, -0.0362, -0.0413,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.6772e-02,  5.7333e-01,  3.9699e-02,  3.2099e-02, -3.4826e-02,
        -3.4426e-02, -4.0577e-03, -2.0451e-02,  4.5874e-03, -1.3822e-02,
         2.9558e-02, -4.5619e-02,  6.4132e-02,  3.6112e-02,  5.1540e-04,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0735,  0.0413,  0.0029,  0.0094,  0.1433,  0.1316, -0.0171,  0.0609,
         0.1730,  0.0156,  0.0010,  0.0535,  0.0538,  0.0122,  0.0140, -0.0032,
        -0.0520,  0.0277, -0.0221, -0.0920,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.6057e-02,  7.2553e-04, -9.6292e-03, -1.4832e-02, -1.4031e-01,
        -1.1986e-03, -3.0948e-03, -1.5562e-02, -4.9805e-03,  2.4504e-03,
        -4.7823e-02, -5.8212e-02, -1.0416e-02,  1.4026e-03, -3.4811e-02,
        -3.1993e-02,  8.4018e-03, -1.2159e-02,  7.4515e-03, -3.2511e-02,
        -1.3774e-02, -1.6329e-03, -7.4838e-03, -7.8387e-03,  3.2431e-03,
        -3.7972e-02, -8.8232e-02,  8.2516e-03,  8.5107e-03, -5.3178e-02,
        -7.9296e-02, -2.8998e-02, -1.0075e-04, -1.4453e-02, -3.7801e-02,
        -5.4745e-02,  2.0591e-02, -2.6863e-02, -5.1547e-03,  1.1880e-02,
        -1.3004e-02,  2.2581e-03, -1.0717e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.7289e-02,  9.7563e-03,  1.9353e-02, -1.1937e-03, -3.0548e-02,
        -1.3539e-02, -2.8499e-02, -7.4835e-02, -2.6014e-02, -1.0123e-03,
         2.6937e-03,  1.5827e-03, -5.0634e-03, -6.5503e-02, -8.3316e-02,
        -9.1756e-03,  4.5690e-05,  2.0097e-04, -1.3109e-02, -1.7402e-02,
        -8.8354e-02, -1.3923e-02,  1.4410e-03,  1.9435e-02, -2.8303e-02,
         9.0147e-03, -1.6866e-02, -1.1744e-02, -6.9437e-02,  4.5082e-03,
        -1.4204e-03, -5.9191e-02, -2.0139e-02, -7.0802e-02, -1.0783e-01,
        -2.1511e-03,  5.7301e-03, -9.5738e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.3901e-02, -2.2881e-01, -9.3117e-02, -1.5903e-01, -4.9073e-02,
        -8.3568e-02,  4.3632e-02, -4.0225e-03, -2.0283e-02, -7.2423e-02,
        -9.1533e-04, -8.8550e-04, -5.4384e-02,  1.9911e-04,  1.3046e-02,
        -1.9840e-02, -1.5586e-02, -5.4474e-02, -1.0158e-03,  9.6371e-03,
         3.9750e-02, -1.2410e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0386, -0.0575,  0.0051, -0.0068,  0.0169, -0.0094,  0.0025,  0.0928,
         0.0674, -0.0690, -0.0112,  0.0029, -0.0131,  0.0021, -0.0123, -0.0388,
        -0.0823, -0.0158, -0.0289, -0.0476, -0.0073,  0.0021, -0.0629, -0.0327,
        -0.0154, -0.0005, -0.0139,  0.0044,  0.0031, -0.0287, -0.0190,  0.0084,
         0.0059, -0.0273, -0.0315, -0.0912,  0.0038, -0.0088,  0.0057, -0.0065,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1528e-04,  8.3821e-04, -1.2325e-02, -2.1657e-02,  7.1535e-03,
        -4.4992e-02, -1.0978e-02, -8.2574e-04, -5.1273e-02, -3.9149e-02,
        -1.0789e-02, -2.9476e-02, -3.3917e-03, -1.4574e-01, -1.0504e-01,
        -1.0201e-02,  6.7237e-03,  1.3293e-02, -2.4094e-02,  2.2314e-02,
        -5.6162e-03, -5.8355e-03,  6.5174e-04, -9.0985e-03, -6.3303e-03,
        -1.1406e-01,  2.2693e-03, -8.8821e-02, -9.7585e-02, -3.1954e-02,
         1.8121e-02, -5.9290e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0142e-02, -4.3062e-01, -1.7024e-02, -1.1760e-02, -1.4590e-02,
        -7.7118e-03, -1.7798e-02, -6.8718e-02, -9.6926e-02,  5.9024e-03,
        -1.3171e-03, -2.6737e-02,  2.7012e-03, -6.4872e-03,  8.5787e-03,
         4.6998e-03, -4.5376e-02, -2.6690e-06, -3.3627e-03,  1.5837e-02,
        -3.3428e-03, -8.2813e-03,  4.3661e-03, -5.3323e-02, -7.5572e-03,
         5.5021e-03, -4.2847e-02, -3.2046e-04, -2.7866e-03, -2.3149e-02,
        -1.3165e-03,  2.7994e-03,  4.7910e-03,  1.0329e-03,  4.2292e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0083, -0.0430, -0.1385, -0.0516, -0.0080, -0.0044, -0.1404, -0.1336,
        -0.0466, -0.1119,  0.0106, -0.0640,  0.0081, -0.0491, -0.0140, -0.0204,
        -0.0453, -0.0196,  0.0089, -0.0280,  0.0458,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0968, -0.3135,  0.0206, -0.0488, -0.0897, -0.0433, -0.0066, -0.0661,
        -0.0528,  0.0254, -0.0124,  0.0155, -0.0029,  0.0032, -0.0497,  0.0068,
        -0.0726, -0.0305,  0.0430,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0193, -0.0410, -0.0652, -0.0137, -0.0721, -0.0013,  0.0109,  0.0109,
        -0.0029, -0.1512, -0.1901, -0.0034, -0.0409, -0.2131,  0.0063, -0.0485,
        -0.0050, -0.0482, -0.0142, -0.0418,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0453, -0.2880,  0.0054, -0.0862,  0.0083, -0.0174, -0.0603, -0.0526,
        -0.0105, -0.0062, -0.0248, -0.0214, -0.0075, -0.0897, -0.0156, -0.0322,
        -0.0602, -0.0189, -0.0223, -0.0322, -0.0003, -0.0530, -0.0416,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0439, -0.1845, -0.0692, -0.0403,  0.0089, -0.0413, -0.0248, -0.0160,
        -0.0797, -0.1092, -0.0219, -0.0442, -0.0134, -0.0830,  0.0320, -0.0034,
         0.0196, -0.0200, -0.0084,  0.0158, -0.0702, -0.0149, -0.0112, -0.0085,
         0.0156,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0446,  0.3041,  0.0289, -0.0142,  0.0099,  0.0115, -0.0032,  0.0015,
        -0.0201, -0.0019,  0.0531,  0.0040,  0.0077, -0.0086, -0.0119,  0.0028,
         0.0104,  0.0060, -0.0090,  0.0089,  0.0869,  0.0582,  0.0433,  0.0345,
         0.0238,  0.0300,  0.0086,  0.0092,  0.0068,  0.0069,  0.0182, -0.0120,
         0.0299,  0.0353, -0.0340,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0379,  0.0160, -0.0065, -0.0869, -0.0175,  0.0096, -0.1099, -0.0168,
        -0.1467,  0.0006,  0.0159, -0.0253,  0.0356, -0.0384, -0.0563, -0.1500,
        -0.0254, -0.0860,  0.0305,  0.0628,  0.0083,  0.0169,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0370,  0.0408, -0.0236, -0.0047, -0.0187,  0.0433, -0.0117,  0.0189,
        -0.0072,  0.3252,  0.1022,  0.0314,  0.0186,  0.0297,  0.0163,  0.1136,
         0.0220, -0.0339, -0.0196,  0.0312, -0.0504,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1572, -0.2560, -0.0209,  0.0088, -0.0168,  0.0053, -0.0785, -0.1431,
        -0.0008, -0.0040,  0.0178,  0.0120, -0.0289, -0.0711, -0.0054, -0.0461,
         0.0218,  0.1019, -0.0035,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4616e-02,  8.6095e-03, -2.2053e-02,  1.6255e-02,  1.4319e-02,
        -5.4802e-02,  4.5826e-03, -1.1770e-02,  6.4218e-03, -6.8549e-03,
         3.6739e-03, -2.1392e-02, -1.4443e-02, -2.0944e-02,  4.4723e-02,
         1.5695e-02, -4.5893e-02, -4.4307e-02,  1.9140e-02, -5.5364e-02,
        -1.5039e-02, -1.5942e-02, -4.0810e-02,  7.8817e-04, -4.2803e-02,
        -1.4571e-02, -1.0684e-02, -4.1737e-02,  1.7785e-02, -1.3714e-01,
        -5.3506e-03, -2.6133e-02, -2.7758e-02, -3.9583e-02, -6.8777e-03,
        -7.1639e-03, -1.4094e-02,  2.3814e-03,  8.7675e-03, -3.2063e-02,
         1.1815e-03,  1.1529e-02, -7.2273e-05, -7.4906e-03, -1.0373e-02,
        -9.4654e-03,  6.5527e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0896, -0.2256, -0.0853, -0.0164,  0.0020,  0.0118, -0.0132, -0.0159,
        -0.0228,  0.0144,  0.0074,  0.0101, -0.0040,  0.0232,  0.0067, -0.0025,
        -0.0909,  0.0033,  0.0016, -0.0082,  0.0039,  0.0037, -0.0352, -0.0987,
         0.0029,  0.0077, -0.0276,  0.0019,  0.0112,  0.0061,  0.0002,  0.0066,
         0.0302, -0.1091,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0216,  0.4230,  0.0106, -0.0184, -0.0044,  0.0360, -0.0127, -0.0077,
         0.0044,  0.0235,  0.0013, -0.0043,  0.0458,  0.0663, -0.0015,  0.0142,
        -0.0062, -0.0108,  0.0100,  0.0844, -0.0204, -0.0022,  0.0161, -0.0057,
        -0.0164, -0.0195, -0.0069,  0.0094,  0.0025,  0.0402,  0.0070, -0.0129,
         0.0098, -0.0237,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0775,  0.4162, -0.0727,  0.0258,  0.0331,  0.1002, -0.0202,  0.0139,
         0.0180,  0.0462,  0.0275,  0.0148,  0.0250,  0.0571,  0.0138, -0.0058,
        -0.0091,  0.0231,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0150, -0.0156, -0.0050, -0.0118, -0.0116, -0.0142, -0.0181,  0.0243,
        -0.1540, -0.1848, -0.0052, -0.0660, -0.0574, -0.0083, -0.0007, -0.0120,
         0.0038,  0.0060, -0.0151, -0.2329,  0.0261, -0.0292, -0.0011, -0.0050,
        -0.0246,  0.0081, -0.0440,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0417, -0.0155,  0.0052,  0.0084, -0.0145, -0.0148, -0.0079, -0.0118,
        -0.0560,  0.0023,  0.0025, -0.0108, -0.2454, -0.0077, -0.0301,  0.0129,
        -0.0162, -0.0260, -0.0595, -0.2366, -0.0062,  0.0096, -0.0905, -0.0038,
         0.0236, -0.0403,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0209, -0.3573,  0.0216, -0.0628, -0.0969,  0.0126, -0.0186,  0.0196,
        -0.0645, -0.0522, -0.0477, -0.0228,  0.0279, -0.0015, -0.0139, -0.0477,
        -0.0128, -0.0236,  0.0032, -0.0207,  0.0511,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0607, -0.3392, -0.1589, -0.1620, -0.0285, -0.0273, -0.0678, -0.0164,
         0.0014,  0.0434, -0.0136, -0.0569, -0.0018,  0.0188,  0.0034,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0995,  0.0126, -0.0122,  0.0161,  0.0032, -0.0286, -0.0284,  0.0345,
         0.0220,  0.0458,  0.0218,  0.0040,  0.4988,  0.0472,  0.0975,  0.0062,
        -0.0194,  0.0021,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0412, -0.2318, -0.1094, -0.1472, -0.0146, -0.0096, -0.0333, -0.0406,
        -0.0170, -0.0543, -0.0757, -0.0456,  0.0117, -0.0555, -0.0089,  0.0060,
         0.0157, -0.0182,  0.0636,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.4267e-05, -7.6943e-02, -6.8339e-02, -1.8281e-02, -3.2345e-05,
        -2.4062e-02,  2.7935e-02,  8.4626e-03, -1.3618e-02,  1.3066e-03,
         2.5300e-03,  2.1718e-03, -2.6180e-02,  6.3094e-03,  5.6587e-03,
         8.3313e-04, -4.5088e-02,  9.9754e-03, -5.1769e-03,  3.8512e-02,
        -9.7192e-02,  3.1540e-03, -5.9834e-02, -2.8530e-03, -7.1896e-02,
         3.6137e-03, -7.8332e-02,  5.0246e-03,  1.2428e-02, -2.2093e-02,
        -6.5966e-03, -9.5364e-03, -5.1580e-02, -1.9628e-03, -3.0629e-02,
        -5.9236e-03, -4.5363e-02,  2.0815e-02, -2.0706e-02, -9.7782e-03,
        -5.2249e-03, -5.2136e-03,  2.1346e-02, -2.7444e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0196, -0.2558, -0.0409, -0.1045, -0.0590, -0.0067, -0.0189, -0.0088,
        -0.0254, -0.0152, -0.0321,  0.0095,  0.0075,  0.0055,  0.0248, -0.0007,
         0.0183, -0.0904,  0.0057,  0.0047, -0.0321,  0.0008,  0.0089,  0.0023,
         0.0144,  0.0065, -0.0132, -0.0720,  0.0150,  0.0093,  0.0095, -0.0162,
         0.0089, -0.0040,  0.0328,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.4963e-02, -4.1328e-01,  2.5347e-03, -1.2750e-01, -3.4870e-03,
        -2.3912e-04, -8.1534e-02, -4.8303e-02, -2.6632e-02,  7.6324e-03,
         7.9589e-03, -2.6790e-02, -3.8119e-02, -3.7942e-03,  1.7188e-02,
        -3.8530e-02, -1.4715e-02,  5.5332e-03,  2.6584e-02, -6.0783e-03,
        -2.2568e-03, -2.4510e-02,  2.2354e-02, -2.9488e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0236, -0.1640, -0.0145, -0.0361, -0.0164,  0.0026,  0.0071, -0.0226,
        -0.0975, -0.0122, -0.0327,  0.0035,  0.0150, -0.0388, -0.0793, -0.0117,
        -0.0008, -0.0028, -0.0451, -0.0073, -0.0030, -0.0423,  0.0030, -0.0220,
        -0.0122, -0.0743, -0.0321, -0.0306, -0.0171, -0.0243, -0.0639,  0.0072,
         0.0247, -0.0095,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0430,  0.0209, -0.0126, -0.0207,  0.0056,  0.0154, -0.1438, -0.0132,
         0.0003, -0.0108,  0.0140, -0.0146, -0.0732, -0.0050, -0.0884, -0.2001,
        -0.0255, -0.0006, -0.0108,  0.0168,  0.0081, -0.0035, -0.0028, -0.0779,
        -0.0315, -0.0180, -0.0548, -0.0098,  0.0070,  0.0512,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.1118,  0.1273, -0.0436, -0.3335,  0.0146, -0.0054, -0.0537, -0.0298,
        -0.0505, -0.1857,  0.0118, -0.0210,  0.0111,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0583, -0.0171, -0.0524, -0.0222,  0.0080,  0.0121,  0.0628,  0.1426,
         0.1504, -0.0067, -0.0248,  0.1197, -0.0703,  0.0764,  0.0456, -0.0343,
        -0.0180,  0.0784,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1279, -0.3319, -0.3067, -0.0293, -0.1051, -0.0599,  0.0276, -0.0116,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0005, -0.2545, -0.0099, -0.0325, -0.0364, -0.0084, -0.0053, -0.0018,
         0.0189, -0.0057,  0.0055,  0.0155, -0.0057, -0.0013,  0.0008, -0.0055,
         0.0041,  0.0015,  0.0019, -0.0038,  0.0154,  0.0051, -0.0328, -0.0212,
         0.0123, -0.0156, -0.0068, -0.0125, -0.1222, -0.0364,  0.0138, -0.0457,
        -0.0099, -0.0918,  0.0011, -0.0440, -0.0210, -0.0032,  0.0054,  0.0040,
        -0.0154, -0.0031, -0.0032,  0.0049,  0.0161,  0.0049,  0.0126],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1241,  0.1756, -0.0203,  0.4564,  0.0410, -0.0783,  0.0018,  0.0044,
         0.0298,  0.0237, -0.0446,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0531,  0.4306,  0.0055,  0.0910,  0.0166,  0.0032,  0.0275,  0.0961,
         0.0291,  0.0628,  0.0274, -0.0099,  0.0837, -0.0052, -0.0358, -0.0005,
         0.0178,  0.0041,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0906, -0.0805, -0.2357,  0.0292,  0.0347,  0.0355, -0.1626, -0.0431,
        -0.0948,  0.0063,  0.0239,  0.0577,  0.0255,  0.0208, -0.0591,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0432, -0.3447,  0.0213, -0.0363,  0.0500, -0.0634, -0.1439,  0.0147,
         0.0015, -0.0949, -0.1469,  0.0017,  0.0272, -0.0039, -0.0063,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.9441e-01,  3.9953e-03, -8.4681e-03, -2.6549e-02, -7.4531e-03,
         7.8271e-03, -1.2029e-02,  2.4448e-02,  3.4536e-02,  5.4321e-03,
        -3.9386e-05,  3.9289e-02,  1.2526e-01,  5.0258e-03, -1.1589e-02,
         1.0325e-02,  3.6731e-02, -2.4388e-02, -2.0740e-02, -6.2714e-03,
         7.3635e-02,  1.3885e-01, -2.6199e-02, -7.7762e-02,  4.6075e-02,
         3.2673e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0832,  0.0194,  0.0183,  0.0674, -0.0172,  0.0080, -0.0152,  0.1471,
         0.3759,  0.0138,  0.0231,  0.0427, -0.0505, -0.0462,  0.0720,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0331,  0.3251,  0.0146,  0.0679,  0.0791,  0.0248,  0.0585, -0.0071,
         0.0006,  0.1122, -0.0094, -0.0207,  0.0311, -0.0008, -0.0361, -0.0015,
         0.0169, -0.0043,  0.0475,  0.0015,  0.0408, -0.0175, -0.0164, -0.0107,
         0.0020, -0.0199,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0097, -0.0220,  0.0089,  0.0068, -0.0227, -0.0495,  0.0192,  0.0105,
        -0.0435, -0.0872, -0.2008, -0.0187, -0.0198,  0.0381, -0.0926, -0.0654,
        -0.0085, -0.0293, -0.0006, -0.0670, -0.0141, -0.0454, -0.0222, -0.0600,
         0.0179, -0.0195,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0424, -0.2701,  0.0483, -0.0123, -0.0076,  0.0049, -0.0496, -0.0338,
         0.0217, -0.0277, -0.0245, -0.0232, -0.0355,  0.0096, -0.0221, -0.0159,
         0.0041, -0.0062, -0.0058, -0.0056,  0.0104, -0.0118,  0.0121, -0.0560,
        -0.0711, -0.0119, -0.0418, -0.0117,  0.0222,  0.0021,  0.0074, -0.0224,
         0.0011,  0.0033,  0.0060, -0.0377,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0407, -0.0880, -0.0301, -0.0164, -0.0436, -0.0205, -0.0465, -0.0836,
         0.0084,  0.0132, -0.0143,  0.0040, -0.0012,  0.0133, -0.0012,  0.0010,
         0.0097, -0.0037,  0.0046,  0.0151, -0.0071, -0.0708, -0.0117, -0.0077,
        -0.0403, -0.0857, -0.0017, -0.0716, -0.0511, -0.0243, -0.0107, -0.0599,
        -0.0668,  0.0021,  0.0061,  0.0238,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0019, -0.0845, -0.0384, -0.0153,  0.0177, -0.2190,  0.0245, -0.0153,
        -0.0130, -0.0062, -0.0511, -0.0954, -0.2679, -0.0024,  0.0413, -0.0055,
        -0.0471,  0.0258, -0.0155, -0.0124,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0762, -0.1403, -0.0264, -0.0146, -0.0217,  0.0028, -0.0124, -0.0456,
        -0.0098, -0.0304, -0.0112, -0.0044, -0.0052, -0.0098, -0.0689, -0.0565,
        -0.0234, -0.0204, -0.0075, -0.0119, -0.0169, -0.0060,  0.0035, -0.0041,
        -0.0141,  0.0046,  0.0016, -0.0090,  0.0024, -0.0344, -0.1489, -0.0019,
        -0.0414, -0.0633,  0.0084,  0.0160, -0.0113, -0.0129,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0203, -0.1620, -0.0895, -0.0873, -0.0112,  0.0091, -0.0144,  0.0377,
         0.0165, -0.0135, -0.0268, -0.0666, -0.0004,  0.0084, -0.0038, -0.0457,
         0.0264, -0.0464,  0.0005, -0.0020, -0.0005,  0.0102,  0.0280,  0.0216,
         0.0023, -0.0013, -0.0103, -0.0513, -0.0769, -0.0096, -0.0166, -0.0113,
         0.0073, -0.0091, -0.0305, -0.0030,  0.0032,  0.0129, -0.0041, -0.0004,
         0.0013], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0754, -0.4555, -0.0024,  0.0268,  0.0130, -0.0493, -0.0130, -0.0140,
        -0.0013,  0.0062, -0.0092, -0.0476, -0.0415, -0.0203,  0.0256, -0.0603,
        -0.0071, -0.0498, -0.0226,  0.0224, -0.0007, -0.0358,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0304,  0.0082, -0.0014,  0.0018, -0.0055, -0.0056, -0.0052,  0.0289,
        -0.0089,  0.0103, -0.0066, -0.0276,  0.0150, -0.0660, -0.0077, -0.0946,
         0.0031, -0.0535, -0.0183, -0.0884, -0.1175,  0.0019,  0.0318, -0.0493,
         0.0050,  0.0009,  0.0034, -0.0256, -0.0976,  0.0094, -0.0653, -0.0558,
        -0.0261,  0.0082, -0.0152,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0332, -0.1012, -0.2860, -0.0259, -0.0823, -0.0270,  0.0194,  0.0251,
         0.0384, -0.0133, -0.0288, -0.0137, -0.0052,  0.0159, -0.0096, -0.0027,
         0.0388, -0.0035, -0.0529,  0.0095, -0.0349,  0.0078, -0.0026, -0.0275,
        -0.0063,  0.0049, -0.0353,  0.0296, -0.0192,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1150e-02, -3.8981e-01, -6.0634e-02, -6.4470e-02,  4.1609e-03,
        -5.7773e-02, -1.7133e-02, -2.5549e-02, -4.8007e-03,  2.6516e-03,
        -5.5747e-02, -5.0095e-02, -1.8010e-02,  1.4891e-03, -1.4812e-04,
        -5.2843e-02, -2.6112e-02, -4.0319e-03, -2.2726e-02,  1.2257e-02,
        -1.1751e-02, -4.3738e-04, -1.5768e-02,  1.1921e-02,  7.8530e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0934,  0.0204, -0.0084,  0.0237, -0.2191,  0.0271,  0.0090, -0.1282,
        -0.1283, -0.0161,  0.0111,  0.0111,  0.0038,  0.0184, -0.0896, -0.0311,
         0.0345, -0.0317, -0.0200, -0.0172, -0.0096,  0.0242,  0.0239,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0050, -0.0007,  0.0080,  0.0035,  0.0076,  0.0013, -0.0048,  0.0016,
         0.0095,  0.0233, -0.0018, -0.0121, -0.0399, -0.1334, -0.0181,  0.0059,
        -0.0026, -0.0477, -0.0026, -0.0007,  0.0030, -0.0130,  0.0190,  0.0073,
        -0.0075,  0.0210, -0.1805,  0.0006, -0.1460, -0.0466, -0.0656, -0.0132,
        -0.0097, -0.0150, -0.0335,  0.0225, -0.0142,  0.0197, -0.0323,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0247, -0.1167, -0.0283, -0.1863, -0.1842, -0.0494, -0.0113, -0.1590,
        -0.1070, -0.0529, -0.0451,  0.0351,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 1.2261e-02, -9.2154e-02, -3.2294e-01,  1.6989e-02, -2.3351e-03,
        -6.0155e-02, -2.3179e-02,  7.2801e-03,  8.3221e-03, -1.7585e-02,
         3.8702e-03,  3.1101e-03, -5.5837e-03, -6.5560e-03, -3.6372e-02,
         2.3389e-02, -7.6376e-03, -7.0369e-02,  6.3515e-04,  1.0005e-02,
        -2.0932e-02, -1.9768e-02, -1.0408e-02,  7.7110e-03,  6.8331e-03,
         5.0009e-03,  6.0538e-05, -1.5088e-02, -6.4225e-03,  7.5219e-03,
         9.7063e-03,  3.8512e-02, -2.2499e-02, -2.3130e-03,  2.8314e-03,
         1.2673e-02, -3.0986e-02, -1.6111e-03, -6.1194e-03, -2.7752e-04,
         7.7724e-03,  2.8400e-03,  1.7969e-02,  4.6722e-04,  1.2955e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0086,  0.4629,  0.0958,  0.1213, -0.0049,  0.0287, -0.0495,  0.0399,
         0.0997, -0.0397,  0.0431,  0.0060,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0305, -0.6801, -0.0697, -0.1191, -0.0298, -0.0488,  0.0117,  0.0103,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0409, -0.2611, -0.0379,  0.0331, -0.0686,  0.0121, -0.0294, -0.0189,
        -0.0143, -0.0508, -0.0042, -0.0093, -0.0016, -0.0027,  0.0021, -0.0024,
        -0.0045, -0.0030,  0.0012, -0.0033,  0.0047,  0.0052, -0.0052, -0.0044,
        -0.0048,  0.0046,  0.0034, -0.0021, -0.0279, -0.0117, -0.0008, -0.0033,
         0.0010, -0.0033, -0.0056,  0.0058, -0.0042, -0.0040,  0.0006,  0.0004,
        -0.0003, -0.0030, -0.0059, -0.0045, -0.0005,  0.0311, -0.0657,  0.0070,
        -0.0394, -0.0399, -0.0122, -0.0121, -0.0166,  0.0151, -0.0144, -0.0011,
        -0.0128,  0.0022, -0.0017, -0.0024,  0.0076], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0079,  0.0005,  0.0174, -0.0221, -0.0982,  0.0253, -0.0482,  0.0099,
         0.0010,  0.0053, -0.0083, -0.0769,  0.0097, -0.0694, -0.1341,  0.0149,
         0.0141, -0.0172, -0.0317, -0.0078, -0.0707, -0.0629, -0.0121, -0.0086,
        -0.0525, -0.0187,  0.0003, -0.0349, -0.0471, -0.0359, -0.0066, -0.0233,
        -0.0064,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0711,  0.0754,  0.2291,  0.0797,  0.0415, -0.0041,  0.0621,  0.0082,
        -0.0041,  0.0340, -0.0028, -0.0181,  0.0055,  0.0183, -0.0006, -0.0357,
         0.0086,  0.0167,  0.1935,  0.0109,  0.0107, -0.0091, -0.0411, -0.0191,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0291, -0.0825, -0.0568, -0.0331, -0.0204, -0.0954,  0.0178, -0.0674,
        -0.0196, -0.0880, -0.0887, -0.0698, -0.1396, -0.0377,  0.0189, -0.0026,
         0.0175,  0.0007,  0.0142, -0.0098, -0.0159, -0.0218, -0.0194, -0.0334,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0718,  0.0089,  0.0584,  0.0874,  0.0012,  0.0100,  0.0134,  0.0558,
         0.0160,  0.0016,  0.0097,  0.0108, -0.0212, -0.0036,  0.0780,  0.1099,
         0.0831,  0.0007, -0.0074, -0.0055,  0.0688,  0.0109,  0.0273,  0.0663,
        -0.0056, -0.0135,  0.0617,  0.0083,  0.0090,  0.0114,  0.0171,  0.0192,
        -0.0059, -0.0204,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0112, -0.0003, -0.0289, -0.0044, -0.1496,  0.0514, -0.0015, -0.0231,
        -0.0387, -0.0058,  0.0032,  0.0069, -0.0245, -0.1856, -0.0050, -0.0382,
        -0.1441,  0.0069, -0.0793, -0.0106, -0.0651, -0.0750,  0.0287, -0.0054,
         0.0067,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0544, -0.0172, -0.0038,  0.0074,  0.0366, -0.0290,  0.1575,  0.0642,
         0.0841,  0.3302, -0.0290,  0.0491,  0.0167,  0.0024, -0.0510,  0.0673,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0258, -0.2935, -0.0180, -0.0893, -0.0402, -0.0581, -0.1455, -0.0244,
         0.0448, -0.0079,  0.0017, -0.0116, -0.0219,  0.0264,  0.0435, -0.0028,
        -0.0183, -0.0414, -0.0129, -0.0105,  0.0104, -0.0474,  0.0035,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0651, -0.0017,  0.0026,  0.1913,  0.4472,  0.0197, -0.0898,  0.0099,
         0.0287, -0.0290,  0.0261, -0.0148, -0.0731, -0.0009,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0607, -0.0983, -0.0753, -0.1304, -0.0189,  0.0723, -0.0469,  0.0237,
        -0.0214, -0.0479, -0.0917, -0.0616, -0.0139, -0.0166, -0.0066,  0.0126,
         0.0128, -0.0709, -0.0133, -0.0355, -0.0271, -0.0103,  0.0170,  0.0145,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.6630e-03,  4.8762e-03, -2.0996e-01, -1.2525e-04, -1.6245e-01,
        -1.2298e-02, -5.0092e-01, -3.4868e-03,  5.5206e-02,  2.1743e-02,
         2.1267e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0663,  0.0178,  0.0202,  0.0161,  0.0522, -0.0928, -0.2624, -0.0039,
        -0.1570, -0.0193, -0.0486,  0.0174, -0.0402,  0.0009, -0.1001, -0.0096,
        -0.0192,  0.0165,  0.0017, -0.0250,  0.0128,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0090, -0.0394,  0.0154,  0.0101,  0.0898,  0.0267, -0.0514, -0.1902,
        -0.1365, -0.2451,  0.0239,  0.0682,  0.0124, -0.0818,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0498,  0.0202,  0.0171, -0.0128,  0.0177,  0.0326, -0.0147, -0.0032,
        -0.0141, -0.0039, -0.0115,  0.0535,  0.4302,  0.0223,  0.0018,  0.0542,
         0.0216, -0.0082,  0.0373, -0.0066,  0.1668,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0318,  0.3056, -0.0179,  0.0896, -0.0592,  0.0837,  0.0157,  0.1984,
         0.0297,  0.0437,  0.0194, -0.0468, -0.0584,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0278, -0.0903, -0.0042, -0.0020, -0.0054, -0.0129, -0.0286, -0.0895,
        -0.0401, -0.0102, -0.0017,  0.0077,  0.0079,  0.0094,  0.0041, -0.0114,
         0.0013, -0.0013, -0.0008,  0.0130, -0.0073,  0.0188,  0.0124, -0.0194,
        -0.1671, -0.0191, -0.0993, -0.0632, -0.0477, -0.0140, -0.0087,  0.0072,
        -0.0022,  0.0015,  0.0060,  0.0112, -0.0266,  0.0003,  0.0039, -0.0061,
        -0.0269, -0.0019,  0.0045,  0.0070,  0.0480], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0848, -0.0998, -0.0931, -0.0282, -0.0043,  0.0194,  0.0062, -0.0149,
         0.0043,  0.0066, -0.0296,  0.0131, -0.0043,  0.0043, -0.0180, -0.0177,
        -0.1417, -0.0030, -0.0018,  0.0099, -0.0099, -0.0602, -0.0118, -0.0850,
        -0.0012,  0.0111, -0.0242, -0.0244, -0.1090, -0.0108, -0.0154, -0.0186,
         0.0042, -0.0093,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0364, -0.0095,  0.0239,  0.0598,  0.0181,  0.0063,  0.0027,  0.0096,
        -0.0411,  0.0031,  0.0069,  0.0112, -0.0151,  0.0040, -0.0035, -0.0121,
        -0.0116, -0.0191,  0.0186,  0.2644,  0.3257,  0.0097, -0.0100,  0.0197,
        -0.0244,  0.0337,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0014, -0.2897, -0.0338, -0.0703,  0.0111, -0.0207,  0.0311, -0.0630,
        -0.0328,  0.0006, -0.0015,  0.0063,  0.0124,  0.0090, -0.0315, -0.1114,
        -0.1304, -0.0153, -0.0391, -0.0088, -0.0534,  0.0032,  0.0230,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0236, -0.1509, -0.0356, -0.0057,  0.0049,  0.0028, -0.0147, -0.0219,
        -0.0567, -0.0188,  0.0098,  0.0068,  0.0097,  0.0006, -0.1261,  0.0322,
        -0.0101, -0.0129, -0.0392, -0.0259, -0.0031, -0.0017,  0.0010,  0.0123,
         0.0051,  0.0036, -0.0512, -0.0958,  0.0127, -0.0084, -0.0222,  0.0045,
        -0.0014,  0.0084, -0.0060, -0.0047, -0.0391,  0.0057, -0.0251,  0.0791,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0123, -0.0083,  0.0101, -0.0123, -0.0615,  0.0041, -0.0236, -0.0747,
         0.0068, -0.0218, -0.0021, -0.0162, -0.0707, -0.0199, -0.0515, -0.1076,
        -0.0427, -0.0126,  0.0013, -0.0511, -0.0198, -0.0714, -0.0322, -0.0037,
        -0.0361, -0.0013, -0.0169, -0.0474, -0.0021, -0.0273, -0.0504, -0.0089,
        -0.0291,  0.0422,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0357, -0.0834, -0.0295, -0.0141, -0.1080,  0.0027,  0.0095, -0.0046,
        -0.0023,  0.0100, -0.0031, -0.0040, -0.0164, -0.0145, -0.1082,  0.0032,
        -0.0037, -0.0051, -0.0446,  0.0042, -0.0106,  0.0064, -0.0136,  0.0173,
        -0.0396, -0.0779,  0.0242,  0.0042, -0.0075, -0.0033, -0.0040,  0.0011,
         0.0035, -0.0108, -0.0347,  0.0009, -0.0186, -0.0362, -0.0117, -0.0168,
        -0.0209,  0.0015, -0.0306, -0.0096, -0.0004, -0.0037, -0.0034,  0.0044,
        -0.0079, -0.0018, -0.0076, -0.0585], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0059, -0.0033,  0.0027,  0.0139,  0.1102,  0.0108, -0.0340, -0.0439,
        -0.0124,  0.0158,  0.0014, -0.0031,  0.0143,  0.1685,  0.0237,  0.0453,
         0.1214,  0.0696,  0.0505, -0.0276,  0.0242,  0.0157,  0.0083,  0.1312,
         0.0040, -0.0121, -0.0012,  0.0067, -0.0038,  0.0083, -0.0065,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0448, -0.1094,  0.0016, -0.0072,  0.0098,  0.0133, -0.0171, -0.0024,
        -0.0149,  0.0304,  0.0041, -0.0483, -0.0022, -0.0292, -0.1097, -0.0852,
        -0.0264, -0.0863, -0.0840, -0.1146, -0.0308, -0.0064, -0.0385,  0.0056,
         0.0777,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0866, -0.0092, -0.0278,  0.0409,  0.0084, -0.0805, -0.5797,  0.0017,
        -0.0958, -0.0207,  0.0339, -0.0146,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0465, -0.1716, -0.0086, -0.0755, -0.0928, -0.0103,  0.0182,  0.0023,
        -0.0065, -0.0199,  0.0110, -0.0177, -0.0516, -0.1101, -0.0135, -0.0664,
        -0.0217, -0.0112, -0.0127,  0.0099, -0.0410, -0.0084, -0.0089,  0.0021,
        -0.0214, -0.0950, -0.0146,  0.0088, -0.0215,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0132,  0.5102,  0.0472, -0.0323,  0.0007,  0.0751,  0.1260, -0.0993,
        -0.0366, -0.0133,  0.0124,  0.0336,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-9.5944e-03, -4.4351e-01, -1.7027e-02, -1.4175e-01, -2.3624e-04,
        -7.0277e-02, -1.8006e-01,  3.2716e-03, -6.4447e-02, -1.1772e-03,
        -1.9059e-02,  4.9594e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0271,  0.0195, -0.1038,  0.0128, -0.1126, -0.0208, -0.0380, -0.0546,
        -0.1775, -0.0406,  0.0127, -0.1356, -0.0174, -0.0466, -0.0168, -0.0998,
        -0.0139, -0.0139,  0.0298, -0.0062,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.0683e-02,  2.2714e-01,  1.0805e-02,  4.7926e-02,  1.2292e-01,
         9.5109e-03,  2.8112e-02, -1.5177e-02,  2.5211e-02,  1.1034e-02,
         1.1599e-02,  1.7301e-02,  1.3628e-02,  1.1197e-02, -5.2692e-03,
         1.8722e-03,  1.3463e-02, -1.0210e-02,  3.2108e-03,  2.9660e-02,
         4.1175e-02, -3.0122e-02, -6.0767e-03,  1.5645e-02,  8.2023e-03,
        -7.9492e-03,  3.9941e-03,  4.0926e-02,  2.6415e-02, -2.8003e-03,
         2.3416e-02,  2.8881e-05,  2.7626e-02,  7.0356e-02,  6.7712e-03,
         2.3174e-02, -3.9385e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1198,  0.6618,  0.0236,  0.0226,  0.0197, -0.0134, -0.0618, -0.0321,
        -0.0452,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0012, -0.0133, -0.0240, -0.1339, -0.0180, -0.0030, -0.0530, -0.0974,
         0.0252, -0.0133, -0.0256, -0.0057, -0.0928, -0.1702, -0.0032, -0.0245,
         0.0088,  0.0016, -0.0813, -0.0508,  0.0453,  0.0129, -0.0052,  0.0024,
        -0.0103,  0.0401, -0.0367,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0556,  0.2408, -0.0102,  0.0471, -0.0028,  0.0039,  0.0025,  0.0015,
         0.0078, -0.0144, -0.0018,  0.0688,  0.0267, -0.0129,  0.0016,  0.0073,
         0.0740,  0.0112, -0.0157, -0.0079, -0.0079,  0.0727,  0.0658, -0.0205,
         0.0077, -0.0166,  0.0071,  0.0352,  0.0842, -0.0024, -0.0024,  0.0234,
         0.0317, -0.0081,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0143, -0.0023, -0.0040,  0.0022, -0.0448, -0.0094,  0.0056,  0.0088,
         0.0150,  0.0461, -0.0212, -0.0987, -0.0160, -0.0142,  0.0065, -0.0371,
        -0.0909,  0.0065, -0.0281, -0.0122,  0.0062, -0.0127, -0.0253, -0.0064,
         0.0259,  0.0118, -0.0018, -0.0045, -0.0027, -0.0069,  0.0215, -0.0728,
        -0.1043,  0.0061, -0.0207, -0.0474, -0.0536,  0.0019,  0.0062,  0.0042,
        -0.0242, -0.0184, -0.0101,  0.0123, -0.0084,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0509, -0.0204, -0.0415, -0.0239, -0.0341, -0.1419, -0.2655, -0.0050,
         0.0175, -0.1543,  0.0010, -0.0363, -0.1290, -0.0157, -0.0315,  0.0314,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0891, -0.0236,  0.0258,  0.0046, -0.0139, -0.0101, -0.0133,  0.0252,
         0.2044,  0.1480,  0.0740,  0.1447,  0.0163,  0.0418,  0.0302, -0.0063,
        -0.0045,  0.0050, -0.0220,  0.0071, -0.0029, -0.0242, -0.0021, -0.0064,
         0.0219, -0.0230, -0.0094,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0320, -0.0315, -0.0477, -0.0008, -0.0406, -0.0007, -0.0158, -0.0101,
        -0.0227,  0.0015, -0.0058, -0.0204, -0.0608, -0.0061, -0.0335, -0.0284,
        -0.0280,  0.0078,  0.0070, -0.0219,  0.0103, -0.0133, -0.0180, -0.0600,
        -0.0147, -0.0091, -0.0245, -0.0519,  0.0084, -0.0479,  0.0104, -0.0446,
        -0.0412, -0.0134,  0.0053, -0.0342, -0.0210, -0.0277,  0.0083,  0.0090,
        -0.0192, -0.0423, -0.0029, -0.0017, -0.0287, -0.0091], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0291, -0.0092, -0.0294, -0.0572,  0.0541,  0.0311,  0.0146,  0.2008,
         0.0073, -0.0283,  0.0491, -0.0063,  0.0030, -0.0058,  0.0111,  0.0112,
         0.0141,  0.0233,  0.0922,  0.1129,  0.0466,  0.0132,  0.0082,  0.0086,
         0.0064, -0.0063,  0.0841, -0.0081,  0.0147, -0.0138,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1633,  0.0361,  0.0275, -0.0275, -0.0666,  0.0872,  0.0617,  0.3788,
         0.0399, -0.0659,  0.0456,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0563, -0.3975,  0.0558, -0.0263, -0.0548, -0.0233, -0.0339, -0.1200,
         0.0142, -0.0142, -0.0078, -0.0112, -0.0414, -0.0219,  0.0086, -0.0667,
        -0.0085, -0.0232, -0.0084,  0.0061,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0685, -0.3629, -0.2129, -0.1106,  0.0277, -0.0481, -0.0890,  0.0557,
        -0.0140,  0.0040,  0.0066,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0126,  0.2692,  0.2016,  0.1635, -0.0081,  0.0007,  0.0162,  0.0518,
         0.0684, -0.0065, -0.0158,  0.0244,  0.0034,  0.0019, -0.0036,  0.0233,
         0.0054,  0.0140,  0.0256, -0.0091, -0.0748,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0292, -0.4090, -0.1159,  0.0307, -0.0138, -0.1956,  0.0391,  0.0276,
         0.0183,  0.0223,  0.0058, -0.0079, -0.0068,  0.0091, -0.0011,  0.0203,
        -0.0020,  0.0455,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0200, -0.0879, -0.1669, -0.0297, -0.0064,  0.0057,  0.0270,  0.0014,
        -0.0018, -0.0051, -0.0008,  0.0118, -0.0238, -0.0453, -0.0627, -0.0045,
        -0.0259, -0.0416, -0.0302, -0.0033, -0.0114, -0.0374, -0.0907, -0.0101,
        -0.0320, -0.0121, -0.0362, -0.0151, -0.0070, -0.0555,  0.0094, -0.0035,
         0.0009,  0.0014, -0.0217,  0.0004,  0.0139,  0.0192,  0.0203,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.3460e-02, -3.8463e-01,  1.5942e-02, -5.7165e-02,  5.6114e-03,
         2.2057e-03,  1.5721e-02,  1.3774e-02, -2.9991e-02, -1.9395e-02,
        -3.2045e-02, -2.3145e-02,  6.9760e-03,  5.7413e-03, -1.8849e-02,
        -7.7604e-03, -8.1087e-02, -8.4933e-03, -5.6937e-03, -1.6338e-02,
        -5.5810e-04,  2.9186e-03,  2.8803e-03,  3.1913e-03, -2.1054e-03,
         3.2114e-04, -2.4626e-02, -1.8854e-03, -4.9115e-02, -8.1926e-02,
        -3.4324e-03, -1.0236e-02, -2.5331e-02,  2.7457e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0209,  0.0580,  0.2899,  0.2708,  0.1119,  0.1049,  0.0313,  0.0445,
         0.0679,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0338,  0.1988,  0.0383,  0.0880,  0.0384,  0.0893,  0.0031,  0.0250,
         0.0041,  0.0074,  0.0320,  0.0117, -0.0014,  0.0181,  0.0111,  0.0554,
         0.0152,  0.1021, -0.0043, -0.0066,  0.0128, -0.0027,  0.0050,  0.0063,
         0.0223,  0.0031,  0.0097, -0.0059,  0.0037,  0.0033,  0.0126,  0.0560,
        -0.0027,  0.0366,  0.0048,  0.0097,  0.0010,  0.0047,  0.0052, -0.0079,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0120, -0.0470, -0.2070,  0.0027, -0.0516, -0.0152, -0.0031, -0.0114,
        -0.0608, -0.0512, -0.0028,  0.0009,  0.0031,  0.0042, -0.0014,  0.0031,
         0.0093, -0.0065, -0.0466,  0.0049, -0.0241, -0.0105, -0.0012, -0.0090,
         0.0102, -0.0204, -0.0151,  0.0002, -0.0615, -0.0545,  0.0054, -0.0253,
        -0.0042,  0.0045,  0.0066,  0.0096, -0.0415, -0.0575, -0.0018, -0.0056,
         0.0117,  0.0010,  0.0006, -0.0008,  0.0047,  0.0013, -0.0056,  0.0013,
         0.0042,  0.0042, -0.0135,  0.0374,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0030,  0.3238, -0.0151,  0.0619,  0.0059,  0.0308,  0.0641,  0.0309,
         0.0337, -0.0071, -0.0142,  0.0355,  0.1044,  0.0197, -0.0024,  0.0706,
         0.0460, -0.0145,  0.0497, -0.0263,  0.0030, -0.0375,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0161,  0.6769,  0.0201,  0.0083, -0.0269,  0.0029, -0.0078,  0.0368,
         0.0148, -0.0216, -0.0173,  0.0220, -0.0059,  0.0141, -0.0228,  0.0503,
         0.0060, -0.0016, -0.0038, -0.0164, -0.0073,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0116, -0.1552, -0.0301, -0.0175, -0.0062,  0.0045, -0.0152,  0.0157,
        -0.0111,  0.0018,  0.0044,  0.0027,  0.0022, -0.0095,  0.0055,  0.0035,
         0.0016,  0.0008, -0.0248,  0.0029,  0.0021,  0.0044, -0.0015,  0.0039,
        -0.0016, -0.0045, -0.0058, -0.0395,  0.0076, -0.0585, -0.1046, -0.0972,
        -0.0070, -0.0408, -0.0345, -0.0060,  0.0119,  0.0197,  0.0075,  0.0050,
        -0.0761, -0.0024,  0.0041, -0.0013, -0.0240,  0.0102, -0.0398, -0.0194,
        -0.0031,  0.0129, -0.0081, -0.0035,  0.0007,  0.0011,  0.0034],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0205, -0.0161,  0.0117, -0.0183, -0.0080, -0.0454, -0.1172,  0.0056,
        -0.0048,  0.0163, -0.0213,  0.0159, -0.0043, -0.0229, -0.0613, -0.0224,
         0.0022, -0.0252, -0.0113, -0.0538, -0.0100, -0.0088, -0.0282,  0.0059,
         0.0018,  0.0046, -0.0028, -0.1186,  0.0138,  0.0012, -0.0081, -0.0042,
        -0.0133, -0.0651,  0.0020, -0.0565, -0.0044, -0.0204, -0.0508, -0.0375,
         0.0030, -0.0157,  0.0021, -0.0108,  0.0058,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0488,  0.2280,  0.0799,  0.0773,  0.0114,  0.0242,  0.0296,  0.0266,
         0.0410, -0.0040, -0.0305,  0.0444,  0.0023,  0.0539,  0.0500, -0.0115,
         0.0247,  0.0489, -0.0011,  0.0398,  0.0858, -0.0282,  0.0007, -0.0075,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0525,  0.0336,  0.0075,  0.0092,  0.0014, -0.0027, -0.0042, -0.0357,
         0.0335, -0.0171, -0.0055,  0.0049, -0.0777, -0.0648,  0.0011,  0.0241,
        -0.0943, -0.0342, -0.0428, -0.0660,  0.0067, -0.0270, -0.0516, -0.0499,
         0.0095, -0.0314, -0.0704, -0.0338, -0.0310, -0.0658, -0.0022,  0.0020,
        -0.0058,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0822,  0.3809,  0.0146,  0.0617,  0.0215,  0.0176,  0.0053, -0.0318,
         0.1590, -0.0271,  0.0058,  0.0110, -0.0392,  0.0482, -0.0281,  0.0186,
         0.0093, -0.0208,  0.0173,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0614,  0.0265, -0.0422,  0.0515,  0.1120, -0.0198,  0.0735,  0.0696,
         0.0805, -0.0070,  0.0066,  0.0049, -0.0090, -0.0204, -0.0069, -0.0039,
         0.0060,  0.0764,  0.0296,  0.0157,  0.0528,  0.1545, -0.0139, -0.0170,
        -0.0384,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.9412e-02, -3.1851e-01, -9.6656e-02,  1.6811e-03,  2.5542e-04,
        -1.3031e-01, -1.7676e-01,  3.2641e-02, -1.3253e-02, -2.1001e-02,
         1.1440e-02,  8.4288e-03, -4.8541e-02, -2.3599e-02, -4.1527e-02,
        -3.5435e-02,  1.0550e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 2.9073e-02, -9.5410e-03,  2.2488e-02,  1.6567e-04,  7.3624e-03,
         6.6250e-03,  5.9029e-03, -3.3242e-02,  4.0591e-02,  5.6076e-02,
         2.7892e-02,  1.6686e-02, -3.5259e-02, -2.3110e-03, -2.3223e-03,
         4.3685e-03,  8.9199e-02,  1.7929e-01,  1.2609e-04,  4.9927e-02,
         1.4821e-02,  1.0427e-01,  1.4635e-01, -2.7611e-03, -1.3657e-02,
         1.9477e-02, -2.2677e-02, -9.7875e-03,  6.3296e-05, -3.6032e-03,
         1.5017e-02, -2.9062e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0597,  0.0708,  0.1665,  0.0102, -0.0033,  0.0013,  0.0784,  0.1121,
         0.0151, -0.0119, -0.0069, -0.0073,  0.0084, -0.0045,  0.0136, -0.0235,
         0.0103,  0.0251,  0.0516,  0.0066,  0.0509,  0.0112,  0.0877,  0.0854,
        -0.0200,  0.0360, -0.0127,  0.0091,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0593,  0.0647, -0.0160,  0.0234,  0.0018,  0.0644,  0.1670,  0.0101,
         0.0074,  0.0192,  0.0113,  0.0895,  0.0138,  0.0187, -0.0009,  0.0677,
         0.0150,  0.0794,  0.0099, -0.0212,  0.0348, -0.0009,  0.0294,  0.0130,
         0.0558, -0.0059,  0.0121, -0.0156,  0.0719,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0049,  0.0665,  0.0208,  0.0550,  0.1360,  0.0183,  0.0173, -0.0018,
         0.0122, -0.0039, -0.0069,  0.0080,  0.0271,  0.1272, -0.0034,  0.0501,
        -0.0091,  0.0920,  0.0206,  0.0423,  0.0191,  0.0079,  0.0385,  0.0099,
         0.0275,  0.0328,  0.0347,  0.0180,  0.0060, -0.0037,  0.0073, -0.0081,
         0.0632,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.8389e-03,  2.9127e-01, -2.2933e-02,  8.1761e-02, -2.4317e-02,
         2.1643e-02, -1.2676e-02,  3.0873e-02, -1.9510e-05,  5.6266e-02,
         4.0185e-02,  4.7507e-03,  1.4400e-02, -1.4806e-02,  7.5227e-03,
        -2.6726e-03,  3.7129e-02, -1.5990e-03,  1.4892e-01,  1.8730e-02,
         5.2944e-02, -2.5719e-02, -2.7712e-02,  5.8314e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.9204e-02, -2.3424e-01, -1.3068e-01, -6.6695e-03, -4.2298e-02,
        -6.4916e-03, -1.2215e-02, -5.7988e-02,  1.9982e-02, -6.4347e-02,
        -2.5745e-02, -2.5658e-02, -3.6595e-02, -4.6385e-03,  1.4954e-02,
        -1.5957e-04,  1.8944e-02, -4.7973e-02, -4.1432e-03, -1.5080e-01,
        -3.2719e-02, -9.0403e-03, -3.4517e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0302, -0.2242, -0.0219, -0.0141, -0.0208, -0.0429,  0.0219,  0.0094,
        -0.0078, -0.0097,  0.0059, -0.0239, -0.0095,  0.0133, -0.0148, -0.0265,
        -0.0671, -0.0194, -0.0182, -0.0025,  0.0009, -0.0069, -0.0148,  0.0116,
        -0.0451, -0.0162, -0.0074, -0.0004, -0.0092, -0.0529, -0.0435,  0.0024,
        -0.0325, -0.0161, -0.0331, -0.0503, -0.0250, -0.0263, -0.0016,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0417,  0.0665,  0.0257,  0.0241,  0.1121,  0.0111, -0.0045, -0.0105,
        -0.0071, -0.0038,  0.0039, -0.0014,  0.0053,  0.0039,  0.0088,  0.0848,
         0.0517,  0.0050,  0.0091, -0.0135,  0.0492,  0.0050, -0.0026, -0.0017,
         0.0733,  0.1265,  0.0084,  0.0128,  0.0125,  0.0521,  0.0046,  0.0257,
        -0.0017,  0.0098,  0.0170,  0.0031,  0.0024,  0.0492, -0.0171, -0.0006,
         0.0105,  0.0068,  0.0129,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.6115e-02,  3.0592e-01,  1.0998e-02, -4.9608e-05, -1.6599e-02,
        -2.0326e-03,  2.0293e-02,  1.1677e-04,  9.1508e-02, -4.4602e-03,
         2.8913e-03, -1.3433e-03,  1.2779e-02,  1.3681e-03, -6.5742e-03,
        -2.8248e-03,  2.5959e-02,  5.9153e-02,  9.3089e-03,  2.3776e-03,
        -6.5048e-03, -1.9297e-03,  1.5866e-02,  6.0380e-03, -4.2560e-03,
         2.3765e-03, -7.3740e-03,  2.5820e-03, -8.0549e-03, -8.9021e-03,
        -5.7842e-03,  2.2943e-02,  7.2321e-02, -2.1187e-02, -6.3617e-03,
         2.0757e-03, -7.1391e-03,  2.0493e-02,  3.1959e-02,  1.2024e-02,
         2.9727e-02,  4.6422e-02,  6.1929e-03, -4.6392e-03,  2.5684e-03,
        -1.3638e-03,  3.5784e-03, -4.1706e-03, -3.7537e-03, -3.2371e-03,
         4.1351e-04, -1.3161e-04, -4.3632e-03,  2.6856e-03,  2.2988e-03,
         4.6683e-03,  1.0940e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0121,  0.4635,  0.0397,  0.0092,  0.0441,  0.0093,  0.0598,  0.0954,
         0.0336,  0.0414,  0.0951, -0.0343, -0.0450,  0.0175,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0578, -0.1551, -0.0530,  0.0106,  0.0041, -0.0331, -0.0040, -0.0026,
        -0.0463, -0.0032, -0.0411, -0.1079,  0.0117, -0.0098, -0.0373, -0.0021,
         0.0020, -0.0266, -0.0157, -0.0545, -0.1375, -0.0101, -0.0383, -0.0052,
         0.0154,  0.0260, -0.0042,  0.0162, -0.0443,  0.0241,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1144, -0.2117,  0.0333, -0.2826, -0.1101,  0.0998, -0.0719,  0.0762,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0957, -0.1087, -0.0219,  0.0034, -0.0208, -0.1058, -0.0080, -0.0024,
         0.0053, -0.0169,  0.0064,  0.0032, -0.0270, -0.0457, -0.0099, -0.0039,
         0.0117,  0.0073, -0.0023, -0.0038,  0.0225, -0.0074, -0.0129, -0.0342,
        -0.0654, -0.0073, -0.0507, -0.0101, -0.0664, -0.1326,  0.0235,  0.0006,
        -0.0562,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0130, -0.2086, -0.0766, -0.0077, -0.0093, -0.0074, -0.0065, -0.0131,
         0.0009, -0.0155,  0.0042, -0.0046, -0.0171, -0.0044,  0.0019, -0.0054,
         0.0007, -0.0003, -0.0189,  0.0055,  0.0012,  0.0008, -0.0076,  0.0031,
         0.0012, -0.0049, -0.0084, -0.0300,  0.0034, -0.0003, -0.0068, -0.0095,
        -0.0347, -0.0461,  0.0219, -0.0152, -0.0183,  0.0016,  0.0054, -0.0015,
        -0.0013, -0.0153, -0.0057, -0.0348, -0.0018,  0.0053, -0.0006, -0.0039,
        -0.0077, -0.0034,  0.0003, -0.0178, -0.0135, -0.0261, -0.0060,  0.0132,
         0.0013, -0.0059,  0.0033, -0.0288, -0.0101, -0.0104, -0.0736, -0.0003,
        -0.0002, -0.0096, -0.0022,  0.0073, -0.0094,  0.0203, -0.0174],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0416, -0.0071, -0.0245, -0.0044, -0.0777, -0.0195, -0.1744, -0.0109,
         0.0027, -0.0031, -0.0016,  0.0068,  0.0072,  0.0091,  0.0162,  0.0021,
         0.0017,  0.0022, -0.0024, -0.0802, -0.0022, -0.0006,  0.0192, -0.0458,
        -0.0011, -0.0295, -0.0257,  0.0103, -0.0217, -0.0131, -0.0063, -0.0413,
        -0.0797, -0.0744, -0.0137, -0.0084, -0.0519, -0.0171, -0.0142,  0.0071,
        -0.0121,  0.0093,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0311,  0.1691,  0.0020, -0.0034,  0.0041, -0.0068, -0.0106, -0.0015,
         0.0105,  0.0605,  0.0159,  0.0524,  0.0873, -0.0114,  0.0065, -0.0012,
         0.0123,  0.0493,  0.0940,  0.0096,  0.0117,  0.0307, -0.0120, -0.0013,
        -0.0180,  0.0718,  0.0341,  0.0045,  0.0971,  0.0189,  0.0065, -0.0006,
        -0.0043, -0.0002,  0.0139, -0.0190,  0.0161,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0668, -0.5218, -0.0532, -0.1180, -0.0186, -0.0400, -0.0036,  0.0107,
        -0.0153, -0.0329, -0.0081,  0.0185, -0.0499, -0.0075,  0.0350,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0131,  0.0629, -0.0149,  0.0463,  0.1686,  0.1028, -0.0018,  0.0876,
         0.2137,  0.0346, -0.0062, -0.0099,  0.0051, -0.0060, -0.0513, -0.0049,
        -0.0101, -0.0893, -0.0411, -0.0298,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.8946e-02,  1.1065e-02, -8.2718e-03, -2.2572e-02, -6.5465e-02,
         1.2299e-02, -1.4012e-02,  3.3786e-02,  3.9557e-02,  2.8604e-02,
        -1.9019e-02, -2.6415e-02,  1.9980e-02,  2.6916e-02, -1.1617e-02,
        -1.6620e-02, -2.4319e-02, -1.7250e-02,  1.4101e-02, -9.2714e-03,
        -9.8346e-05,  1.3331e-02,  6.4585e-03, -2.3549e-03, -1.0479e-03,
        -2.4668e-02, -4.2172e-02,  3.2611e-03,  1.1718e-02, -2.7876e-02,
        -2.4368e-02,  3.3558e-02,  1.0469e-02,  1.8160e-02, -3.9092e-03,
        -2.0872e-02,  9.3600e-03, -2.4441e-02,  3.2671e-02, -1.4835e-02,
        -1.7447e-02, -8.3650e-02, -7.3186e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.7795e-02, -2.4767e-03, -8.4017e-03, -9.5228e-03, -4.6804e-02,
        -3.0115e-02, -4.4317e-02, -1.5636e-01, -5.3613e-03,  1.8557e-03,
         4.2277e-03, -1.9062e-02, -1.2613e-02, -6.2606e-02, -9.7492e-02,
         1.2375e-03,  3.7960e-03, -3.1946e-04, -9.4552e-03,  2.3208e-02,
        -4.7137e-02, -2.9461e-03,  4.0622e-02,  6.1099e-03,  1.9804e-02,
         1.2293e-02,  8.2153e-03, -2.5347e-03, -8.1486e-02, -2.7477e-03,
        -7.2995e-03, -3.7281e-02, -1.0475e-02, -4.7468e-02, -7.6990e-02,
        -2.9959e-03, -7.4214e-05, -6.4943e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0292, -0.3429, -0.1163, -0.1332,  0.0237, -0.0436,  0.0311,  0.0023,
        -0.0127, -0.0707, -0.0079, -0.0099, -0.0244,  0.0058,  0.0094,  0.0100,
         0.0169, -0.0406, -0.0042,  0.0274,  0.0147,  0.0232,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0438, -0.1018, -0.0052,  0.0055,  0.0184, -0.0078, -0.0304,  0.0221,
        -0.0236, -0.0799,  0.0121,  0.0088, -0.0047,  0.0056, -0.0108, -0.0388,
        -0.0504, -0.0042, -0.0231, -0.0380, -0.0034, -0.0184, -0.0604, -0.0233,
        -0.0299,  0.0047, -0.0026,  0.0084, -0.0046, -0.0655, -0.0055, -0.0495,
        -0.0110, -0.0018, -0.0360, -0.0373,  0.0152,  0.0009, -0.0161, -0.0704,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0512, -0.0252,  0.0019, -0.0311,  0.0034, -0.0521, -0.0030,  0.0044,
        -0.0430, -0.0158,  0.0025, -0.0125, -0.0144, -0.1072, -0.0787,  0.0030,
         0.0197, -0.0086, -0.0507, -0.0132,  0.0129,  0.0024,  0.0065,  0.0102,
        -0.0083, -0.1345, -0.0243, -0.0952, -0.1070, -0.0160,  0.0031, -0.0383,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.9461e-02, -3.3280e-01, -2.2299e-02,  2.1203e-02, -1.6942e-02,
         1.9497e-02, -1.2683e-03, -4.1485e-02, -1.1393e-01, -1.3113e-02,
        -2.5853e-03, -4.6112e-02,  5.6751e-03, -1.2454e-02, -1.4460e-02,
         1.7439e-03, -4.5412e-02, -5.4915e-03, -4.9257e-05,  2.0216e-02,
         4.4463e-03, -1.5836e-02, -3.4608e-03, -7.9938e-02, -7.1682e-03,
        -1.2685e-02, -4.8366e-02,  4.2263e-03,  2.0314e-03, -2.1225e-02,
        -6.1434e-03,  1.5379e-03, -2.6765e-03, -2.9554e-02,  4.5078e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0484,  0.0353,  0.1091,  0.0972,  0.0365, -0.0427,  0.1445,  0.1084,
         0.0099,  0.0962, -0.0376,  0.0414, -0.0194,  0.0444,  0.0129,  0.0240,
         0.0123, -0.0052,  0.0009, -0.0378,  0.0358,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0820, -0.3630,  0.0079, -0.0668, -0.1148, -0.0594, -0.0149, -0.0557,
        -0.0679,  0.0098, -0.0101, -0.0020,  0.0045,  0.0275, -0.0392,  0.0170,
        -0.0203, -0.0032,  0.0342,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0170, -0.0653, -0.1544, -0.0175, -0.0236, -0.0041,  0.0077, -0.0049,
        -0.0043, -0.0668, -0.1498, -0.0168, -0.0395, -0.1497, -0.0388, -0.0923,
         0.0143, -0.0783,  0.0137, -0.0413,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0067, -0.3880, -0.0142, -0.1053,  0.0290, -0.0159, -0.0309, -0.0590,
        -0.0086,  0.0013,  0.0011, -0.0186,  0.0063, -0.0432, -0.0013, -0.0586,
        -0.0895, -0.0084, -0.0491, -0.0229,  0.0225, -0.0147,  0.0048,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0551, -0.1868, -0.0693, -0.0249,  0.0177, -0.0539, -0.0083, -0.0548,
        -0.0311, -0.1325, -0.0373, -0.0665, -0.0240, -0.0639,  0.0129, -0.0002,
         0.0104, -0.0098, -0.0077, -0.0178, -0.0472, -0.0121, -0.0129,  0.0295,
         0.0133,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.8180e-02,  3.0637e-01,  3.6910e-02,  3.9387e-02,  6.6251e-03,
        -3.8671e-03, -1.4318e-02, -2.1797e-04, -5.3114e-02, -3.3841e-02,
         4.4298e-02, -8.6187e-05, -4.2415e-04, -6.0201e-03,  4.8478e-03,
         3.1617e-03, -3.8141e-03,  1.8906e-03, -1.6424e-02,  1.1355e-02,
         5.1221e-02,  8.4418e-02,  2.0357e-02,  5.3847e-02,  2.1428e-02,
         1.7483e-02,  2.1568e-03, -3.4028e-03,  7.7309e-03,  1.8042e-02,
         4.7003e-03,  1.1461e-03,  2.2015e-02,  1.5499e-02,  1.1400e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0020,  0.0173, -0.0252, -0.0842, -0.0129, -0.0294, -0.1048, -0.0157,
        -0.0941, -0.0075, -0.0252, -0.0318,  0.0361, -0.0815, -0.0807, -0.1500,
         0.0042, -0.1190, -0.0209,  0.0122,  0.0172, -0.0281,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0768,  0.0067,  0.0384, -0.0250, -0.0149, -0.0167, -0.0047, -0.0160,
        -0.0333,  0.0367,  0.1588,  0.1281, -0.1597, -0.0463, -0.0339, -0.0276,
         0.0318,  0.0259,  0.0266,  0.0695, -0.0227,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0004, -0.4028, -0.0382, -0.0047,  0.0330, -0.0045, -0.1028, -0.1579,
        -0.0042, -0.0062,  0.0017,  0.0057,  0.0043, -0.0641,  0.0061, -0.0800,
         0.0267,  0.0357, -0.0212,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0456, -0.0163, -0.0132,  0.0150,  0.0125, -0.0716,  0.0053, -0.0029,
         0.0151, -0.0004, -0.0095, -0.0121, -0.0063, -0.0286, -0.0024, -0.0370,
        -0.0675, -0.0560,  0.0011, -0.0435, -0.0187,  0.0088,  0.0075,  0.0198,
        -0.0424, -0.0106, -0.0070, -0.0161, -0.0283, -0.0770,  0.0004,  0.0124,
        -0.0519, -0.1287, -0.0033, -0.0073, -0.0037, -0.0020, -0.0055, -0.0377,
        -0.0023, -0.0009,  0.0014, -0.0024, -0.0016,  0.0271, -0.0133],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0171, -0.3262, -0.1195, -0.0021,  0.0184, -0.0146,  0.0102, -0.0037,
        -0.0467, -0.0070,  0.0013,  0.0062,  0.0118, -0.0024,  0.0078, -0.0028,
        -0.0570,  0.0005,  0.0097, -0.0045, -0.0063, -0.0241, -0.1308, -0.0897,
         0.0109, -0.0073,  0.0036,  0.0033,  0.0018,  0.0018,  0.0017,  0.0078,
        -0.0184, -0.0230,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.1726e-02, -3.8154e-01, -2.2793e-02,  4.3818e-02, -8.4418e-03,
        -5.7026e-02,  1.2670e-02,  1.7087e-03,  1.8729e-03, -1.1215e-02,
         2.6606e-02, -3.3430e-03, -4.2663e-02, -7.4114e-02,  1.0202e-02,
        -1.1427e-02,  7.6261e-03, -5.5789e-03,  2.2395e-02, -6.7417e-02,
        -6.7444e-03, -7.1348e-03,  3.4752e-04, -1.3203e-03,  1.8080e-03,
         5.7623e-04, -3.9807e-04, -2.0339e-02, -1.1440e-02, -4.4874e-02,
        -3.2912e-03,  4.8272e-03, -4.3169e-02,  9.5425e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0539, -0.3687,  0.0437,  0.0199, -0.0430, -0.1133,  0.0095, -0.0143,
         0.0033, -0.1450, -0.0031, -0.0093, -0.0428, -0.0990, -0.0134, -0.0064,
        -0.0037, -0.0076,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0427, -0.0205,  0.0199, -0.0202, -0.0021, -0.0103, -0.0265, -0.0148,
         0.1217,  0.2289, -0.0052, -0.0294,  0.0209,  0.0258,  0.0097,  0.0228,
         0.0456, -0.0004,  0.0310,  0.1716,  0.0144,  0.0219,  0.0282,  0.0015,
        -0.0074, -0.0302,  0.0266,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0152, -0.0216,  0.0067, -0.0020, -0.0030, -0.0070, -0.0154, -0.0375,
        -0.2059, -0.0016,  0.0098, -0.0318, -0.1490,  0.0106,  0.0091, -0.0051,
        -0.0189, -0.0620, -0.0003, -0.1929,  0.0104,  0.0256, -0.0672,  0.0252,
        -0.0330,  0.0330,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0026, -0.2749,  0.0052, -0.0987, -0.1083, -0.0095, -0.0150, -0.0204,
        -0.0521, -0.0355, -0.0427, -0.0379,  0.0138, -0.0218, -0.0309, -0.0686,
        -0.0601, -0.0514, -0.0090, -0.0014,  0.0405,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0345, -0.3242, -0.1807, -0.1093, -0.0056, -0.0488, -0.1242, -0.0019,
        -0.0065, -0.0016,  0.0007, -0.0700,  0.0382, -0.0301, -0.0236,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0780,  0.0345,  0.0065,  0.0277, -0.0130, -0.0336, -0.0126, -0.0786,
        -0.0072, -0.0191,  0.0284, -0.0676,  0.2985,  0.0929,  0.0284,  0.0177,
        -0.0852,  0.0706,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0247, -0.3182, -0.1072, -0.1259,  0.0169, -0.0096,  0.0327, -0.0126,
         0.0182, -0.0428, -0.0782, -0.0327,  0.0085, -0.0797,  0.0011,  0.0202,
        -0.0045,  0.0214,  0.0448,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.7717e-02, -1.1877e-01, -9.6786e-02, -3.7935e-02, -4.2266e-02,
        -2.1065e-02,  2.0750e-02, -1.0743e-02, -8.3407e-03, -4.6169e-05,
         4.0651e-03, -7.3841e-03, -2.1551e-02, -4.2212e-03, -2.6441e-03,
        -3.0896e-03, -3.6655e-02,  7.2782e-03,  1.1322e-02,  6.2466e-03,
        -7.1034e-02, -8.4891e-03, -5.8665e-02, -1.1699e-02, -4.7770e-02,
        -1.1861e-02, -5.3234e-02,  1.2414e-03,  5.0379e-03, -3.3978e-02,
         4.9704e-03, -1.0408e-02, -4.7421e-02, -3.0301e-03, -2.6289e-02,
        -6.6294e-03, -5.8616e-02,  1.2571e-02, -9.5062e-03, -3.5551e-03,
         2.7801e-03,  3.7227e-03, -1.7136e-03, -6.9051e-03], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.5929e-03, -2.4398e-01, -2.6230e-02, -7.0275e-02, -4.2583e-02,
        -9.7017e-03, -2.2182e-02, -9.5023e-03, -9.5873e-02,  4.2612e-03,
        -2.8100e-02,  1.7900e-04,  1.9809e-02,  5.9990e-04,  1.1829e-02,
         1.0596e-02, -1.2532e-02, -6.5037e-02,  1.4552e-03, -3.0463e-02,
        -3.6656e-02, -1.5992e-03, -8.3761e-03,  5.2098e-03,  1.0228e-02,
        -1.8836e-03, -8.8907e-03, -1.1905e-01,  6.0627e-03, -2.0063e-02,
         1.3110e-02, -3.0506e-02,  1.4231e-02,  3.3579e-03, -7.9932e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.7403e-02, -4.0993e-01, -1.6763e-02, -6.0205e-02, -1.9317e-02,
         1.3541e-02, -8.8846e-02, -7.9092e-02, -2.1372e-02, -6.8997e-03,
        -1.1446e-02, -3.0768e-02, -4.8129e-02, -1.0028e-02, -4.9234e-03,
        -2.2103e-02, -1.1628e-02,  6.1228e-03,  2.3002e-02, -2.0367e-03,
         1.5644e-03,  3.9267e-03,  6.0747e-02, -2.0292e-04,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0231, -0.3399, -0.0073, -0.0602, -0.0256, -0.0142, -0.0091, -0.0083,
        -0.0752,  0.0035, -0.0252,  0.0062,  0.0133, -0.0210, -0.0384,  0.0036,
         0.0187, -0.0028, -0.0427, -0.0015,  0.0071, -0.0454,  0.0228, -0.0415,
         0.0061, -0.0429, -0.0207, -0.0140, -0.0047, -0.0188, -0.0191, -0.0061,
        -0.0087, -0.0023,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0126,  0.0088,  0.0031, -0.0439,  0.0120,  0.0095, -0.1981, -0.0570,
         0.0009, -0.0131,  0.0043, -0.0238, -0.0917, -0.0059, -0.0928, -0.1582,
        -0.0233, -0.0052,  0.0014, -0.0085,  0.0023, -0.0043, -0.0067, -0.0605,
        -0.0349, -0.0219, -0.0397, -0.0072, -0.0119,  0.0364,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0236, -0.1673,  0.0039, -0.1522, -0.0194, -0.0039, -0.0016, -0.0537,
        -0.1330, -0.3195,  0.0306, -0.0184, -0.0729,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0634, -0.0062,  0.0749,  0.0191, -0.0205, -0.0473, -0.0309, -0.0726,
        -0.2397, -0.0251,  0.0093, -0.0840, -0.0597, -0.0701, -0.1110, -0.0091,
        -0.0254,  0.0318,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0639,  0.2426,  0.4454,  0.1081,  0.0807, -0.0398,  0.0183, -0.0011,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.2063e-03, -1.8335e-01, -1.2098e-02, -2.2138e-02, -2.5142e-02,
        -3.2319e-03, -1.3481e-02, -5.4390e-03,  1.3599e-02, -5.0683e-03,
         5.7579e-03,  2.0216e-02, -1.2228e-02, -5.6241e-03, -1.4743e-03,
        -1.0423e-02,  2.7242e-03, -1.8949e-03,  6.8416e-05, -9.7543e-04,
         6.6879e-03, -7.8677e-03, -3.5039e-02, -3.1692e-02, -1.5011e-03,
        -2.6460e-02, -8.6175e-03, -4.7390e-03, -6.8173e-02, -4.5664e-02,
        -1.4170e-02, -3.6923e-02, -8.5315e-03, -1.1196e-01, -2.6140e-02,
        -5.9293e-02, -6.2576e-02, -1.3391e-02,  1.7333e-02,  4.6319e-03,
        -1.0427e-02, -3.7988e-03, -3.0373e-03,  1.1085e-02,  1.0989e-02,
        -4.7632e-03,  1.8363e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1842,  0.2112, -0.0936,  0.3373,  0.0540, -0.0112,  0.0163,  0.0359,
        -0.0036,  0.0416,  0.0109,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0950, -0.2393, -0.0262, -0.0794, -0.0188, -0.0173, -0.0353, -0.2193,
         0.0159, -0.0963, -0.0534,  0.0009, -0.0632,  0.0013, -0.0190, -0.0073,
         0.0079, -0.0040,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0463, -0.1084, -0.2681,  0.0440, -0.0012,  0.0032, -0.2061, -0.0303,
        -0.1085, -0.0521,  0.0014, -0.0259, -0.0144,  0.0361, -0.0540,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0403, -0.4129, -0.0350, -0.0588, -0.0073, -0.0650, -0.1073,  0.0235,
        -0.0157, -0.0673, -0.0992, -0.0039, -0.0399, -0.0075, -0.0166,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1133,  0.0056, -0.0031,  0.0036, -0.0040,  0.0295,  0.0082,  0.0332,
        -0.1099, -0.0202, -0.0368, -0.0595, -0.1369, -0.0301, -0.0079,  0.0282,
        -0.0493,  0.0147,  0.0227, -0.0175, -0.0695, -0.1265,  0.0148, -0.0246,
        -0.0177,  0.0126,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1821, -0.0669, -0.0039,  0.0111, -0.0194, -0.0101,  0.0416, -0.2040,
        -0.3633,  0.0092, -0.0126, -0.0278,  0.0034, -0.0104, -0.0341,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0257, -0.3120,  0.0088, -0.0587, -0.0528, -0.0253, -0.0339,  0.0057,
         0.0017, -0.1635,  0.0169,  0.0164, -0.0640, -0.0116,  0.0022, -0.0020,
        -0.0192, -0.0043, -0.0623,  0.0196, -0.0143,  0.0012,  0.0144,  0.0149,
         0.0425, -0.0061,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0280, -0.0123,  0.0057,  0.0051, -0.0064, -0.0366,  0.0034,  0.0258,
        -0.0413, -0.0945, -0.2170, -0.0367, -0.0203, -0.0050, -0.1040, -0.0842,
         0.0312, -0.0394,  0.0021, -0.0513, -0.0344, -0.0283, -0.0189, -0.0388,
        -0.0055, -0.0239,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0491, -0.2917,  0.0301,  0.0121, -0.0074,  0.0069, -0.0257, -0.0647,
        -0.0123, -0.0355, -0.0081, -0.0139, -0.0217,  0.0117, -0.0205, -0.0188,
        -0.0047,  0.0043, -0.0102, -0.0052,  0.0253, -0.0096,  0.0298, -0.0774,
        -0.0903,  0.0037, -0.0166, -0.0100,  0.0012,  0.0013, -0.0046, -0.0305,
        -0.0008, -0.0056,  0.0126,  0.0260,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.9224e-03, -3.9020e-02, -4.0161e-02,  2.1278e-02, -3.9966e-02,
        -9.4425e-04, -4.8771e-02, -1.0656e-01, -1.9799e-03,  7.1471e-03,
        -2.4408e-02, -8.2118e-04,  5.4754e-04,  1.6201e-02,  7.5480e-05,
        -7.6315e-03,  7.3552e-03, -2.1363e-03,  2.0711e-03,  1.6070e-02,
        -1.9239e-02, -7.0800e-02, -1.5730e-02, -2.2489e-03, -4.0048e-02,
        -1.1451e-01,  2.3672e-03, -6.5663e-02, -3.1792e-02, -5.1825e-03,
        -2.6713e-02, -5.0990e-02, -1.0529e-01, -1.0269e-02, -4.1027e-02,
         1.3068e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0314, -0.0573, -0.0128, -0.0391,  0.0498, -0.1374, -0.0655, -0.0387,
        -0.0016, -0.0061, -0.0128, -0.0634, -0.2305,  0.0324,  0.0110,  0.0044,
        -0.1057, -0.0105, -0.0274,  0.0622,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0674, -0.1012, -0.0334, -0.0236,  0.0042, -0.0116, -0.0148, -0.0378,
        -0.0261, -0.0232,  0.0009,  0.0045, -0.0030, -0.0117, -0.1557, -0.0474,
        -0.0311, -0.0670, -0.0124,  0.0058, -0.0157,  0.0003,  0.0010,  0.0033,
        -0.0060,  0.0133,  0.0010,  0.0133, -0.0160, -0.0194, -0.0928, -0.0170,
        -0.0250, -0.0410,  0.0060, -0.0320,  0.0073,  0.0068,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.3233e-02, -1.0168e-01, -1.0259e-01, -1.0069e-01, -6.9229e-03,
        -7.9103e-04, -2.1652e-02,  4.2201e-03,  1.6259e-02, -1.0206e-02,
        -5.8072e-02, -9.2569e-02, -5.1128e-03,  2.6425e-03, -1.2864e-02,
        -5.8042e-02, -4.8502e-03, -7.0358e-02, -1.5745e-03,  8.1858e-03,
         1.5200e-03,  4.2775e-03,  2.0997e-02,  8.2469e-03,  7.4309e-03,
        -5.2916e-03,  1.0457e-03, -2.9848e-02, -6.9008e-02, -8.9152e-03,
        -3.1018e-03, -8.7640e-03,  6.2801e-05, -3.8689e-03, -2.9408e-02,
        -3.6525e-03, -1.5478e-02,  1.7290e-02, -1.0161e-03,  1.0382e-02,
        -7.8812e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.6596e-02, -4.9991e-01, -6.8582e-04,  2.5483e-02, -2.3676e-02,
        -8.2102e-02, -4.9115e-03, -2.6259e-04, -1.8733e-02, -5.0845e-03,
         2.6501e-03, -2.0023e-02, -6.1321e-02, -1.3602e-02,  1.0934e-02,
        -5.7883e-02, -2.0830e-03, -4.3114e-02, -6.3918e-03,  1.2385e-02,
        -2.3027e-03, -2.9866e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0290,  0.0081,  0.0118, -0.0014,  0.0008, -0.0199, -0.0164,  0.0218,
        -0.0031,  0.0104,  0.0081, -0.0767, -0.0024, -0.0370,  0.0106, -0.1116,
         0.0092, -0.0463, -0.0047, -0.0541, -0.0778,  0.0101,  0.0237, -0.0646,
        -0.0093, -0.0098, -0.0151, -0.0372, -0.0894, -0.0183, -0.0485, -0.0365,
        -0.0084,  0.0275, -0.0402,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0588, -0.0750, -0.2656, -0.0339, -0.0846, -0.0096,  0.0051,  0.0080,
         0.0069, -0.0132, -0.0322,  0.0060, -0.0086,  0.0052,  0.0026, -0.0090,
         0.0696, -0.0155, -0.0462, -0.0008, -0.0208,  0.0264, -0.0242, -0.0844,
        -0.0042, -0.0007, -0.0358, -0.0267,  0.0205,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0046, -0.3275, -0.0758, -0.0512,  0.0037, -0.0555,  0.0225, -0.0220,
        -0.0027, -0.0137, -0.0157, -0.0570,  0.0017, -0.0169, -0.0345, -0.0755,
        -0.0344, -0.0135, -0.0593, -0.0146, -0.0126, -0.0084, -0.0155,  0.0237,
         0.0377,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0379,  0.0071,  0.0033,  0.0275, -0.1702,  0.0078, -0.0093, -0.1478,
        -0.1630,  0.0129,  0.0066,  0.0456,  0.0010, -0.0038, -0.0985, -0.1100,
         0.0337, -0.0338,  0.0273,  0.0035, -0.0256, -0.0171, -0.0067,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0074,  0.0007,  0.0139, -0.0071,  0.0151,  0.0212,  0.0113,  0.0056,
        -0.0061,  0.0073,  0.0012, -0.0027, -0.0829, -0.1666, -0.0241, -0.0404,
        -0.0047, -0.0387,  0.0112,  0.0082, -0.0022,  0.0265,  0.0041,  0.0038,
         0.0067, -0.0258, -0.0763,  0.0207, -0.0928, -0.0090, -0.1051, -0.0193,
        -0.0087, -0.0327, -0.0486, -0.0154, -0.0111, -0.0046, -0.0098,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0482, -0.0636, -0.0658,  0.1461,  0.2330,  0.0265,  0.0020,  0.0862,
         0.2228, -0.0243, -0.0250,  0.0565,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.0896, -0.1195, -0.2261,  0.0031, -0.0131, -0.0075, -0.0682,  0.0151,
         0.0018, -0.0277, -0.0044,  0.0227,  0.0050, -0.0094, -0.0438,  0.0028,
        -0.0110, -0.0742,  0.0006, -0.0043, -0.0155, -0.0405,  0.0063, -0.0045,
        -0.0055,  0.0051, -0.0176, -0.0140,  0.0169, -0.0171,  0.0087,  0.0050,
        -0.0112,  0.0100, -0.0036,  0.0102, -0.0106, -0.0040,  0.0018, -0.0009,
         0.0088,  0.0069, -0.0065, -0.0136, -0.0051,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0698,  0.3153,  0.0763,  0.1248, -0.0508,  0.0071,  0.0252,  0.0458,
         0.0885,  0.0760,  0.0429, -0.0773,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0661,  0.3373, -0.0412,  0.1558, -0.1646,  0.0259,  0.0560, -0.1530,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.8871e-02,  2.0205e-01,  3.1165e-02, -6.3684e-03,  6.4871e-02,
        -2.5573e-03,  4.1596e-03,  5.6644e-03,  1.5764e-02,  5.8122e-02,
         1.6113e-02, -3.7800e-03,  2.5116e-03, -2.6011e-03, -2.9784e-03,
        -4.0967e-03, -7.8800e-03, -3.9957e-03, -1.3260e-03,  1.9667e-03,
        -1.0898e-03, -5.4577e-03,  2.8933e-03, -1.2598e-03, -1.1029e-04,
         6.8259e-04,  3.2818e-03, -1.4773e-03,  1.8672e-02,  1.9479e-03,
         5.3026e-03,  6.9270e-03,  5.1729e-04,  2.3285e-03,  1.7663e-04,
        -6.0716e-03, -3.4184e-03, -3.1195e-03,  3.0698e-03, -4.6616e-04,
         1.3096e-04, -6.4879e-03,  2.5875e-03,  1.0580e-03, -6.7998e-03,
        -4.5668e-02,  1.0148e-01,  4.1526e-03,  6.3409e-02,  1.0594e-01,
         4.2864e-02,  2.3804e-04,  3.2145e-02,  2.7215e-03,  1.9618e-02,
         2.0137e-03,  3.2285e-03, -7.7076e-03,  5.3863e-03, -1.5873e-02,
        -5.3758e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0342,  0.0020,  0.0165, -0.0188, -0.0901,  0.0062, -0.0251, -0.0038,
        -0.0015,  0.0086, -0.0019, -0.1290, -0.0231, -0.0642, -0.0539, -0.0039,
        -0.0004,  0.0250, -0.0443, -0.0410, -0.0613, -0.0976, -0.0062, -0.0206,
        -0.0707, -0.0181, -0.0117, -0.0260, -0.0368, -0.0228,  0.0165, -0.0020,
        -0.0164,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0515, -0.0441, -0.1522, -0.0646, -0.0063, -0.0111, -0.0547, -0.0121,
         0.0059, -0.0090,  0.0162,  0.0108, -0.0226,  0.0070, -0.0117,  0.1171,
        -0.0040, -0.0938, -0.2045, -0.0045,  0.0180, -0.0166, -0.0386,  0.0231,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0375, -0.0772, -0.0957, -0.0232, -0.0084, -0.0918,  0.0318, -0.0652,
         0.0153, -0.1328, -0.0840, -0.0607, -0.1003, -0.0033, -0.0177, -0.0011,
        -0.0042, -0.0152,  0.0316,  0.0063, -0.0383, -0.0027,  0.0063, -0.0492,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0400,  0.0041, -0.0867, -0.1053, -0.0106,  0.0019, -0.0144, -0.0704,
         0.0024, -0.0019, -0.0032, -0.0014,  0.0002, -0.0036, -0.0601, -0.0572,
        -0.0662,  0.0022, -0.0129, -0.0064, -0.0677, -0.0168, -0.0362, -0.1105,
         0.0045, -0.0178, -0.0917, -0.0185, -0.0092, -0.0120, -0.0236, -0.0195,
         0.0135,  0.0077,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0244,  0.0026,  0.0023, -0.0102, -0.0904,  0.0025,  0.0014, -0.0045,
         0.0107,  0.0053, -0.0004,  0.0518,  0.0440, -0.1562, -0.0338, -0.0451,
        -0.1080,  0.0084, -0.0729, -0.0165, -0.0762, -0.1546,  0.0282, -0.0187,
        -0.0307,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0855,  0.0304, -0.0387, -0.0024, -0.0278, -0.0284, -0.1490, -0.0281,
        -0.2114, -0.2712, -0.0313, -0.0351, -0.0264, -0.0093, -0.0126, -0.0125,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.3533e-02, -3.3387e-01,  5.8651e-04, -6.7368e-02, -6.2216e-02,
        -5.5792e-02, -2.0733e-01, -2.2039e-02,  2.4533e-02, -6.4692e-03,
         9.7427e-03, -3.9054e-03, -3.8970e-02,  5.2694e-03,  6.7982e-03,
         2.3018e-04, -2.1629e-02, -4.3422e-02, -2.5280e-03, -2.3680e-04,
         2.1992e-02,  5.3957e-03,  2.6138e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1010,  0.0302, -0.0436, -0.1197, -0.4082,  0.0383,  0.0666,  0.0080,
        -0.0239,  0.0179, -0.0039,  0.0469,  0.0160, -0.0759,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0862, -0.1100, -0.0924, -0.1378, -0.0194,  0.0459, -0.0243,  0.0062,
        -0.0306, -0.0032, -0.0532, -0.0896, -0.0125, -0.0014,  0.0007,  0.0519,
         0.0309, -0.0804,  0.0089, -0.0836, -0.0022, -0.0126, -0.0070, -0.0092,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0064, -0.0149, -0.1077,  0.0209, -0.1662, -0.0638, -0.4482,  0.0553,
        -0.0363,  0.0348, -0.0454,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0293,  0.0162,  0.0535,  0.0045,  0.0130, -0.0754, -0.2091, -0.0154,
        -0.0983, -0.0399, -0.1010,  0.0039, -0.0559, -0.0024, -0.0745, -0.0145,
        -0.0218, -0.1020,  0.0459, -0.0063, -0.0171,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0664,  0.0257,  0.0244,  0.0208, -0.0158,  0.0006,  0.0131, -0.3354,
        -0.1710, -0.2167, -0.0391,  0.0278,  0.0074, -0.0357,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0680,  0.0399,  0.0140, -0.0027,  0.0276,  0.0036,  0.0273, -0.0123,
         0.0006,  0.0649,  0.0531,  0.0143,  0.4242,  0.0181,  0.0279,  0.0208,
         0.0685,  0.0287,  0.0307,  0.0178,  0.0350,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0033,  0.3135,  0.0006,  0.1056, -0.0678,  0.1164,  0.0607,  0.1853,
         0.0217,  0.0473,  0.0393, -0.0105, -0.0280,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0188, -0.1083, -0.0188, -0.0155, -0.0136, -0.0187, -0.0544, -0.1023,
        -0.0425,  0.0029, -0.0140,  0.0133,  0.0121,  0.0036,  0.0010, -0.0069,
        -0.0007, -0.0047, -0.0066,  0.0073,  0.0016,  0.0093,  0.0097, -0.0006,
        -0.1308,  0.0255, -0.0790, -0.0619, -0.0740, -0.0043, -0.0209,  0.0034,
         0.0047, -0.0121,  0.0006,  0.0150, -0.0261,  0.0027, -0.0016, -0.0060,
        -0.0201,  0.0046,  0.0049, -0.0032,  0.0118], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0118,  0.0334,  0.0684, -0.0127, -0.0039, -0.0185,  0.0018,  0.0304,
        -0.0015,  0.0127,  0.1278,  0.0254,  0.0167,  0.0223,  0.0756,  0.0189,
         0.1605, -0.0204,  0.0051, -0.0051,  0.0096,  0.0660,  0.0129,  0.0598,
        -0.0148, -0.0081,  0.0114,  0.0111,  0.0713,  0.0214, -0.0071,  0.0106,
         0.0096, -0.0134,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0871,  0.0759,  0.0096, -0.0012,  0.0015,  0.0143, -0.0162, -0.0023,
        -0.0330,  0.0040,  0.0198, -0.0030,  0.0015,  0.0003, -0.0157, -0.0141,
         0.0040, -0.0140,  0.0312,  0.2189,  0.2800,  0.0257, -0.0169,  0.0478,
         0.0099, -0.0520,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0381, -0.3359,  0.0558, -0.1024, -0.0240, -0.0148, -0.0106, -0.0820,
         0.0124, -0.0075,  0.0047,  0.0194,  0.0090,  0.0076,  0.0007, -0.0485,
        -0.1071,  0.0114, -0.0394, -0.0018, -0.0382,  0.0027,  0.0259,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0198, -0.1376, -0.0182, -0.0361, -0.0035,  0.0077,  0.0042, -0.0464,
        -0.0974, -0.0219,  0.0141, -0.0174,  0.0024, -0.0052, -0.1388,  0.0149,
        -0.0169, -0.0231, -0.0639, -0.0257, -0.0103, -0.0086,  0.0055,  0.0036,
         0.0015, -0.0082, -0.0328, -0.0888, -0.0033, -0.0150,  0.0007,  0.0065,
        -0.0128,  0.0063, -0.0005, -0.0172, -0.0376, -0.0063,  0.0034,  0.0159,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0228,  0.0129,  0.0088, -0.0338, -0.0809, -0.0114, -0.0281, -0.0731,
         0.0566, -0.0239, -0.0087, -0.0002, -0.0351,  0.0030, -0.0347, -0.0630,
        -0.0418, -0.0335, -0.0172, -0.0548,  0.0074, -0.1099, -0.0520, -0.0079,
        -0.0173,  0.0037, -0.0004, -0.0489, -0.0163, -0.0338, -0.0310,  0.0082,
        -0.0101,  0.0090,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0751, -0.0364, -0.0179, -0.0375, -0.0924,  0.0132,  0.0158,  0.0045,
         0.0039,  0.0066,  0.0074, -0.0069,  0.0233,  0.0199, -0.1136,  0.0251,
         0.0009, -0.0016, -0.0549, -0.0077, -0.0041, -0.0033, -0.0003,  0.0142,
        -0.0307, -0.0957, -0.0074,  0.0095, -0.0073,  0.0053,  0.0156, -0.0014,
         0.0004, -0.0024, -0.0421, -0.0053, -0.0157, -0.0343, -0.0112, -0.0234,
        -0.0272,  0.0063, -0.0235, -0.0056,  0.0030,  0.0031, -0.0063,  0.0053,
        -0.0039, -0.0070,  0.0036,  0.0109], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0138, -0.0024,  0.0224,  0.0120,  0.0540,  0.0004, -0.0070, -0.0122,
         0.0112,  0.0017, -0.0180, -0.0065,  0.0211,  0.1989,  0.0398,  0.0484,
         0.1631,  0.0215,  0.0505,  0.0044,  0.0281,  0.0031,  0.0322,  0.1552,
         0.0041,  0.0062,  0.0142,  0.0083, -0.0074, -0.0008, -0.0313,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.2981e-02, -1.4625e-01,  6.6995e-03, -1.3473e-02,  1.4598e-02,
         8.7479e-03, -2.3526e-02, -2.0660e-02, -7.5718e-03,  2.7412e-02,
        -8.9296e-05, -4.7107e-02, -1.1115e-03,  3.8903e-03, -7.6047e-02,
        -7.2152e-02, -6.6004e-03, -1.0085e-01, -1.1084e-01, -1.4791e-01,
        -3.1869e-02,  6.7788e-03, -1.6975e-02,  1.5333e-02,  7.0529e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1045,  0.0380, -0.0133,  0.0346, -0.0145, -0.2061, -0.4147, -0.0391,
         0.0129, -0.0317, -0.0481, -0.0426,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0210, -0.2627, -0.0322, -0.0576, -0.1090, -0.0267,  0.0149,  0.0029,
         0.0048,  0.0047, -0.0071,  0.0065, -0.0532, -0.1011,  0.0200, -0.0490,
        -0.0167, -0.0034, -0.0024, -0.0097, -0.0355,  0.0003, -0.0129, -0.0032,
        -0.0083, -0.0859, -0.0087, -0.0064,  0.0333,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.7236e-02, -3.5843e-01,  3.1030e-02,  1.6127e-03, -7.2584e-05,
        -1.1888e-01, -2.9546e-01,  1.5716e-02, -2.9227e-02,  1.0956e-02,
        -4.3468e-02, -7.9152e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1039, -0.3291, -0.0224, -0.1209,  0.0137, -0.0948, -0.1358, -0.0161,
        -0.0792, -0.0246,  0.0194,  0.0403,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0490,  0.0455, -0.1138, -0.0169, -0.1532, -0.0005, -0.0194, -0.0124,
        -0.2226,  0.0064,  0.0178, -0.1069, -0.0091, -0.0380,  0.0003, -0.1046,
        -0.0256,  0.0035,  0.0155, -0.0389,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0202, -0.2132,  0.0138, -0.0861, -0.1731,  0.0113, -0.0259, -0.0114,
        -0.0200, -0.0113, -0.0016, -0.0612, -0.0377, -0.0138,  0.0044, -0.0031,
        -0.0139,  0.0078, -0.0016, -0.0059, -0.0228,  0.0173,  0.0147,  0.0024,
        -0.0033,  0.0164, -0.0080, -0.0418, -0.0245, -0.0003,  0.0109, -0.0072,
        -0.0243, -0.0505, -0.0008,  0.0072,  0.0107,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0660,  0.6652,  0.0314, -0.0822,  0.0291,  0.0104,  0.0244, -0.0430,
         0.0483,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0253,  0.0498,  0.0022, -0.1400, -0.0235, -0.0135, -0.0564, -0.0892,
         0.0094, -0.0117, -0.0090, -0.0124, -0.1318, -0.1020,  0.0013, -0.0267,
         0.0028,  0.0056, -0.0628, -0.0977,  0.0310,  0.0020, -0.0148,  0.0177,
         0.0156, -0.0028,  0.0430,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.2547e-02, -2.9242e-01,  1.0145e-02, -6.6144e-02, -8.7107e-03,
        -2.8557e-02,  2.5182e-03, -1.3003e-03, -6.8767e-03,  7.6680e-03,
         5.0811e-02, -3.7540e-02, -4.0689e-03,  1.7980e-02, -9.3165e-03,
         5.1288e-03, -4.3850e-02,  4.9006e-03,  1.5851e-02,  1.0487e-02,
         1.8225e-02, -4.3692e-02, -4.0607e-02,  4.7787e-03,  1.5726e-03,
         4.8514e-03, -8.7088e-03, -4.3281e-02, -1.1948e-01, -5.4389e-03,
         1.1705e-02,  1.9499e-04,  2.5266e-02, -5.3759e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0367, -0.0071,  0.0048, -0.0065, -0.0278, -0.0095,  0.0063,  0.0067,
         0.0085,  0.0277, -0.0036, -0.0820,  0.0017, -0.0063,  0.0017, -0.0387,
        -0.1070, -0.0071, -0.0318, -0.0094, -0.0003, -0.0094, -0.0388, -0.0103,
         0.0149,  0.0068,  0.0153, -0.0016, -0.0023, -0.0034,  0.0161, -0.0860,
        -0.1249, -0.0056, -0.0336, -0.0331, -0.0710, -0.0046,  0.0100, -0.0082,
        -0.0132, -0.0429,  0.0013,  0.0097,  0.0056,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0702, -0.0008, -0.0390, -0.0104, -0.0338, -0.1664, -0.2900,  0.0207,
         0.0412, -0.1137,  0.0159, -0.0329, -0.0880,  0.0384,  0.0050, -0.0337,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0291, -0.0117,  0.0415,  0.0209,  0.0076, -0.0129, -0.0091, -0.0144,
         0.2237,  0.1632,  0.0023,  0.1637, -0.0075,  0.0562,  0.1071,  0.0022,
        -0.0145,  0.0031, -0.0191, -0.0006, -0.0079, -0.0019,  0.0102,  0.0008,
        -0.0050,  0.0170,  0.0467,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0251, -0.0025, -0.0707, -0.0040, -0.0438,  0.0005, -0.0350, -0.0178,
        -0.0302, -0.0153,  0.0017, -0.0059, -0.0478, -0.0073, -0.0194, -0.0039,
        -0.0282,  0.0119, -0.0071, -0.0554,  0.0140, -0.0333, -0.0101, -0.0725,
        -0.0048,  0.0123, -0.0364, -0.0374,  0.0007, -0.0561,  0.0156, -0.0218,
        -0.0414, -0.0101,  0.0007, -0.0491,  0.0023, -0.0174,  0.0086,  0.0201,
        -0.0165, -0.0390, -0.0071, -0.0007,  0.0319, -0.0064], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0898,  0.0347,  0.0010, -0.0106,  0.0048,  0.0017,  0.0038, -0.0781,
        -0.0138, -0.0242, -0.0044, -0.0134,  0.0043,  0.0073, -0.0023,  0.0057,
        -0.0014, -0.0195, -0.1046, -0.2756, -0.0019, -0.0329, -0.0174,  0.0040,
        -0.0009, -0.0438, -0.1440, -0.0202, -0.0321,  0.0015,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0180, -0.0817, -0.0830, -0.0312,  0.0394,  0.0160,  0.1072,  0.4017,
        -0.1023, -0.0314,  0.0881,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0476, -0.3686, -0.0489, -0.0307, -0.0770, -0.0230, -0.0400, -0.1174,
         0.0134, -0.0343, -0.0071,  0.0065, -0.0158, -0.0220,  0.0013, -0.0598,
         0.0016, -0.0036, -0.0407,  0.0407,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0098, -0.4246, -0.1912, -0.0881,  0.0031, -0.0675, -0.1130, -0.0235,
        -0.0045, -0.0462,  0.0285,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0033, -0.2957, -0.1623, -0.1592,  0.0040, -0.0298, -0.0152, -0.0569,
        -0.0658, -0.0253, -0.0058, -0.0321, -0.0111,  0.0032, -0.0014, -0.0332,
         0.0106, -0.0101, -0.0254,  0.0354, -0.0142,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0794, -0.1983, -0.1336, -0.0232, -0.0325, -0.3398,  0.0198, -0.0117,
        -0.0156, -0.0064,  0.0153, -0.0711, -0.0287,  0.0009,  0.0047, -0.0024,
         0.0007,  0.0158,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0261, -0.1318, -0.0989,  0.0011,  0.0162,  0.0182,  0.0126,  0.0104,
        -0.0016,  0.0003,  0.0030,  0.0101, -0.0049, -0.0723, -0.0743,  0.0028,
        -0.0232, -0.0514, -0.0271,  0.0119, -0.0075, -0.0390, -0.0889, -0.0044,
        -0.0395,  0.0038, -0.0250, -0.0089, -0.0109, -0.0836, -0.0022, -0.0060,
        -0.0078,  0.0103, -0.0095,  0.0040,  0.0008,  0.0082,  0.0415,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0494, -0.2904, -0.0226, -0.0524, -0.0066,  0.0028,  0.0020,  0.0060,
         0.0009, -0.0264, -0.0200, -0.1001,  0.0022,  0.0104,  0.0257, -0.0016,
        -0.1140, -0.0024,  0.0050, -0.0157, -0.0153,  0.0008,  0.0052,  0.0046,
        -0.0036, -0.0040, -0.0095, -0.0019, -0.0358, -0.0857, -0.0098, -0.0048,
        -0.0290, -0.0334,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 6.5269e-06, -2.1815e-01, -8.3378e-02, -3.8781e-01,  7.1178e-02,
        -2.9057e-02,  8.3669e-03,  5.0146e-02, -1.5191e-01,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0178, -0.2429,  0.0127, -0.0417, -0.0312, -0.0837, -0.0007, -0.0285,
        -0.0366,  0.0008, -0.0277, -0.0088, -0.0039, -0.0107,  0.0091, -0.0103,
         0.0101, -0.1243, -0.0079,  0.0103, -0.0064, -0.0050, -0.0209,  0.0016,
        -0.0391, -0.0011, -0.0131,  0.0031, -0.0102,  0.0062,  0.0020, -0.0773,
        -0.0054, -0.0319,  0.0140,  0.0016,  0.0010,  0.0269,  0.0126, -0.0008,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0469, -0.0532, -0.1866, -0.0121, -0.0409, -0.0049,  0.0135, -0.0093,
        -0.0297, -0.0391,  0.0011, -0.0026,  0.0027, -0.0003,  0.0069, -0.0040,
         0.0047, -0.0234, -0.0457,  0.0119, -0.0111,  0.0215,  0.0044, -0.0095,
         0.0173, -0.0148, -0.0389,  0.0112, -0.0405, -0.0928, -0.0114,  0.0056,
        -0.0038,  0.0066,  0.0080, -0.0007, -0.0347, -0.0366, -0.0010, -0.0049,
         0.0131,  0.0120,  0.0106,  0.0045, -0.0025,  0.0077, -0.0059,  0.0006,
         0.0037,  0.0041,  0.0140,  0.0066,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0019, -0.2676, -0.0102, -0.0783, -0.0293,  0.0162, -0.0800, -0.0192,
        -0.0521, -0.0489,  0.0344,  0.0152, -0.0533, -0.0171, -0.0344, -0.0779,
        -0.0359, -0.0166, -0.0583,  0.0113, -0.0185,  0.0234,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.8735e-02,  5.0141e-01, -3.7909e-02, -1.5630e-02, -2.1247e-02,
        -1.0086e-02,  1.6115e-02,  3.2529e-02,  1.1725e-02,  2.0742e-05,
         2.3865e-02,  4.6510e-02,  2.2232e-02,  2.4820e-02, -6.4917e-02,
         7.6531e-02,  6.1473e-03,  1.6738e-02, -1.0842e-02,  7.8479e-04,
        -3.1210e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0361, -0.1611, -0.0481, -0.0188, -0.0027,  0.0068, -0.0146,  0.0021,
        -0.0321, -0.0051,  0.0014, -0.0022,  0.0054,  0.0154,  0.0042, -0.0061,
         0.0011, -0.0006, -0.0192,  0.0008,  0.0053,  0.0036, -0.0073,  0.0070,
        -0.0033, -0.0098, -0.0267, -0.0357, -0.0025, -0.0313, -0.0742, -0.0829,
         0.0045, -0.0332, -0.0238, -0.0028,  0.0024,  0.0007,  0.0177,  0.0075,
        -0.0733, -0.0116,  0.0057, -0.0049, -0.0320,  0.0018, -0.0328, -0.0069,
         0.0044,  0.0039, -0.0065, -0.0055,  0.0135, -0.0253,  0.0062],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.3057e-02,  3.4194e-04,  1.7105e-02, -9.7406e-03, -1.0920e-02,
        -2.5022e-02, -1.4554e-01, -2.7063e-02,  1.0221e-02,  9.3696e-03,
        -1.2660e-02,  1.0684e-02, -8.9375e-03, -2.6943e-02, -5.6490e-02,
        -2.2623e-02, -9.2291e-03, -1.0879e-02, -1.1262e-02, -7.4344e-02,
        -2.9204e-02, -2.6397e-02, -2.7667e-02,  5.1075e-03, -1.3839e-04,
         1.9882e-03, -1.5380e-02, -5.9844e-02, -3.0935e-02,  1.9728e-04,
         3.5068e-03,  4.1201e-03, -1.8844e-02, -7.2399e-02, -4.6642e-03,
        -1.1066e-02,  1.4051e-03, -2.2035e-02, -2.9878e-02, -2.5026e-02,
        -8.6892e-03, -2.9364e-03,  1.5667e-02,  8.4705e-03,  1.1998e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0293, -0.2317, -0.0854, -0.1113, -0.0163, -0.0155, -0.0218, -0.0115,
        -0.0276, -0.0071,  0.0133, -0.0252, -0.0054, -0.0872, -0.0590, -0.0070,
        -0.0094, -0.0521, -0.0137, -0.0125, -0.0885,  0.0185,  0.0225, -0.0283,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0368,  0.0169,  0.0101, -0.0107, -0.0086, -0.0121, -0.0656, -0.0620,
         0.0227,  0.0109, -0.0144, -0.0057, -0.0074, -0.0286, -0.0041,  0.0533,
        -0.0331, -0.0356, -0.0241, -0.1030,  0.0214, -0.0093, -0.0598, -0.0875,
         0.0103, -0.0383, -0.0752, -0.0113, -0.0179, -0.0539,  0.0113, -0.0121,
        -0.0261,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0328, -0.4298, -0.0069, -0.0906,  0.0057, -0.0165, -0.0229,  0.0035,
        -0.1503,  0.0049, -0.0042, -0.0072,  0.0307, -0.0605, -0.0045, -0.0254,
        -0.0258,  0.0433, -0.0345,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0401,  0.0095,  0.0641, -0.0130, -0.1087, -0.0282, -0.1033, -0.1116,
        -0.1188,  0.0056, -0.0115, -0.0235,  0.0107, -0.0077,  0.0027,  0.0116,
         0.0011, -0.0843, -0.0374, -0.0194, -0.0739, -0.0917, -0.0119,  0.0066,
         0.0032,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0216, -0.2236, -0.1544, -0.0115, -0.0123, -0.0803, -0.2361,  0.0133,
        -0.0066, -0.0299,  0.0142, -0.0005, -0.0704, -0.0054,  0.0120,  0.0367,
         0.0710,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0157,  0.0057, -0.0100,  0.0076,  0.0033, -0.0079,  0.0020,  0.0063,
         0.0095, -0.0685, -0.0225, -0.0084,  0.0220,  0.0010,  0.0044, -0.0208,
        -0.1163, -0.0971, -0.0142, -0.0624,  0.0114, -0.1578, -0.1581, -0.0039,
        -0.0572, -0.0054, -0.0050, -0.0049, -0.0144,  0.0018, -0.0148,  0.0595,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0162,  0.1226,  0.1554,  0.0124, -0.0027,  0.0068,  0.0607,  0.1582,
         0.0018, -0.0083, -0.0192, -0.0122, -0.0386, -0.0105,  0.0254, -0.0177,
         0.0093, -0.0044,  0.0457, -0.0089,  0.0226, -0.0075,  0.0553,  0.0857,
         0.0270,  0.0120,  0.0139, -0.0391,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1085, -0.0136, -0.0058, -0.0027,  0.0125, -0.0550, -0.1676, -0.0424,
        -0.0163, -0.0225,  0.0014, -0.1417, -0.0177,  0.0132, -0.0077, -0.1015,
        -0.0086, -0.0175,  0.0014, -0.0040, -0.0185,  0.0090, -0.0009, -0.0176,
        -0.0748,  0.0305,  0.0309,  0.0302, -0.0262,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0129, -0.0966, -0.0082,  0.0018, -0.0817, -0.0285,  0.0039,  0.0106,
        -0.0102,  0.0210,  0.0303, -0.0075, -0.0287, -0.1034,  0.0306, -0.0370,
         0.0274, -0.1095, -0.0122, -0.0494,  0.0011, -0.0009, -0.0480, -0.0056,
        -0.0403, -0.0682, -0.0359, -0.0115, -0.0216, -0.0076, -0.0095,  0.0203,
        -0.0182,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0081, -0.2663, -0.0155, -0.0834,  0.0140, -0.0547,  0.0278, -0.0422,
        -0.0132, -0.0684, -0.0522, -0.0141, -0.0105,  0.0011, -0.0107, -0.0005,
        -0.0628, -0.0222, -0.1068, -0.0018, -0.0707, -0.0121, -0.0294,  0.0114,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0492, -0.1726, -0.1878, -0.0060, -0.0672, -0.0060, -0.0276, -0.0830,
         0.0281, -0.0401, -0.0063, -0.0216, -0.0541, -0.0087,  0.0142,  0.0146,
        -0.0124, -0.0728,  0.0089, -0.0417, -0.0408,  0.0210,  0.0153,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.3763e-03, -1.7327e-01, -1.1095e-02, -2.5803e-02, -2.7411e-02,
        -3.5436e-02,  8.1604e-03,  7.8909e-04,  7.3997e-03,  9.6205e-03,
         1.2868e-02,  2.0853e-03, -5.5958e-03,  4.0302e-03,  1.2054e-04,
        -6.2124e-03, -7.2423e-02, -5.7474e-02, -5.0807e-02,  1.8546e-03,
        -1.0432e-02, -3.3097e-02, -6.8705e-03, -8.5686e-03, -7.2897e-02,
        -1.9433e-02, -1.5394e-02, -9.8774e-03, -1.1368e-02, -7.5492e-02,
        -5.1132e-02, -1.7354e-02, -4.9295e-02, -7.2495e-03, -1.9599e-02,
        -3.6269e-02, -1.0760e-02, -2.3456e-02,  7.6207e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0865, -0.0584, -0.0035, -0.0231, -0.0921, -0.0133,  0.0060,  0.0088,
         0.0109,  0.0144,  0.0096,  0.0042,  0.0075, -0.0066, -0.0056, -0.0420,
        -0.0176, -0.0138, -0.0046,  0.0019, -0.0515,  0.0077,  0.0028, -0.0117,
        -0.0864, -0.1228, -0.0140, -0.0449, -0.0156, -0.0647, -0.0199, -0.0117,
         0.0047,  0.0006, -0.0097, -0.0090, -0.0131, -0.0402,  0.0052,  0.0005,
        -0.0059, -0.0055, -0.0215,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.2319e-02, -1.9619e-01, -1.2455e-03, -7.9958e-03,  1.5412e-02,
        -1.1584e-03, -1.5063e-02,  3.2552e-03, -4.7298e-02,  1.5578e-03,
         8.1096e-03,  1.9906e-03, -1.9508e-02,  5.8277e-03, -2.3740e-03,
         3.3316e-03, -2.8461e-02, -4.9332e-02, -2.4841e-03, -9.8812e-03,
        -1.0449e-03, -7.2164e-03, -6.4785e-03, -7.7382e-03,  6.1151e-03,
         1.7170e-03,  9.0569e-03, -9.1415e-04,  5.3971e-03,  8.2191e-03,
         1.6649e-02, -2.6557e-02, -4.7535e-02,  2.4796e-03,  2.2346e-03,
        -7.8506e-03,  2.1763e-02, -5.8628e-02, -7.8799e-02, -5.2202e-02,
        -2.9487e-02, -2.2777e-02,  5.1942e-03, -3.0179e-03, -8.9696e-03,
        -6.6999e-03, -2.7925e-03, -4.4597e-03,  6.9812e-05, -2.6141e-03,
        -1.0890e-02, -1.3157e-03, -1.2578e-03,  5.2618e-03, -3.4536e-03,
        -3.5747e-02, -4.6027e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0468,  0.3980, -0.0273,  0.0124,  0.0462, -0.0260,  0.0577,  0.1286,
        -0.0195, -0.0167,  0.1070, -0.0908, -0.0149,  0.0081,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0116, -0.2103, -0.0074,  0.0158,  0.0016,  0.0003,  0.0013,  0.0031,
        -0.0438, -0.0128, -0.0459, -0.1127, -0.0447, -0.0158, -0.0702, -0.0275,
        -0.0279, -0.0635,  0.0107, -0.0633, -0.1020, -0.0180, -0.0252,  0.0030,
         0.0094,  0.0165,  0.0045,  0.0042, -0.0179, -0.0092,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0400,  0.1977, -0.1301,  0.4407,  0.1283,  0.0051,  0.0493, -0.0088,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0674, -0.0989,  0.0098,  0.0029, -0.0288, -0.1178, -0.0044,  0.0171,
         0.0018, -0.0087,  0.0132,  0.0077, -0.0801, -0.1388, -0.0117,  0.0136,
         0.0013, -0.0073, -0.0091, -0.0021, -0.0045,  0.0143,  0.0093,  0.0077,
        -0.0810, -0.0247, -0.0249, -0.0038, -0.0569, -0.0911, -0.0212, -0.0024,
        -0.0156,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.1766e-02, -1.4781e-01, -3.5212e-02, -2.1746e-02, -4.0383e-03,
        -2.1290e-02, -4.5300e-03, -2.6474e-03,  7.3768e-03, -2.5126e-02,
        -2.4910e-03, -7.9539e-03, -1.9255e-02, -5.1671e-03,  1.3972e-02,
         1.1050e-02,  8.0348e-03, -1.6303e-04, -6.2478e-03,  7.5378e-03,
        -3.1012e-03,  4.5152e-03, -2.1836e-03,  1.1856e-04,  1.5261e-02,
         9.1116e-04,  1.8350e-05, -4.2868e-02, -1.0545e-02, -1.3863e-02,
        -2.2821e-03, -4.0517e-02, -4.9594e-02, -7.0453e-02,  1.0364e-02,
        -1.6226e-04, -2.8284e-02, -6.3407e-03,  4.9850e-03,  7.9178e-03,
         1.3771e-03,  4.0980e-03, -4.0724e-03, -5.3573e-02, -1.7704e-02,
        -1.2249e-03, -3.7301e-03, -1.6728e-03, -5.1532e-03,  3.5300e-03,
        -2.0472e-02, -1.8707e-02, -1.0317e-02, -1.6226e-02, -4.2144e-03,
         8.0516e-03, -3.7803e-03,  1.3953e-03,  3.9598e-03, -1.8662e-02,
        -5.6561e-03, -6.9032e-03, -5.8446e-02,  4.5163e-05,  3.7663e-03,
         1.1584e-03,  2.1797e-03, -1.1099e-03,  3.5122e-03, -2.0061e-02,
         7.5446e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0486,  0.0219,  0.0084, -0.0024,  0.0263, -0.0038,  0.0982,  0.0083,
        -0.0052, -0.0234, -0.0174, -0.0024, -0.0060, -0.0078, -0.0160, -0.0015,
         0.0064, -0.0060,  0.0237,  0.0860,  0.0226, -0.0003, -0.0281,  0.0478,
        -0.0065,  0.0413,  0.0105, -0.0027,  0.0019,  0.0028,  0.0072,  0.0211,
         0.0769,  0.0672,  0.0372,  0.0158,  0.1093,  0.0416,  0.0029,  0.0043,
        -0.0163,  0.0161,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.8715e-02,  3.2826e-01,  5.9822e-04,  9.2894e-03,  5.6198e-03,
        -5.8294e-03, -1.7106e-02, -1.7683e-02,  1.6531e-03,  2.9544e-02,
         1.3812e-03,  4.1614e-02,  7.2566e-02, -1.4009e-02, -7.8243e-03,
         1.0174e-03, -6.1303e-03,  4.4495e-02,  8.4697e-02, -3.5015e-03,
         1.2554e-02,  1.8753e-02, -4.7691e-03, -9.4251e-03,  1.1881e-04,
         4.3685e-02,  1.4612e-02,  1.2804e-02,  7.0675e-02,  6.5174e-03,
        -1.8455e-02,  6.8780e-03, -1.2124e-02,  2.8111e-03, -4.4116e-03,
         1.7764e-02, -1.2108e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0070,  0.5016,  0.0620,  0.0586, -0.0019, -0.2001,  0.0037, -0.0017,
         0.0011,  0.0027,  0.0185, -0.0101,  0.0531, -0.0069,  0.0709,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0127, -0.0351,  0.0096, -0.0022, -0.1375, -0.2056, -0.0301, -0.0976,
        -0.1878, -0.0164,  0.0127, -0.0362, -0.0869,  0.0120,  0.0073, -0.0016,
         0.0115,  0.0543,  0.0146,  0.0284,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0349,  0.0193,  0.0064, -0.0052, -0.1048, -0.0073, -0.0020, -0.0024,
         0.0059,  0.0070, -0.0710, -0.0744, -0.0108, -0.0090, -0.0514, -0.0392,
         0.0036, -0.0326, -0.0005, -0.0739,  0.0019, -0.0013, -0.0104,  0.0012,
         0.0075, -0.0177, -0.0688, -0.0030,  0.0083, -0.0384, -0.0876, -0.0033,
         0.0105, -0.0051, -0.0254, -0.0828,  0.0033, -0.0190, -0.0056,  0.0012,
        -0.0095, -0.0107,  0.0163,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0126,  0.0048,  0.0035, -0.0040, -0.0513, -0.0392, -0.0567, -0.0591,
        -0.0422, -0.0071,  0.0281,  0.0012, -0.0069, -0.0678, -0.1139, -0.0157,
         0.0033, -0.0025, -0.0060, -0.0213, -0.0635, -0.0179,  0.0146,  0.0040,
        -0.0061, -0.0009, -0.0122, -0.0047, -0.0982,  0.0002, -0.0099, -0.0357,
        -0.0011, -0.0400, -0.0726, -0.0285,  0.0150,  0.0274,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0052, -0.2406, -0.0849, -0.1050,  0.0080, -0.0727, -0.0009,  0.0250,
        -0.0547, -0.0602, -0.0087, -0.0176, -0.0432, -0.0280,  0.0038, -0.0040,
        -0.0292, -0.0852, -0.0378,  0.0428,  0.0155,  0.0268,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0746, -0.1286, -0.0284, -0.0032,  0.0089,  0.0100,  0.0097,  0.0023,
        -0.0073, -0.1267, -0.0202, -0.0217, -0.0161,  0.0016, -0.0306, -0.0492,
        -0.1199, -0.0063, -0.0015, -0.0607, -0.0114, -0.0041, -0.0476, -0.0294,
        -0.0098, -0.0041,  0.0008,  0.0037, -0.0122, -0.0300,  0.0020, -0.0057,
         0.0195,  0.0067, -0.0160, -0.0543,  0.0091,  0.0004,  0.0019,  0.0039,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0038, -0.0206,  0.0107,  0.0225, -0.0034, -0.0672, -0.0174, -0.0104,
        -0.0462, -0.0146, -0.0070, -0.0134, -0.0018, -0.1266, -0.0532, -0.0460,
        -0.0100,  0.0036, -0.1108, -0.0071, -0.0083, -0.0005,  0.0177,  0.0019,
        -0.0155, -0.1306, -0.0218, -0.0720, -0.0980, -0.0191,  0.0082, -0.0102,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0466, -0.2958, -0.0273,  0.0179, -0.0231, -0.0162, -0.0258, -0.0528,
        -0.0677,  0.0023,  0.0078, -0.0482,  0.0229, -0.0143,  0.0076, -0.0063,
        -0.0434, -0.0006,  0.0037,  0.0115,  0.0038, -0.0133, -0.0135, -0.0903,
        -0.0113, -0.0020, -0.0515, -0.0126, -0.0092, -0.0097, -0.0052, -0.0004,
         0.0004,  0.0034,  0.0317,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.1276, -0.0297, -0.1334, -0.1065, -0.0236,  0.0182, -0.1609, -0.1108,
        -0.0187, -0.0688,  0.0066, -0.0229,  0.0044, -0.0307, -0.0122, -0.0070,
        -0.0183, -0.0138,  0.0148,  0.0146,  0.0567,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0253, -0.4204, -0.0281, -0.0876, -0.0918, -0.0268, -0.0023, -0.0320,
        -0.0657,  0.0278, -0.0058,  0.0228,  0.0117,  0.0094, -0.0391,  0.0142,
        -0.0338, -0.0371,  0.0183,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0611,  0.0044, -0.1717,  0.0082, -0.0565, -0.0290,  0.0049,  0.0403,
         0.0014, -0.0295, -0.1221, -0.0324, -0.0519, -0.1349, -0.0040, -0.0801,
        -0.0474, -0.0442,  0.0144, -0.0614,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0720, -0.4000, -0.0298, -0.0837,  0.0285,  0.0071, -0.0092, -0.0374,
        -0.0190, -0.0117, -0.0074, -0.0119, -0.0076, -0.0703, -0.0086, -0.0243,
        -0.0320,  0.0030, -0.0291, -0.0605, -0.0094,  0.0059, -0.0313,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0095, -0.2209, -0.0642, -0.0239, -0.0044, -0.0344, -0.0273, -0.0167,
        -0.0849, -0.1225, -0.0270, -0.0693, -0.0137, -0.0676,  0.0224, -0.0061,
         0.0076,  0.0016,  0.0046, -0.0052, -0.0378, -0.0092,  0.0277, -0.0296,
         0.0619,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0425, -0.3003, -0.0246,  0.0042, -0.0011, -0.0007,  0.0190, -0.0026,
         0.0213, -0.0332, -0.1060, -0.0041, -0.0188,  0.0140, -0.0069, -0.0348,
        -0.0079, -0.0031, -0.0064, -0.0022, -0.0753, -0.0283, -0.0424, -0.0393,
         0.0053, -0.0148, -0.0150,  0.0116, -0.0251, -0.0195, -0.0155, -0.0008,
        -0.0233,  0.0121, -0.0180,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0133, -0.0130, -0.0132, -0.0970, -0.0135, -0.0154, -0.1261, -0.0227,
        -0.1438, -0.0260, -0.0346, -0.0323, -0.0255, -0.0718, -0.0463, -0.1298,
         0.0014, -0.0807,  0.0244,  0.0071, -0.0207, -0.0415,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0182, -0.0508, -0.0210, -0.0125,  0.0701,  0.0096,  0.0007,  0.0407,
         0.0232,  0.2543, -0.0358,  0.0033,  0.0397,  0.0688,  0.0085,  0.0337,
         0.0291,  0.0628, -0.1031,  0.0566, -0.0576,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0106,  0.4132,  0.0868, -0.0084,  0.0279,  0.0043,  0.0761,  0.0925,
         0.0020,  0.0009,  0.0012, -0.0169,  0.0127,  0.0890, -0.0278,  0.0308,
        -0.0076, -0.0742,  0.0171,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0482,  0.0103, -0.0205,  0.0141,  0.0133, -0.0779, -0.0097,  0.0022,
         0.0105, -0.0047, -0.0114, -0.0272, -0.0017, -0.0372, -0.0030, -0.0019,
        -0.0154, -0.0512,  0.0015, -0.0225,  0.0002,  0.0194, -0.0130,  0.0182,
        -0.0184, -0.0185, -0.0062, -0.0216, -0.0101, -0.0882, -0.0044, -0.0186,
        -0.0614, -0.1492, -0.0059, -0.0184, -0.0286,  0.0159, -0.0051, -0.0309,
         0.0118,  0.0079,  0.0008, -0.0016,  0.0071,  0.0191, -0.0155],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1143, -0.2387, -0.1730, -0.0142,  0.0086,  0.0136,  0.0026, -0.0147,
        -0.0418,  0.0033,  0.0037,  0.0047,  0.0129,  0.0132,  0.0313,  0.0021,
        -0.0464,  0.0015,  0.0003, -0.0169,  0.0171,  0.0087, -0.0383, -0.0732,
         0.0024,  0.0140,  0.0146,  0.0034,  0.0139,  0.0101,  0.0144,  0.0222,
        -0.0042,  0.0058,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.5518e-02,  3.1707e-01, -5.4801e-03,  1.2276e-03,  3.9773e-02,
         3.4880e-02, -3.8508e-03,  7.6911e-03,  2.0437e-03,  2.2894e-02,
        -4.8265e-05, -1.5590e-02,  5.6708e-02,  8.5089e-02,  2.7475e-02,
        -4.4108e-03,  2.6965e-02,  9.3340e-04, -5.9690e-03,  4.5844e-02,
         1.6515e-02,  4.6168e-02,  1.0072e-02,  1.3257e-02,  1.1630e-02,
        -1.1746e-02,  1.2082e-02,  2.4354e-02,  1.3898e-02,  4.0037e-02,
         6.0447e-03, -3.1419e-03,  1.1175e-02, -5.0420e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0510, -0.3920, -0.0035, -0.0090, -0.0502, -0.1028, -0.0047, -0.0071,
        -0.0011, -0.0887, -0.0201, -0.0211, -0.0351, -0.0987, -0.0376, -0.0202,
        -0.0135, -0.0436,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.2902e-02,  2.4417e-02, -2.7476e-02,  1.4977e-02,  8.0688e-03,
         1.4253e-02,  3.4422e-04, -2.9858e-02, -1.3929e-01, -2.2098e-01,
        -1.0561e-02, -1.4803e-02,  6.5273e-03, -1.7866e-02,  2.8588e-02,
        -7.8897e-03, -2.2151e-02, -8.8469e-04,  1.8348e-02, -2.5026e-01,
        -9.2583e-03,  5.6323e-03, -2.3032e-02,  4.0655e-03, -1.6267e-04,
        -5.8198e-04, -5.6817e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0373,  0.0121,  0.0039, -0.0027, -0.0483,  0.0036, -0.0151, -0.0078,
        -0.1523, -0.0141, -0.0143, -0.0320, -0.1146, -0.0406,  0.0019, -0.0093,
         0.0181, -0.0182,  0.0180, -0.2370, -0.0004,  0.0046, -0.0838, -0.0245,
         0.0401, -0.0453,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0962, -0.2588, -0.0289, -0.1053, -0.0876, -0.0258, -0.0190,  0.0067,
        -0.0600, -0.0404, -0.0452, -0.0389,  0.0070, -0.0012, -0.0326, -0.0589,
        -0.0295, -0.0207, -0.0073, -0.0240, -0.0061,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0403, -0.2337, -0.1708, -0.1400, -0.0510, -0.0163, -0.1581, -0.0047,
        -0.0229,  0.0081, -0.0211, -0.0309,  0.0510, -0.0283,  0.0227,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0987, -0.0119,  0.0280,  0.0170, -0.1246, -0.0044,  0.0489,  0.0052,
        -0.0027,  0.0363,  0.0196, -0.0627, -0.2373, -0.0914,  0.0406,  0.0257,
         0.0175, -0.1277,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0433,  0.3947,  0.0644,  0.0821, -0.0085,  0.0128,  0.0298,  0.0327,
         0.0048,  0.0401,  0.0654,  0.0520, -0.0033,  0.0828,  0.0020,  0.0086,
         0.0027, -0.0089, -0.0609,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0399, -0.1351, -0.1034, -0.0228, -0.0358, -0.0279,  0.0117, -0.0093,
        -0.0146, -0.0050, -0.0015,  0.0086, -0.0210, -0.0006,  0.0006, -0.0115,
        -0.0580,  0.0045,  0.0074,  0.0134, -0.0811, -0.0006, -0.0516, -0.0011,
        -0.0589, -0.0092, -0.0773,  0.0168, -0.0204, -0.0230, -0.0024, -0.0027,
        -0.0182, -0.0018, -0.0254, -0.0029, -0.0394, -0.0036, -0.0114,  0.0034,
         0.0002, -0.0011,  0.0017, -0.0130], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0181,  0.2400,  0.0321,  0.1163,  0.0627,  0.0245,  0.0174,  0.0124,
         0.0733,  0.0068,  0.0199,  0.0131, -0.0159, -0.0008, -0.0173, -0.0093,
        -0.0343,  0.0821, -0.0056,  0.0083,  0.0223, -0.0091,  0.0006,  0.0044,
        -0.0027,  0.0034,  0.0017,  0.0258, -0.0022,  0.0016,  0.0076,  0.0130,
        -0.0194, -0.0084,  0.0675,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0520, -0.3266, -0.0266, -0.0854, -0.0278, -0.0094, -0.1215, -0.0621,
        -0.0033, -0.0138,  0.0070, -0.0177, -0.0492,  0.0006, -0.0061, -0.0310,
        -0.0104, -0.0063,  0.0180,  0.0005, -0.0017, -0.0007, -0.0238,  0.0983,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0365,  0.2752,  0.0092,  0.0333,  0.0198,  0.0119, -0.0179,  0.0075,
         0.0728,  0.0093,  0.0278, -0.0065, -0.0073,  0.0333,  0.0701,  0.0164,
        -0.0215,  0.0048,  0.0511,  0.0111,  0.0067,  0.0268, -0.0077,  0.0214,
        -0.0005,  0.0274,  0.0279,  0.0319,  0.0051,  0.0206,  0.0213,  0.0008,
        -0.0553,  0.0034,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0224,  0.0173, -0.0086, -0.0215, -0.0159, -0.0255, -0.1661, -0.0301,
        -0.0068, -0.0079, -0.0025, -0.0015, -0.0884, -0.0128, -0.1133, -0.1616,
        -0.0083,  0.0144,  0.0052,  0.0328,  0.0174,  0.0002,  0.0340, -0.0605,
        -0.0442, -0.0016, -0.0391, -0.0045,  0.0310, -0.0046,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.1500, -0.0119, -0.0374, -0.3716, -0.0148,  0.0247, -0.0339, -0.0268,
        -0.0425, -0.2123,  0.0434, -0.0094,  0.0213,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0351,  0.0060,  0.0293, -0.0112,  0.0104, -0.0286,  0.0014, -0.0933,
        -0.1972,  0.0370, -0.0303, -0.1478, -0.0042, -0.1222, -0.1533, -0.0296,
         0.0440, -0.0191,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1114, -0.2303, -0.3100,  0.0752, -0.1250, -0.0403, -0.0474, -0.0604,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0033, -0.1328,  0.0116, -0.0377, -0.0135, -0.0209, -0.0210, -0.0223,
         0.0095, -0.0041, -0.0020,  0.0100, -0.0073, -0.0106,  0.0055, -0.0070,
         0.0018,  0.0035, -0.0054, -0.0067,  0.0062,  0.0138, -0.0333, -0.0337,
         0.0006, -0.0329, -0.0102, -0.0041, -0.0881, -0.0481,  0.0055, -0.0260,
        -0.0149, -0.1385,  0.0003, -0.0456, -0.0309,  0.0087,  0.0187, -0.0028,
        -0.0114, -0.0027, -0.0184,  0.0075,  0.0086,  0.0292,  0.0227],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0415,  0.1521,  0.1609,  0.3494, -0.0045, -0.0615,  0.0429, -0.0068,
         0.0762, -0.0085, -0.0956,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0331, -0.3770, -0.0605, -0.1154, -0.0136, -0.0084, -0.0353, -0.1094,
        -0.0128, -0.0388, -0.0369,  0.0089, -0.0680, -0.0043,  0.0164, -0.0076,
         0.0204, -0.0332,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0620, -0.0635, -0.4010,  0.0057,  0.0217,  0.0204, -0.2109,  0.0114,
        -0.0775,  0.0069, -0.0009,  0.0272,  0.0370,  0.0318, -0.0220,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0702, -0.4185, -0.0276, -0.0188, -0.0229, -0.0557, -0.1070, -0.0582,
        -0.0099, -0.0966, -0.0736, -0.0056, -0.0032,  0.0074, -0.0247,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0925,  0.0158,  0.0157,  0.0284, -0.0087, -0.0166, -0.0133, -0.0010,
        -0.0896, -0.0512, -0.0251, -0.0721, -0.1497, -0.0318,  0.0206, -0.0022,
        -0.0478, -0.0061,  0.0070,  0.0132, -0.0794, -0.1430,  0.0028,  0.0321,
        -0.0241,  0.0100,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0204, -0.0250, -0.0301, -0.0108,  0.0239, -0.0141, -0.0343, -0.2699,
        -0.2894, -0.0344,  0.0261,  0.0278,  0.0419,  0.0405, -0.1113,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0067, -0.3491, -0.0309, -0.0729, -0.0910, -0.0043, -0.0406, -0.0111,
         0.0020, -0.0819,  0.0139, -0.0090, -0.0410, -0.0057, -0.0121, -0.0018,
        -0.0164, -0.0030, -0.0557, -0.0039, -0.0449,  0.0205,  0.0014,  0.0121,
         0.0125,  0.0557,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0207,  0.0185,  0.0103,  0.0273,  0.0020, -0.0126, -0.0006,  0.0100,
        -0.0398, -0.0462, -0.1427, -0.0042, -0.0104, -0.0132, -0.1025, -0.0643,
        -0.0112, -0.0364, -0.0044, -0.1312, -0.0618, -0.0272, -0.0253, -0.0798,
        -0.0694, -0.0280,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-5.9806e-02, -2.3348e-01,  1.7123e-02, -5.0602e-03, -1.0639e-02,
         1.6537e-02, -3.6945e-02, -7.6648e-02, -3.2838e-03, -3.4513e-02,
        -2.1690e-02, -3.8737e-02, -7.2464e-02,  8.1032e-03, -2.3746e-02,
        -1.8432e-02, -1.1112e-02, -6.9924e-03, -1.5813e-02, -4.6727e-03,
         2.5998e-02, -2.2192e-02,  1.4956e-03, -3.9691e-02, -7.0534e-02,
         2.3290e-04, -3.4156e-02, -9.6663e-03,  3.1606e-03, -1.8378e-03,
        -1.0170e-03, -1.7202e-02, -7.0310e-03,  8.0035e-03,  5.2666e-03,
        -3.6712e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.8952e-03, -1.3032e-01, -3.6775e-02,  5.7285e-03, -4.7463e-02,
        -1.8366e-02, -4.2871e-02, -6.5925e-02,  1.3784e-03, -7.3987e-05,
        -4.5116e-02, -1.5665e-03, -1.7285e-03,  1.0198e-02, -2.1669e-03,
        -1.5438e-02,  3.8393e-03, -7.9395e-03,  6.5596e-03,  4.7851e-03,
        -1.4395e-02, -5.6656e-02, -7.9974e-03, -2.0925e-02, -6.0469e-02,
        -8.2211e-02, -3.4263e-03, -6.7058e-02, -5.7304e-02,  2.2297e-02,
         5.5222e-03, -5.4313e-02, -3.9757e-02,  8.4209e-03, -3.8707e-02,
         4.4055e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0183, -0.0367, -0.0169, -0.0222,  0.0368, -0.1355, -0.0524, -0.0206,
        -0.0221,  0.0222, -0.0333, -0.1147, -0.2664, -0.0446,  0.0447, -0.0383,
        -0.0259,  0.0094,  0.0023,  0.0366,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.0877e-02, -1.8239e-01, -6.4466e-02, -5.7041e-02,  2.9000e-03,
        -6.7135e-03, -3.4594e-03, -3.6602e-02, -2.5362e-03, -2.5221e-02,
         5.6825e-03,  1.5664e-03, -1.7356e-02, -5.1625e-05, -4.0916e-02,
        -2.0010e-02, -1.5893e-02, -4.7232e-02, -4.4757e-03, -2.4117e-03,
        -5.1782e-03, -6.2632e-03,  6.4369e-03,  2.4131e-02, -5.6211e-03,
         8.8447e-03,  2.7867e-03,  1.4669e-02, -3.1876e-03, -5.7853e-02,
        -2.8946e-02, -6.1980e-02,  3.1576e-02, -8.9064e-02,  3.7264e-02,
         4.6327e-02,  7.4617e-04,  1.3281e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0494, -0.1318, -0.0576, -0.0348, -0.0098,  0.0171, -0.0097, -0.0086,
        -0.0028, -0.0081, -0.0670, -0.1093,  0.0020,  0.0079, -0.0111, -0.0812,
         0.0037, -0.0628, -0.0063,  0.0060,  0.0007,  0.0045,  0.0093,  0.0003,
        -0.0010, -0.0081, -0.0126, -0.0337, -0.0736,  0.0075, -0.0092, -0.0159,
         0.0053, -0.0021, -0.0264, -0.0029, -0.0095,  0.0229,  0.0164,  0.0099,
        -0.0409], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.6306e-02, -4.9988e-01, -3.2670e-02,  4.7860e-02, -8.3927e-03,
        -3.6131e-02,  1.0666e-02,  4.7538e-03, -1.6413e-02, -3.4423e-03,
        -3.2424e-04, -4.9480e-02, -6.7378e-02,  2.7356e-03,  1.6182e-02,
        -5.1185e-02,  1.6739e-03, -2.9058e-02, -7.7901e-03,  3.1087e-03,
         3.6177e-03, -4.0954e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0637,  0.0102,  0.0007, -0.0036,  0.0194, -0.0159, -0.0030,  0.0140,
        -0.0145,  0.0022,  0.0215, -0.0747,  0.0017, -0.0306, -0.0162, -0.1211,
        -0.0036, -0.0584, -0.0417, -0.0853, -0.0781,  0.0025,  0.0041, -0.0590,
         0.0039, -0.0017,  0.0091, -0.0332, -0.0959, -0.0133, -0.0146, -0.0548,
        -0.0062,  0.0153,  0.0063,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1154, -0.0428, -0.1413, -0.0235, -0.1018,  0.0058,  0.0054,  0.0187,
        -0.0037,  0.0058, -0.0639,  0.0158,  0.0016, -0.0045, -0.0157, -0.0440,
         0.0420, -0.0103, -0.0642, -0.0025, -0.0560,  0.0151, -0.0179, -0.0540,
        -0.0242, -0.0209, -0.0700,  0.0026,  0.0104,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0634, -0.3177, -0.1258, -0.0623, -0.0458, -0.0515, -0.0220, -0.0080,
        -0.0202, -0.0140, -0.0199, -0.0190,  0.0016,  0.0098, -0.0159, -0.0596,
        -0.0279, -0.0100, -0.0211, -0.0059, -0.0071, -0.0315, -0.0054,  0.0206,
        -0.0139,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.6452e-02,  3.7692e-03, -2.1771e-02, -1.8748e-02, -1.9132e-01,
        -1.9130e-03, -2.4466e-02, -1.6566e-01, -1.6741e-01, -6.5615e-05,
        -1.6492e-02,  3.1691e-02, -2.4792e-02,  3.1973e-02, -9.6468e-02,
        -8.3388e-02,  1.0824e-02, -3.6103e-02,  6.3646e-05,  1.6706e-02,
        -2.0215e-02, -1.7078e-03,  1.8010e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0736, -0.0250,  0.0011, -0.0065,  0.0018,  0.0233, -0.0053, -0.0097,
         0.0034, -0.0068,  0.0032, -0.0161, -0.0683, -0.0632, -0.0136, -0.0137,
        -0.0007, -0.0350,  0.0051, -0.0040,  0.0093, -0.0258, -0.0109, -0.0160,
        -0.0392,  0.0042, -0.1419,  0.0044, -0.0847, -0.0110, -0.0787, -0.0185,
        -0.0089, -0.0201, -0.0364, -0.0248, -0.0275, -0.0002, -0.0583,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0869, -0.0138, -0.0129, -0.1791, -0.1566, -0.0130, -0.0350, -0.1267,
        -0.2183,  0.1060,  0.0500,  0.0018,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0290, -0.0795, -0.2964,  0.0094, -0.0184, -0.0323, -0.0373,  0.0107,
        -0.0028, -0.0356,  0.0064, -0.0080, -0.0089,  0.0023, -0.0269, -0.0024,
        -0.0266, -0.0564,  0.0014, -0.0020, -0.0259, -0.0298,  0.0147,  0.0073,
         0.0007,  0.0071,  0.0102, -0.0142, -0.0235, -0.0071,  0.0074,  0.0072,
        -0.0522, -0.0040,  0.0003,  0.0085, -0.0481, -0.0059, -0.0046,  0.0007,
         0.0067, -0.0046,  0.0140, -0.0008,  0.0017,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0215, -0.4097, -0.0810, -0.1545, -0.0424,  0.0017, -0.0185, -0.0463,
        -0.1095, -0.0177, -0.0420,  0.0552,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0673, -0.5538, -0.0740, -0.1190,  0.0326,  0.0670, -0.0610, -0.0253,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.2290e-03,  2.5547e-01,  1.4942e-02,  1.7887e-03,  4.7884e-02,
        -5.6782e-03,  1.9628e-02,  1.1163e-02,  2.9998e-02,  7.4310e-02,
         1.9118e-02,  1.0659e-02,  1.8330e-03,  1.2870e-03, -8.6043e-04,
        -6.9148e-03, -3.1667e-03, -4.8034e-03, -1.7777e-03,  3.1685e-03,
        -8.6439e-03, -7.1103e-03, -1.1610e-03,  5.0400e-03,  2.3673e-03,
        -2.8663e-03,  7.3060e-04,  1.7669e-05,  3.1305e-02, -2.3668e-03,
         2.9542e-03,  7.7217e-03, -2.5648e-03, -6.8361e-03, -4.6343e-03,
        -2.6309e-03, -3.3404e-03, -5.8557e-04, -9.0823e-03,  9.4776e-04,
        -6.8426e-04, -1.9703e-03,  2.3391e-03,  1.0504e-02, -2.6789e-04,
        -2.2867e-02,  7.2345e-02, -2.2291e-02,  5.3215e-02,  8.9390e-02,
         1.5197e-02,  3.0761e-04,  2.6988e-02,  1.1270e-02,  2.2490e-02,
         1.1838e-03,  1.5047e-02,  4.1643e-03, -1.8408e-03, -2.8128e-03,
        -1.2412e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0373, -0.0129,  0.0241, -0.0362, -0.1059,  0.0029, -0.0320,  0.0080,
        -0.0081,  0.0039,  0.0043, -0.0837,  0.0031, -0.0615, -0.0947,  0.0119,
         0.0008, -0.0039, -0.0413, -0.0122, -0.0671, -0.0733, -0.0372, -0.0199,
        -0.0603, -0.0213, -0.0048, -0.0287, -0.0238, -0.0155,  0.0087, -0.0095,
        -0.0412,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0218, -0.0682, -0.1496, -0.0569, -0.0166,  0.0198, -0.0383,  0.0129,
         0.0060, -0.0070, -0.0074, -0.0009, -0.0219, -0.0096, -0.0021,  0.1271,
        -0.0034, -0.0906, -0.2379,  0.0178, -0.0061, -0.0150,  0.0283, -0.0349,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0287, -0.1282, -0.1267, -0.0819, -0.0142, -0.1131, -0.0016, -0.0636,
        -0.0035, -0.0514, -0.0863, -0.0610, -0.1107, -0.0032, -0.0198,  0.0068,
        -0.0047,  0.0042, -0.0058,  0.0062, -0.0341, -0.0205, -0.0033, -0.0204,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0408, -0.0042, -0.0393, -0.0861, -0.0122,  0.0111, -0.0303, -0.0476,
         0.0022, -0.0142,  0.0034, -0.0067,  0.0113, -0.0025, -0.0604, -0.1072,
        -0.1028,  0.0163, -0.0111, -0.0315, -0.0586,  0.0188, -0.0212, -0.0806,
         0.0079, -0.0081, -0.0494, -0.0050,  0.0067, -0.0246, -0.0191, -0.0241,
         0.0042,  0.0305,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1171,  0.0253, -0.0140,  0.0170, -0.1299, -0.0024,  0.0094,  0.0072,
         0.0036,  0.0132, -0.0107,  0.0016, -0.0026, -0.1366,  0.0339,  0.0109,
        -0.0806,  0.0159, -0.0452,  0.0081, -0.0600, -0.1345,  0.0787, -0.0182,
         0.0234,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0191,  0.0155, -0.0157, -0.0167, -0.0007, -0.0143, -0.1522, -0.0803,
        -0.1733, -0.2134,  0.0568, -0.0248, -0.0227,  0.0591, -0.0839,  0.0513,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0214, -0.2158,  0.0269, -0.1042, -0.0497, -0.0754, -0.1565, -0.0166,
         0.0341,  0.0070,  0.0206, -0.0189, -0.0433, -0.0147,  0.0196,  0.0064,
        -0.0177, -0.0376, -0.0230, -0.0108, -0.0063,  0.0315,  0.0420,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0106, -0.0143,  0.0251,  0.1759,  0.3477, -0.0570, -0.0525, -0.0221,
         0.0073, -0.0599, -0.0412, -0.0362, -0.0634,  0.0869,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0130, -0.0721, -0.0734, -0.1511, -0.0234, -0.0072, -0.0356,  0.0301,
        -0.0445, -0.0435, -0.0649, -0.1039, -0.0424,  0.0036, -0.0033,  0.0162,
        -0.0317, -0.0940, -0.0170, -0.0736, -0.0147, -0.0035,  0.0145,  0.0229,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0460, -0.0489, -0.1905, -0.0237, -0.1868,  0.0176, -0.3275,  0.0200,
         0.0028,  0.0750, -0.0612,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0370,  0.0264,  0.0397, -0.0089, -0.0029, -0.1020, -0.1857,  0.0089,
        -0.1433, -0.0457, -0.0499,  0.0152, -0.0711, -0.0181, -0.0753,  0.0006,
        -0.0278, -0.0877,  0.0116,  0.0299, -0.0124,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0026,  0.0039,  0.0173,  0.0883,  0.0375,  0.0264,  0.0286, -0.3069,
        -0.1748, -0.1657, -0.0322,  0.0235, -0.0712, -0.0213,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0399, -0.0406, -0.0079, -0.0011, -0.0071,  0.0176, -0.0150, -0.0329,
        -0.0234,  0.0094,  0.1016,  0.1073,  0.3306, -0.0448,  0.0313,  0.0650,
        -0.0299,  0.0388, -0.0372,  0.0100,  0.0083,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0536, -0.3438, -0.0142, -0.1348,  0.0131, -0.0952, -0.0351, -0.1599,
        -0.0469, -0.0654,  0.0010, -0.0114,  0.0257,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0151, -0.1113, -0.0128, -0.0059, -0.0128, -0.0059, -0.0335, -0.0633,
        -0.0172, -0.0038, -0.0014, -0.0058,  0.0129, -0.0023,  0.0039, -0.0135,
         0.0036, -0.0063,  0.0019,  0.0021, -0.0091,  0.0147, -0.0171, -0.0178,
        -0.1494,  0.0229, -0.0818, -0.0656, -0.0754, -0.0100, -0.0284, -0.0023,
        -0.0071, -0.0031, -0.0007,  0.0046, -0.0661, -0.0084,  0.0008, -0.0062,
        -0.0208,  0.0045,  0.0041, -0.0024, -0.0416], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0299,  0.0305,  0.0813, -0.0021,  0.0043, -0.0007, -0.0010,  0.0146,
        -0.0023, -0.0022,  0.0864, -0.0017,  0.0152, -0.0046,  0.0456, -0.0012,
         0.1058, -0.0094,  0.0060,  0.0046,  0.0147,  0.0468, -0.0148,  0.0925,
         0.0027, -0.0046,  0.0498,  0.0490,  0.1745,  0.0358,  0.0122,  0.0079,
         0.0188,  0.0263,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0779, -0.0260, -0.0428, -0.0216, -0.0161, -0.0129,  0.0143,  0.0056,
         0.0121,  0.0096, -0.0233, -0.0132, -0.0040, -0.0079,  0.0068,  0.0168,
        -0.0008, -0.0227, -0.0189, -0.2413, -0.3672,  0.0131, -0.0065,  0.0079,
         0.0071, -0.0036,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.6959e-02, -4.4716e-01, -9.2164e-03, -6.9586e-02,  2.1331e-02,
        -1.1087e-02,  1.4910e-02, -4.2940e-02, -1.3277e-04,  1.2265e-02,
         2.0316e-02,  4.6391e-03,  1.6889e-02, -1.0507e-02, -1.4006e-02,
        -6.8312e-02, -8.7659e-02,  3.0042e-02, -4.1380e-02, -3.9268e-03,
        -2.3472e-02,  8.7713e-03,  1.4487e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0468, -0.0914, -0.0252, -0.0267,  0.0080,  0.0154,  0.0025, -0.0429,
        -0.0837, -0.0091,  0.0264,  0.0007,  0.0087,  0.0065, -0.1498, -0.0005,
        -0.0082, -0.0485, -0.0916, -0.0298, -0.0211, -0.0042,  0.0039,  0.0266,
        -0.0008, -0.0069, -0.0157, -0.0592,  0.0181,  0.0052, -0.0104,  0.0062,
        -0.0051,  0.0356, -0.0099,  0.0066, -0.0245,  0.0041,  0.0055,  0.0079,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0133,  0.0033, -0.0046, -0.0475, -0.0719,  0.0021, -0.0050, -0.0563,
         0.0182, -0.0112, -0.0052, -0.0102, -0.0300, -0.0067, -0.0496, -0.0489,
        -0.0479, -0.0126, -0.0440, -0.0400, -0.0029, -0.0772, -0.0492,  0.0084,
        -0.0152, -0.0082, -0.0130, -0.0828, -0.0092, -0.0559, -0.1078, -0.0191,
        -0.0134, -0.0091,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0591, -0.0520, -0.0272, -0.0086, -0.0629, -0.0296,  0.0058,  0.0140,
         0.0029,  0.0040, -0.0059, -0.0164,  0.0253, -0.0095, -0.1018,  0.0054,
        -0.0117, -0.0054, -0.0556, -0.0007,  0.0013, -0.0017,  0.0011,  0.0018,
        -0.0346, -0.0707,  0.0220,  0.0001,  0.0155, -0.0046, -0.0200,  0.0025,
         0.0054,  0.0030, -0.0513,  0.0080, -0.0173, -0.0314, -0.0024, -0.0299,
        -0.0391, -0.0142, -0.0484, -0.0190,  0.0008, -0.0001, -0.0008, -0.0025,
        -0.0112, -0.0070, -0.0216, -0.0065], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0204, -0.0093,  0.0106, -0.0212,  0.1991,  0.0352,  0.0056, -0.0452,
        -0.0043,  0.0135, -0.0121, -0.0091,  0.0116,  0.1191,  0.0380,  0.0393,
         0.1319,  0.0251,  0.0406, -0.0040,  0.0242,  0.0080,  0.0243,  0.1019,
        -0.0004,  0.0027,  0.0117,  0.0071, -0.0077,  0.0096, -0.0073,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.6584e-02, -1.7825e-01,  5.3984e-03, -3.0129e-02,  1.0332e-02,
         3.2164e-02, -1.3556e-02, -1.3882e-02,  6.0746e-03,  3.8322e-02,
         4.0607e-03, -8.5403e-02,  5.5815e-05, -1.0075e-02, -7.8327e-02,
        -5.5797e-02, -1.4153e-02, -7.7071e-02, -1.3630e-01, -1.1108e-01,
        -2.4677e-02, -3.2983e-03, -1.3912e-02, -4.2841e-03, -2.6815e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0162,  0.0682,  0.0055,  0.0144, -0.0396, -0.1362, -0.4262, -0.0562,
        -0.0045, -0.0749,  0.0654, -0.0927,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.8672e-03, -3.3949e-01, -3.2883e-02, -3.3451e-02, -6.0367e-02,
        -1.9853e-02, -7.1159e-03, -8.8121e-03, -9.6575e-05, -7.3423e-03,
         1.3614e-02, -1.9397e-02, -6.5499e-02, -1.0103e-01,  7.3078e-03,
        -4.4841e-02,  5.4825e-05, -6.4314e-03, -1.5941e-02,  2.6177e-03,
        -2.8786e-02, -1.4331e-02, -2.9102e-02, -2.9872e-02, -4.3836e-02,
        -2.9156e-02,  1.5529e-03,  5.5956e-03,  2.5758e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1241,  0.2576, -0.0019,  0.0188, -0.0150,  0.0460,  0.2025,  0.0170,
         0.0017, -0.0412, -0.0291,  0.2451,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1131, -0.3579,  0.0512, -0.0963,  0.0055, -0.0824, -0.1407, -0.0298,
        -0.0634, -0.0434,  0.0056, -0.0108,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0219,  0.0532, -0.1297, -0.0066, -0.1702,  0.0071,  0.0166, -0.0150,
        -0.1579, -0.0583,  0.0031, -0.1414, -0.0308, -0.0092,  0.0229, -0.1023,
        -0.0060, -0.0222,  0.0070, -0.0187,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.1410e-02, -2.1493e-01,  9.7250e-03, -7.3688e-02, -1.2716e-01,
        -2.5261e-02, -2.8378e-02, -1.4551e-02, -1.5666e-02,  3.1198e-04,
        -9.4734e-03, -4.5839e-02, -1.3069e-02, -1.2556e-02, -1.5642e-02,
         6.1055e-03, -2.8122e-02,  6.4663e-03, -6.4300e-03, -1.1616e-02,
        -1.4282e-02,  9.9807e-03, -8.1542e-04, -2.5731e-05,  1.7866e-03,
         1.1710e-02,  1.4545e-02, -3.0568e-02, -3.8050e-02,  3.5848e-03,
         9.6464e-03,  1.0249e-02, -1.9820e-02, -7.4445e-02, -1.6868e-02,
         1.8729e-02, -1.8495e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0721, -0.6870, -0.0570,  0.0604, -0.0336, -0.0010,  0.0498, -0.0297,
        -0.0095,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1038,  0.0196, -0.0128, -0.1722,  0.0026,  0.0117, -0.0872, -0.0750,
         0.0127,  0.0009, -0.0196, -0.0045, -0.0871, -0.1230, -0.0134,  0.0224,
         0.0018,  0.0244, -0.0686, -0.0597,  0.0097,  0.0149, -0.0017,  0.0013,
         0.0170,  0.0278, -0.0047,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0390, -0.2638,  0.0333, -0.0723, -0.0198, -0.0145, -0.0097, -0.0059,
         0.0024,  0.0056,  0.0048, -0.0901, -0.0021,  0.0123, -0.0048,  0.0165,
        -0.0087,  0.0067,  0.0037,  0.0120,  0.0080, -0.0454, -0.0482, -0.0123,
        -0.0041, -0.0131, -0.0051, -0.0263, -0.1157, -0.0088,  0.0090, -0.0005,
         0.0065,  0.0687,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0167,  0.0002,  0.0009, -0.0061, -0.0407, -0.0171,  0.0023,  0.0046,
         0.0090,  0.0397, -0.0067, -0.0936,  0.0416,  0.0092,  0.0053, -0.0322,
        -0.1146,  0.0517, -0.0369, -0.0123,  0.0058,  0.0006, -0.0335, -0.0117,
         0.0207, -0.0016,  0.0007, -0.0047,  0.0050,  0.0052, -0.0111, -0.0730,
        -0.0718,  0.0160, -0.0155, -0.0135, -0.0561, -0.0084,  0.0090,  0.0103,
        -0.0092, -0.0373,  0.0031,  0.0110,  0.0239,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0602, -0.0020, -0.0285, -0.0012, -0.0739, -0.1894, -0.1507, -0.0059,
         0.0060, -0.1567, -0.0835, -0.0182, -0.1098, -0.0437, -0.0512, -0.0192,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1063,  0.0228, -0.0307, -0.0131,  0.0039, -0.0240,  0.0045, -0.0517,
        -0.1975, -0.1662, -0.0455, -0.1024, -0.0014, -0.0340, -0.0693,  0.0071,
        -0.0071, -0.0065,  0.0039, -0.0145, -0.0184,  0.0155,  0.0058,  0.0044,
        -0.0054,  0.0203, -0.0177,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0007, -0.0141, -0.0704,  0.0024, -0.0246, -0.0039, -0.0185, -0.0117,
        -0.0234, -0.0035,  0.0011, -0.0028, -0.0355, -0.0076, -0.0196, -0.0048,
        -0.0293,  0.0032, -0.0007, -0.0467,  0.0114, -0.0349, -0.0083, -0.0970,
         0.0078,  0.0022, -0.0311, -0.0574,  0.0101, -0.0478,  0.0031, -0.0339,
        -0.0688, -0.0206, -0.0260, -0.0420, -0.0193, -0.0250, -0.0030,  0.0106,
        -0.0188, -0.0505, -0.0057, -0.0126,  0.0127,  0.0149], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.3933e-03,  2.0961e-02,  3.7804e-02,  6.5616e-03,  1.1695e-02,
        -1.9170e-02,  5.0838e-03, -1.0260e-01,  1.7138e-02,  2.7855e-03,
         3.0612e-03,  1.6157e-04,  1.6094e-02,  4.3590e-03,  4.8072e-03,
         7.7728e-03,  1.8286e-03, -2.3035e-02, -1.4324e-01, -2.8563e-01,
        -5.0512e-02,  8.8748e-03,  2.0017e-04, -3.1094e-03,  5.3752e-04,
        -5.9220e-03, -5.4380e-02,  8.0163e-02,  3.4354e-02, -4.3773e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0625,  0.0066, -0.1002,  0.0107,  0.0130,  0.0985,  0.1159,  0.3647,
         0.0633,  0.0593,  0.1053,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0693, -0.3258, -0.0379, -0.0366, -0.0581, -0.0264, -0.0449, -0.1492,
        -0.0012,  0.0035, -0.0033,  0.0085, -0.0121, -0.0210, -0.0174, -0.0785,
        -0.0108, -0.0304, -0.0155, -0.0496,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0485,  0.3624,  0.2018,  0.1296, -0.0316,  0.0335,  0.0949, -0.0058,
         0.0053,  0.0404, -0.0461,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.4408e-02,  2.9362e-01,  1.5059e-01,  1.6174e-01,  4.6691e-03,
         5.2563e-02,  2.1953e-02,  3.0075e-02,  6.1570e-02, -1.1695e-02,
        -3.9318e-03,  6.3376e-03,  1.2974e-04,  1.7180e-02,  9.3529e-03,
         5.2746e-02, -1.5383e-02,  9.9764e-04, -7.8324e-04, -3.1951e-04,
        -5.9955e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0740, -0.2783, -0.1061,  0.0615,  0.0044, -0.3284,  0.0137, -0.0126,
        -0.0104, -0.0144,  0.0133, -0.0380,  0.0096, -0.0081, -0.0124, -0.0037,
        -0.0075,  0.0037,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0056, -0.1380, -0.0952,  0.0067, -0.0003,  0.0162, -0.0067,  0.0031,
         0.0013, -0.0064, -0.0029,  0.0016,  0.0050, -0.0468, -0.0575,  0.0157,
        -0.0257, -0.1120, -0.0394, -0.0420, -0.0219, -0.0464, -0.0906, -0.0026,
        -0.0374,  0.0015, -0.0319, -0.0075, -0.0079, -0.0596,  0.0018, -0.0024,
        -0.0090,  0.0097, -0.0155, -0.0037, -0.0027,  0.0057,  0.0143,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.5268e-02, -3.5195e-01, -6.2677e-03, -5.7960e-02, -1.0473e-02,
        -5.3436e-03,  7.6517e-03,  1.6377e-02, -6.8024e-02, -7.0929e-03,
         2.5242e-03, -1.0053e-01,  1.7898e-02, -1.8262e-03,  7.4783e-03,
         5.9353e-03, -5.6503e-02,  1.4283e-03,  1.1746e-02,  4.3562e-03,
        -7.0142e-03,  8.0395e-03,  3.7326e-03,  6.1507e-03, -9.2466e-04,
        -7.0995e-04, -2.1361e-02,  1.2145e-03, -4.0878e-02, -6.9361e-02,
        -3.1786e-04,  7.7772e-04,  8.0861e-03, -3.4794e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.1638, -0.0103, -0.3024, -0.2397, -0.0287, -0.0968, -0.0358, -0.0642,
        -0.0584,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.7418e-02, -2.2071e-01, -1.2298e-02, -5.0499e-02, -6.8937e-03,
        -7.2810e-02,  4.3390e-03, -2.3533e-02, -2.9863e-02, -1.1363e-03,
        -3.0722e-02, -2.4317e-02, -7.9541e-04, -9.2765e-03, -2.2367e-02,
        -8.9288e-02, -8.2632e-03, -1.1241e-01, -1.4837e-02, -5.7754e-03,
         1.1064e-02,  8.0640e-03,  6.1630e-03,  2.4836e-03, -4.1646e-02,
        -2.2129e-05, -9.4384e-03, -4.0125e-05,  1.1055e-03,  1.4530e-03,
        -9.4604e-03, -3.5103e-02,  1.0680e-02, -3.2175e-02, -1.1903e-02,
        -7.3663e-03,  1.0391e-02,  4.0118e-03, -1.6790e-02,  3.0860e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0464, -0.0998, -0.1659, -0.0230, -0.0513, -0.0103,  0.0054, -0.0078,
        -0.0427, -0.0351,  0.0062,  0.0067,  0.0029,  0.0038, -0.0038,  0.0130,
         0.0106, -0.0151, -0.0397,  0.0003, -0.0373, -0.0014, -0.0008, -0.0236,
         0.0220, -0.0230, -0.0341,  0.0034, -0.0319, -0.0602,  0.0020,  0.0006,
         0.0030, -0.0090,  0.0032,  0.0095, -0.0407, -0.0354, -0.0007, -0.0010,
         0.0059, -0.0046,  0.0019,  0.0034,  0.0022,  0.0013,  0.0046, -0.0019,
         0.0048,  0.0071, -0.0249,  0.0050,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.3600e-03, -3.1015e-01, -3.2456e-02, -4.2219e-02,  1.4478e-02,
        -1.7594e-02, -9.5136e-02, -1.5345e-02, -2.6866e-02,  7.3971e-03,
        -3.5515e-03, -4.0684e-02, -1.0626e-01,  2.7859e-04, -3.4042e-02,
        -7.4536e-02, -5.0802e-02, -9.8356e-03, -6.1957e-02, -1.1158e-03,
         1.5695e-02,  3.7234e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.4254e-02,  5.5448e-01, -1.6350e-02, -3.1936e-05, -8.8002e-03,
        -1.4599e-02,  3.8114e-03,  4.9463e-02,  1.0049e-02,  5.2349e-03,
         3.6926e-02,  7.9059e-03,  9.9770e-03,  9.3170e-03, -3.5640e-02,
         6.7033e-02,  2.2115e-02,  1.9098e-02,  1.0493e-02, -3.3543e-02,
         3.0877e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1069e-02, -1.5263e-01, -4.0215e-02, -3.3924e-02, -1.8053e-02,
        -1.6601e-03,  7.0056e-03,  1.3877e-02, -1.2889e-02,  9.4007e-04,
        -1.2139e-03, -3.7408e-03,  7.6846e-03,  1.3401e-02,  3.3394e-03,
        -8.6159e-03, -1.2905e-04, -1.9681e-03, -1.8521e-02, -1.0881e-03,
         5.9071e-03,  3.1467e-03, -2.7870e-03,  2.3909e-03,  5.1968e-04,
        -6.3849e-03, -1.3987e-02, -2.6628e-02, -3.1355e-03, -3.3845e-02,
        -7.2876e-02, -1.0151e-01, -4.2489e-03, -2.8618e-02, -5.0665e-02,
        -6.7443e-03,  7.4574e-03,  1.6193e-03,  1.3705e-02,  3.2521e-03,
        -4.9449e-02,  2.4656e-04, -5.5479e-04, -1.7287e-02, -3.7986e-02,
        -6.9538e-03, -4.5714e-02, -1.8290e-02,  2.0824e-03,  1.6504e-02,
        -1.8436e-02, -2.4865e-02,  4.5045e-03,  1.2766e-02, -2.9718e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0249, -0.0242,  0.0011, -0.0263, -0.0189, -0.0284, -0.1094, -0.0197,
        -0.0011, -0.0008, -0.0029, -0.0037,  0.0031, -0.0209, -0.0603, -0.0372,
        -0.0118, -0.0235,  0.0016, -0.0407, -0.0068, -0.0362, -0.0445, -0.0020,
         0.0055,  0.0013,  0.0050, -0.0543, -0.0068,  0.0111,  0.0190,  0.0011,
        -0.0135, -0.0712,  0.0227, -0.0327,  0.0069, -0.0506, -0.0461, -0.0377,
        -0.0091,  0.0016,  0.0146, -0.0071,  0.0322,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0286, -0.2322, -0.1143, -0.0642,  0.0372, -0.0172, -0.0181, -0.0010,
        -0.0769,  0.0022,  0.0065, -0.0449, -0.0005, -0.0552, -0.0825,  0.0008,
        -0.0111, -0.0514, -0.0098, -0.0296, -0.0682,  0.0005, -0.0353,  0.0118,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0450,  0.0347,  0.0019,  0.0050, -0.0096, -0.0030, -0.0323, -0.0456,
        -0.0086, -0.0029,  0.0169,  0.0013, -0.0148, -0.0728,  0.0040,  0.0295,
        -0.0504, -0.0606, -0.0517, -0.0720,  0.0105,  0.0122, -0.0427, -0.0697,
         0.0366, -0.0167, -0.0797, -0.0174, -0.0220, -0.0575, -0.0206, -0.0307,
         0.0213,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0448, -0.4444, -0.0095, -0.0955,  0.0006, -0.0105, -0.0172, -0.0067,
        -0.0978,  0.0057,  0.0128, -0.0032,  0.0570, -0.0594,  0.0269, -0.0248,
        -0.0211, -0.0063, -0.0558,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0667, -0.0009,  0.0153, -0.0710, -0.1324, -0.0278, -0.0692, -0.0851,
        -0.1272, -0.0089, -0.0088, -0.0041,  0.0096, -0.0019, -0.0040, -0.0056,
         0.0016, -0.0360, -0.0102, -0.0184, -0.0901, -0.0872, -0.0615,  0.0012,
        -0.0554,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0816, -0.2629, -0.1469, -0.0249, -0.0160, -0.0794, -0.1896,  0.0533,
        -0.0098, -0.0023, -0.0125,  0.0184, -0.0779, -0.0014, -0.0006, -0.0114,
        -0.0112,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.0046, -0.0055,  0.0228,  0.0062,  0.0006,  0.0177,  0.0136, -0.0048,
        -0.0068,  0.0587,  0.0195,  0.0089, -0.0133,  0.0011, -0.0115,  0.0193,
         0.1109,  0.1747, -0.0349,  0.0557,  0.0147,  0.1078,  0.0999, -0.0083,
        -0.0290, -0.0140, -0.0226, -0.0135,  0.0027, -0.0065,  0.0215, -0.0683,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0315, -0.1160, -0.2144, -0.0287, -0.0170, -0.0222, -0.0528, -0.0919,
         0.0023,  0.0006,  0.0134, -0.0128, -0.0013,  0.0066, -0.0171,  0.0234,
        -0.0124,  0.0157, -0.0597, -0.0363, -0.0253, -0.0122, -0.0302, -0.1020,
        -0.0102, -0.0105, -0.0214, -0.0120,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0797,  0.0226, -0.0170,  0.0076,  0.0013,  0.0887,  0.2013,  0.0115,
         0.0020, -0.0135, -0.0160,  0.1361, -0.0062, -0.0228, -0.0191,  0.0602,
        -0.0014,  0.0512,  0.0414, -0.0040, -0.0036, -0.0009,  0.0100, -0.0039,
         0.0666, -0.0493, -0.0169,  0.0162,  0.0289,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0273, -0.0899,  0.0045, -0.0661, -0.1239,  0.0094, -0.0142, -0.0036,
        -0.0048,  0.0201,  0.0107,  0.0049, -0.0339, -0.1165,  0.0335, -0.0423,
        -0.0083, -0.1128, -0.0253, -0.0210,  0.0148,  0.0108, -0.0236, -0.0024,
        -0.0232, -0.0365, -0.0501,  0.0026, -0.0008, -0.0157, -0.0095,  0.0183,
         0.0184,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0441, -0.3017, -0.0128, -0.0718, -0.0182, -0.0513, -0.0138, -0.0737,
        -0.0149, -0.0503, -0.0700, -0.0084,  0.0165,  0.0021, -0.0037,  0.0103,
        -0.0149, -0.0109, -0.0946, -0.0100, -0.0522, -0.0075,  0.0103,  0.0362,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0334, -0.1384, -0.1573, -0.0039, -0.0753, -0.0223, -0.0373, -0.0469,
         0.0144, -0.0466, -0.0111, -0.0542, -0.0498, -0.0228, -0.0013,  0.0020,
         0.0089, -0.0306,  0.0105, -0.0901, -0.0434, -0.0178,  0.0818,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.5191e-03, -2.0595e-01,  1.3374e-04,  1.8565e-02, -8.6087e-03,
        -2.3924e-02,  8.3653e-03,  3.8815e-03, -8.3153e-03, -8.4767e-03,
        -2.3267e-02, -2.2744e-03, -1.8606e-02,  2.0466e-02, -2.1658e-03,
        -5.5669e-03, -8.2808e-02, -4.6968e-02, -3.9748e-02, -6.3645e-03,
         1.7650e-02, -2.5009e-02, -2.1139e-03, -2.3715e-02, -6.2665e-02,
        -4.6064e-03, -5.3028e-03, -3.6723e-03,  2.0614e-03, -5.4210e-02,
        -6.2586e-02, -1.4592e-02, -3.2978e-02, -2.2464e-03, -3.3249e-02,
        -6.5320e-02, -8.4948e-03, -2.0518e-02, -1.6037e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0279, -0.0680,  0.0005, -0.0373, -0.1698, -0.0103,  0.0184, -0.0044,
        -0.0092,  0.0071,  0.0017,  0.0103,  0.0108,  0.0104,  0.0059, -0.0577,
         0.0066,  0.0004,  0.0022,  0.0182, -0.0370,  0.0008, -0.0055, -0.0038,
        -0.0686, -0.1080, -0.0163, -0.0337, -0.0130, -0.0654,  0.0027, -0.0085,
         0.0205, -0.0041, -0.0031, -0.0040, -0.0147, -0.0466,  0.0178,  0.0090,
        -0.0104, -0.0139, -0.0156,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0221, -0.1746,  0.0041,  0.0064,  0.0146, -0.0048, -0.0264,  0.0009,
        -0.0696,  0.0046,  0.0064,  0.0034, -0.0202, -0.0116, -0.0029, -0.0105,
        -0.0415, -0.0556, -0.0143,  0.0004,  0.0002,  0.0082, -0.0020,  0.0094,
        -0.0004, -0.0011,  0.0035,  0.0195,  0.0107,  0.0067,  0.0084, -0.0495,
        -0.0934, -0.0266, -0.0054, -0.0106, -0.0081, -0.0202, -0.0262, -0.0252,
        -0.0350, -0.0435, -0.0049, -0.0015, -0.0075, -0.0092, -0.0020, -0.0046,
        -0.0045, -0.0037, -0.0096,  0.0026, -0.0112, -0.0045,  0.0010,  0.0067,
        -0.0181], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0931,  0.4123,  0.0314,  0.0238,  0.0528,  0.0032,  0.1121,  0.1092,
         0.0017,  0.0388,  0.0806,  0.0063, -0.0298,  0.0049,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0426, -0.2569, -0.0062,  0.0066, -0.0048, -0.0236, -0.0077, -0.0014,
        -0.0421, -0.0010, -0.0635, -0.1010, -0.0182,  0.0064, -0.0421, -0.0211,
        -0.0139, -0.0335,  0.0166, -0.0637, -0.1160, -0.0120, -0.0402, -0.0023,
         0.0202,  0.0061,  0.0067,  0.0035, -0.0013,  0.0187,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0358, -0.4457,  0.0565, -0.2441, -0.0991,  0.0900,  0.0009,  0.0279,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0795, -0.1653, -0.0160, -0.0405, -0.0420, -0.0929,  0.0015, -0.0020,
        -0.0091, -0.0058,  0.0281,  0.0010, -0.0574, -0.0841, -0.0036,  0.0080,
        -0.0021, -0.0023,  0.0049,  0.0039, -0.0054,  0.0091,  0.0029, -0.0065,
        -0.0630,  0.0096, -0.0444, -0.0095, -0.0564, -0.0752,  0.0074, -0.0206,
        -0.0401,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4777e-02, -1.8664e-01, -8.2013e-02, -1.0982e-02, -1.4252e-04,
        -2.4344e-02, -1.0660e-02, -5.7736e-03,  9.1103e-03, -1.0088e-02,
         7.2838e-04, -1.2503e-02, -1.1167e-02,  5.9500e-03,  1.2667e-03,
        -3.7021e-03, -8.7692e-04,  1.7319e-04, -1.1485e-02,  1.5771e-03,
        -8.6919e-04, -2.6561e-03, -8.7564e-03, -6.1013e-05,  1.0757e-02,
        -5.9011e-03, -1.2171e-02, -4.7648e-02, -3.0229e-03,  5.5089e-04,
        -2.8403e-03, -1.0056e-02, -4.5239e-02, -9.3359e-02, -9.9296e-04,
        -7.5127e-03, -1.7553e-02, -2.3370e-03, -4.4389e-03,  1.1383e-03,
         3.5425e-03,  2.6164e-05, -6.1775e-03, -4.9864e-02, -8.8934e-03,
        -2.2655e-03, -3.7108e-04, -9.8566e-04, -7.0322e-03, -9.0274e-03,
        -8.2426e-03, -2.8218e-02, -5.8602e-03, -2.1861e-02, -7.0974e-03,
         1.5800e-03, -1.2123e-03,  8.1600e-04,  1.7866e-04, -1.0249e-02,
        -9.4286e-03, -1.0870e-02, -6.8346e-02, -4.7589e-03, -9.3882e-04,
        -7.0265e-03, -5.4169e-03,  1.2932e-02,  1.2411e-02, -8.5004e-04,
         1.1699e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0276, -0.0061, -0.0266,  0.0038, -0.0456, -0.0080, -0.0429, -0.0066,
         0.0046,  0.0208, -0.0067,  0.0065,  0.0096,  0.0049,  0.0178,  0.0133,
        -0.0092,  0.0062, -0.0259, -0.1296, -0.0058, -0.0105,  0.0090, -0.0532,
         0.0362, -0.0840, -0.0157, -0.0009,  0.0021, -0.0011, -0.0078, -0.0112,
        -0.0769, -0.0675, -0.0112, -0.0068, -0.0802, -0.0673, -0.0052, -0.0065,
         0.0176,  0.0010,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0320, -0.2349,  0.0202,  0.0089, -0.0056, -0.0039,  0.0051, -0.0298,
        -0.0096, -0.0360, -0.0239, -0.0505, -0.0551, -0.0022,  0.0075, -0.0022,
        -0.0024, -0.0668, -0.0879, -0.0053, -0.0087, -0.0303,  0.0061, -0.0011,
        -0.0068, -0.0659, -0.0286, -0.0193, -0.0822,  0.0107, -0.0046, -0.0071,
         0.0011, -0.0074, -0.0179,  0.0051,  0.0072,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0930, -0.4834, -0.0502, -0.1314,  0.0073,  0.0254,  0.0008,  0.0154,
         0.0264,  0.0027, -0.0406,  0.0118, -0.0467, -0.0290, -0.0360,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0305,  0.0400,  0.0025,  0.0055,  0.1880,  0.1663,  0.0308,  0.0956,
         0.1957,  0.0428, -0.0279,  0.0126,  0.0135, -0.0019, -0.0429,  0.0028,
         0.0149,  0.0115,  0.0184, -0.0559,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0122,  0.0059, -0.0082,  0.0089, -0.1062, -0.0101, -0.0090, -0.0273,
        -0.0109,  0.0052, -0.0528, -0.0883, -0.0093, -0.0098, -0.0449, -0.0319,
         0.0115, -0.0373,  0.0149, -0.0427, -0.0101, -0.0028, -0.0095, -0.0075,
        -0.0012, -0.0450, -0.0751,  0.0012, -0.0101, -0.0267, -0.0801, -0.0008,
         0.0038,  0.0003, -0.0470, -0.0371,  0.0080, -0.0311, -0.0051, -0.0044,
        -0.0083, -0.0078,  0.0297,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0051,  0.0018, -0.0151, -0.0052, -0.0321, -0.0058, -0.0474, -0.1402,
        -0.0115, -0.0030,  0.0176, -0.0003,  0.0043, -0.1114, -0.1340,  0.0136,
         0.0075, -0.0012, -0.0070, -0.0020, -0.0765, -0.0072,  0.0121, -0.0004,
        -0.0128,  0.0012, -0.0186, -0.0080, -0.0516, -0.0100, -0.0126, -0.0450,
        -0.0018, -0.0535, -0.0698, -0.0246,  0.0110, -0.0173,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0446, -0.2950, -0.1036, -0.0946,  0.0090, -0.0792, -0.0434, -0.0402,
        -0.0128, -0.0613, -0.0099, -0.0179, -0.0554, -0.0124, -0.0055, -0.0229,
         0.0037, -0.0448, -0.0140,  0.0141,  0.0093,  0.0064,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5626e-02, -1.0409e-01, -2.7005e-02, -1.0874e-02, -8.5481e-03,
        -3.6118e-02,  3.9952e-03, -1.9378e-03, -1.3404e-02, -1.2904e-01,
        -1.1584e-02,  3.0113e-02, -6.1669e-03, -2.0885e-02, -4.9581e-02,
        -4.6745e-02, -7.4887e-02,  4.7773e-03, -5.5444e-03, -2.7224e-02,
        -5.8006e-03, -1.1302e-02, -4.7489e-02, -1.7488e-02, -9.6609e-05,
        -1.5057e-03, -5.0306e-03,  2.9252e-02, -4.2306e-03, -6.0030e-02,
        -3.0487e-02,  7.8812e-03, -4.8417e-03, -8.6158e-03, -2.1502e-02,
        -7.5821e-02,  8.3707e-03, -7.1147e-04,  9.2183e-03, -2.2180e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0552,  0.0006,  0.0055, -0.0077, -0.0052, -0.0355, -0.0078,  0.0219,
        -0.0328, -0.0018, -0.0044,  0.0013, -0.0124, -0.1211, -0.0820, -0.0090,
         0.0218,  0.0239, -0.0669, -0.0091, -0.0013, -0.0049, -0.0041,  0.0194,
        -0.0012, -0.1121, -0.0245, -0.0868, -0.1411, -0.0499, -0.0060, -0.0227,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.6653e-02, -3.7575e-01, -2.4515e-02, -1.5802e-02, -2.8819e-02,
         8.0693e-03, -1.6126e-02, -7.5033e-02, -8.7492e-02, -1.8030e-03,
        -1.5728e-02, -2.0681e-02, -1.7511e-03, -1.8362e-03, -2.1338e-03,
        -3.1340e-03, -4.2769e-02, -1.7990e-03, -3.9867e-03,  1.7893e-04,
        -2.8613e-03,  4.6460e-03, -1.0764e-02, -5.2601e-02,  5.2311e-04,
         1.1965e-02, -3.8511e-02, -3.9132e-03,  3.7515e-03, -2.3104e-02,
         2.9621e-03, -5.2340e-04, -7.6759e-03,  1.0442e-02, -7.1696e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0116, -0.0028, -0.1569, -0.1175, -0.0288, -0.0146, -0.1929, -0.1163,
        -0.0499, -0.1123,  0.0124, -0.0578,  0.0094, -0.0463, -0.0181, -0.0024,
        -0.0210, -0.0057,  0.0034,  0.0107,  0.0091,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0336, -0.4245,  0.0269, -0.0747, -0.1118, -0.0403,  0.0021, -0.0689,
        -0.0427,  0.0424,  0.0012,  0.0050,  0.0055, -0.0105, -0.0209,  0.0152,
        -0.0282, -0.0121,  0.0334,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0169, -0.0781, -0.1089,  0.0009, -0.0328, -0.0026,  0.0130,  0.0326,
         0.0183, -0.0875, -0.1645, -0.0058, -0.0737, -0.1465, -0.0287, -0.0535,
         0.0274, -0.0312,  0.0319, -0.0453,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0090, -0.3100,  0.0330, -0.0732,  0.0207, -0.0176, -0.0591, -0.0342,
        -0.0127, -0.0048, -0.0097, -0.0234,  0.0049, -0.0613, -0.0412, -0.0496,
        -0.0769, -0.0121, -0.0446, -0.0575,  0.0315,  0.0055,  0.0076,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0171, -0.2064, -0.0826, -0.0152,  0.0040, -0.0668,  0.0368, -0.0136,
        -0.0464, -0.0407, -0.0013, -0.0963, -0.0064, -0.1310,  0.0251,  0.0323,
         0.0018,  0.0009, -0.0106, -0.0414, -0.0796,  0.0049,  0.0162, -0.0066,
         0.0160,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0351,  0.3093,  0.0351,  0.0477,  0.0067, -0.0049, -0.0231,  0.0084,
        -0.0263, -0.0165,  0.0687, -0.0089,  0.0096,  0.0265,  0.0020, -0.0044,
         0.0076,  0.0021, -0.0040,  0.0048,  0.0379,  0.0907,  0.0281,  0.0563,
         0.0360,  0.0175,  0.0037,  0.0085,  0.0106,  0.0156,  0.0039,  0.0017,
         0.0111, -0.0098,  0.0173,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0100,  0.0050, -0.0337, -0.0735, -0.0078,  0.0209, -0.0876,  0.0103,
        -0.1175, -0.0081, -0.0186, -0.0154, -0.0115, -0.1028, -0.0779, -0.1848,
        -0.0396, -0.1177,  0.0247,  0.0118, -0.0058,  0.0152,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0403, -0.0109, -0.0474,  0.0030,  0.0237,  0.0267,  0.0251,  0.0237,
        -0.0217,  0.2565,  0.0087,  0.0688,  0.0230,  0.0257, -0.0046,  0.1276,
         0.0337,  0.0214, -0.0262,  0.0093, -0.1718,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0594, -0.3802,  0.0038, -0.0113, -0.0231,  0.0054, -0.0824, -0.1371,
        -0.0016, -0.0153,  0.0155,  0.0128,  0.0083, -0.0706,  0.0082, -0.0559,
         0.0287,  0.0720, -0.0085,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0671,  0.0171, -0.0660, -0.0103,  0.0093, -0.0815,  0.0079, -0.0084,
        -0.0032, -0.0060, -0.0124, -0.0155, -0.0007, -0.0229, -0.0005, -0.0097,
        -0.0387, -0.0127, -0.0046, -0.0297, -0.0061,  0.0327, -0.0354,  0.0078,
        -0.0516, -0.0019, -0.0299, -0.0575,  0.0011, -0.1192, -0.0081,  0.0038,
        -0.0347, -0.0792, -0.0037, -0.0044, -0.0177,  0.0013, -0.0087, -0.0302,
         0.0015,  0.0115,  0.0027, -0.0012, -0.0065,  0.0041,  0.0134],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0490, -0.2703, -0.1944, -0.0203,  0.0127,  0.0009,  0.0005, -0.0283,
        -0.0420,  0.0051, -0.0026, -0.0073,  0.0027,  0.0082,  0.0017, -0.0022,
        -0.1164,  0.0009,  0.0026, -0.0117,  0.0129,  0.0042, -0.0627, -0.0458,
         0.0112,  0.0165,  0.0043,  0.0024, -0.0018,  0.0075,  0.0062,  0.0065,
         0.0171, -0.0210,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0246,  0.4121,  0.0156, -0.0153,  0.0270,  0.0405, -0.0026, -0.0005,
         0.0049,  0.0176, -0.0127,  0.0045,  0.0493,  0.0862, -0.0046,  0.0023,
         0.0112, -0.0336, -0.0052,  0.0943,  0.0076, -0.0014,  0.0145, -0.0015,
         0.0060, -0.0049, -0.0026,  0.0195,  0.0126,  0.0472,  0.0112,  0.0029,
        -0.0012, -0.0023,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-2.8381e-03, -4.0066e-01,  1.4233e-02,  7.1243e-03, -3.0694e-02,
        -1.0032e-01,  4.8579e-02, -1.5332e-03, -3.6384e-04, -1.0947e-01,
        -2.8421e-02, -3.8554e-03, -4.7405e-02, -1.2685e-01, -3.4873e-02,
        -1.7832e-02,  5.3632e-03, -1.9589e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0695, -0.0071,  0.0021, -0.0011, -0.0028, -0.0085,  0.0127, -0.0443,
        -0.1038, -0.2247, -0.0109,  0.0134, -0.0226, -0.0133,  0.0158, -0.0233,
        -0.0450,  0.0036,  0.0280, -0.2082,  0.0180, -0.0097, -0.0242,  0.0227,
         0.0244, -0.0131, -0.0270,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1157, -0.0050,  0.0013,  0.0210, -0.0045, -0.0012, -0.0123, -0.0475,
        -0.1324, -0.0125, -0.0059, -0.0266, -0.2649, -0.0180, -0.0073,  0.0199,
        -0.0054, -0.0154, -0.0053, -0.1584,  0.0287,  0.0051, -0.0171,  0.0094,
         0.0172, -0.0420,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0963, -0.2270,  0.0013, -0.0832, -0.1015, -0.0347, -0.0149, -0.0042,
        -0.0571, -0.0375, -0.0686, -0.0743,  0.0162,  0.0152, -0.0036, -0.0727,
        -0.0126, -0.0272, -0.0051,  0.0257,  0.0208,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0197, -0.2939, -0.1784, -0.1964,  0.0059, -0.0372, -0.0865, -0.0101,
        -0.0173, -0.0106,  0.0121, -0.0635,  0.0112, -0.0133,  0.0437,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0824,  0.0051,  0.0294,  0.0082,  0.0072, -0.0136,  0.0294, -0.0925,
        -0.0093, -0.0409,  0.0023, -0.0244, -0.3512, -0.0954,  0.0917,  0.0038,
         0.1014, -0.0117,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0324, -0.2519, -0.0840, -0.1211, -0.0132, -0.0340,  0.0110, -0.0610,
        -0.0321, -0.0584, -0.0852, -0.0486,  0.0083, -0.0534, -0.0189, -0.0080,
        -0.0043,  0.0190,  0.0550,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0272, -0.1613, -0.0961, -0.0223, -0.0004, -0.0304,  0.0070,  0.0081,
        -0.0301,  0.0015, -0.0036,  0.0155, -0.0321,  0.0069,  0.0024, -0.0181,
        -0.0522, -0.0162,  0.0173,  0.0003, -0.0751,  0.0024, -0.0375, -0.0118,
        -0.0355, -0.0016, -0.0702,  0.0058, -0.0028, -0.0255,  0.0032, -0.0093,
        -0.0413, -0.0045, -0.0153, -0.0205, -0.0378,  0.0005, -0.0202, -0.0081,
         0.0023,  0.0019,  0.0033, -0.0146], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0444, -0.2170, -0.0171, -0.1040, -0.0654, -0.0433, -0.0191, -0.0251,
        -0.0713, -0.0128, -0.0190,  0.0175,  0.0214, -0.0009, -0.0007, -0.0060,
         0.0037, -0.0527, -0.0052, -0.0114, -0.0153,  0.0006, -0.0069, -0.0043,
        -0.0224,  0.0084, -0.0007, -0.0955,  0.0010,  0.0055, -0.0102, -0.0135,
         0.0065, -0.0285,  0.0227,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.8209e-03,  3.9392e-01,  2.8261e-02,  5.7193e-02,  4.7822e-03,
         2.3238e-02,  9.9619e-02,  4.9257e-02,  2.4922e-02,  1.4442e-03,
         3.7335e-03,  3.4890e-02,  9.3158e-02,  4.0410e-03, -7.8108e-03,
         5.9129e-02,  2.7645e-04,  3.5893e-03,  4.1392e-03,  1.1095e-02,
        -2.7728e-03, -3.1704e-03, -1.2942e-02, -7.0799e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0188, -0.2098,  0.0037, -0.0514,  0.0015, -0.0037,  0.0111, -0.0173,
        -0.0692, -0.0048, -0.0154,  0.0074,  0.0085, -0.0467, -0.0592, -0.0010,
         0.0160, -0.0020, -0.0587,  0.0029,  0.0037, -0.0363, -0.0044, -0.0390,
        -0.0123, -0.0592, -0.0802, -0.0328, -0.0145, -0.0266, -0.0398, -0.0038,
        -0.0191, -0.0192,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0372,  0.0112, -0.0065, -0.0365,  0.0123,  0.0021, -0.1718, -0.0319,
         0.0221, -0.0149,  0.0146, -0.0021, -0.0942, -0.0042, -0.0930, -0.0975,
        -0.0143,  0.0017, -0.0075, -0.0062,  0.0064, -0.0188,  0.0108, -0.1187,
        -0.0502, -0.0178, -0.0495,  0.0193, -0.0198, -0.0070,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.2901,  0.0529, -0.0372, -0.1778, -0.0007, -0.0146, -0.0257, -0.0059,
        -0.1249, -0.2249, -0.0165,  0.0076, -0.0211,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0403,  0.0228, -0.0520,  0.0698,  0.0082, -0.0375, -0.0914, -0.1272,
        -0.1910, -0.0073,  0.0147, -0.0796,  0.0757, -0.0574, -0.1116, -0.0013,
        -0.0058, -0.0063,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0596, -0.4214, -0.2711, -0.0411, -0.1648,  0.0086,  0.0072, -0.0262,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0023, -0.2161, -0.0012, -0.0288, -0.0289, -0.0322, -0.0137, -0.0173,
         0.0096, -0.0012,  0.0113,  0.0307, -0.0063, -0.0027,  0.0049, -0.0021,
         0.0023,  0.0086,  0.0095, -0.0083,  0.0142,  0.0040, -0.0164, -0.0180,
         0.0032, -0.0213, -0.0145, -0.0023, -0.0454, -0.0597,  0.0046, -0.0649,
        -0.0112, -0.0762, -0.0170, -0.0503, -0.0436, -0.0163,  0.0027,  0.0063,
        -0.0168, -0.0103, -0.0109,  0.0012,  0.0261, -0.0013,  0.0033],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0731, -0.2727, -0.0979, -0.3193,  0.0264,  0.0173,  0.0035, -0.0780,
        -0.0361, -0.0624,  0.0131,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0493, -0.2895, -0.0345, -0.1242, -0.0359, -0.0042, -0.0456, -0.1295,
        -0.0063, -0.0788, -0.0250,  0.0200, -0.0826,  0.0010, -0.0166,  0.0188,
        -0.0152, -0.0230,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0426, -0.0193, -0.4565,  0.1070,  0.0224,  0.0335, -0.0641, -0.0057,
        -0.1111, -0.0438, -0.0193, -0.0045,  0.0303,  0.0090, -0.0310,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1223,  0.2843,  0.0347,  0.0289,  0.0428,  0.0647,  0.0915,  0.0032,
         0.0310,  0.0854,  0.0884,  0.0088, -0.0119,  0.0996,  0.0026,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0816,  0.0107,  0.0166, -0.0273, -0.0096,  0.0297,  0.0037,  0.0112,
        -0.1209, -0.0064, -0.0158, -0.0540, -0.1084,  0.0112, -0.0089, -0.0202,
        -0.0672,  0.0077,  0.0226,  0.0009, -0.1015, -0.1535,  0.0675, -0.0217,
        -0.0082, -0.0131,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1242, -0.0095,  0.0444,  0.0275,  0.0380,  0.0171,  0.0828, -0.1182,
        -0.4315, -0.0111,  0.0043,  0.0012,  0.0391,  0.0291, -0.0219,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.9163e-03, -3.6238e-01,  2.3072e-03, -4.7672e-02, -6.5533e-02,
        -5.9506e-02, -4.1611e-02,  1.2611e-02, -8.1270e-03, -1.5941e-01,
         8.3105e-03,  1.4458e-03, -2.5797e-02, -1.1962e-02, -5.0217e-03,
         1.2005e-02, -1.1072e-02, -3.2889e-03, -2.9132e-02,  1.3319e-02,
        -3.7649e-02, -7.2086e-03, -6.3831e-03, -3.5003e-04,  3.5877e-02,
        -2.3115e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0397, -0.0090, -0.0074,  0.0351, -0.0101, -0.0263,  0.0145,  0.0112,
        -0.0540, -0.0537, -0.1187, -0.0045, -0.0213,  0.0005, -0.1425, -0.1342,
        -0.0084, -0.0669, -0.0119, -0.1020, -0.0121, -0.0468, -0.0004, -0.0469,
         0.0209, -0.0011,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.0275, -0.2727, -0.0225, -0.0205, -0.0217, -0.0035, -0.0289, -0.0736,
         0.0062, -0.0393, -0.0078, -0.0389, -0.0580, -0.0024, -0.0244,  0.0009,
         0.0010, -0.0021, -0.0219, -0.0120, -0.0048, -0.0187,  0.0124, -0.0459,
        -0.0919, -0.0070, -0.0338, -0.0141,  0.0048,  0.0093, -0.0161, -0.0284,
        -0.0074,  0.0057,  0.0047,  0.0092,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0125, -0.0940, -0.0476,  0.0051, -0.0633,  0.0048, -0.0611, -0.0696,
         0.0103,  0.0112, -0.0321, -0.0017, -0.0135,  0.0253, -0.0026, -0.0152,
         0.0004, -0.0043,  0.0109,  0.0127, -0.0014, -0.0630, -0.0166, -0.0148,
        -0.0494, -0.0617,  0.0046, -0.0580, -0.0517,  0.0063,  0.0042, -0.0605,
        -0.0776, -0.0094, -0.0141,  0.0084,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0503, -0.0643, -0.0004, -0.0492,  0.0555, -0.1992, -0.0029, -0.0448,
        -0.0550, -0.0113, -0.0072, -0.1222, -0.1950, -0.0244,  0.0148, -0.0057,
        -0.0167, -0.0166, -0.0536, -0.0110,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.6527e-02, -1.4605e-01, -1.4629e-02, -2.5130e-02,  1.6449e-02,
        -7.8431e-03, -1.3210e-02, -1.5511e-02, -6.8925e-03, -3.7983e-02,
        -1.9507e-03, -1.5910e-02, -1.9116e-02, -1.1590e-04, -1.0129e-01,
        -3.8564e-02, -2.5902e-02, -4.0552e-02, -8.7639e-03, -5.2472e-03,
        -1.8802e-02, -9.9453e-03,  1.4779e-03,  2.0595e-02, -4.0239e-03,
         6.5667e-04, -2.7466e-03, -9.4932e-04, -1.3920e-02, -5.0689e-02,
        -6.6411e-02, -2.2676e-02, -5.6570e-03, -9.7397e-02,  9.8413e-03,
        -1.7803e-03,  7.0697e-02, -4.0937e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.8091e-02, -1.8365e-01, -8.6023e-02, -8.8536e-02, -1.4481e-02,
        -6.4307e-04, -9.0914e-03,  3.5752e-04, -1.2771e-02, -1.4226e-02,
        -6.6030e-02, -9.2384e-02, -8.7443e-03,  1.3743e-03, -9.7955e-03,
        -8.2023e-02, -1.1006e-02, -3.7820e-02, -6.7419e-04,  1.7520e-04,
         8.2366e-03,  5.3173e-03,  1.2298e-02,  3.5133e-03,  9.6752e-03,
        -2.3001e-02, -7.4143e-03, -2.7689e-02, -6.5664e-02, -1.9334e-03,
        -8.0076e-03, -1.8335e-02,  1.8193e-03, -5.5140e-03, -4.0948e-02,
        -3.4023e-03, -7.4744e-03, -7.1082e-04, -2.6955e-03,  4.9212e-03,
         3.5284e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0086, -0.3636,  0.0077, -0.0367,  0.0067, -0.1263, -0.0153,  0.0059,
        -0.0014,  0.0044,  0.0067, -0.0643, -0.0780, -0.0054, -0.0184, -0.1088,
        -0.0142, -0.0475, -0.0034, -0.0252,  0.0372,  0.0146,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0245,  0.0110,  0.0036,  0.0040,  0.0098, -0.0361, -0.0016,  0.0034,
        -0.0129, -0.0013, -0.0070, -0.0615, -0.0196, -0.0625,  0.0011, -0.0960,
         0.0016, -0.0687, -0.0135, -0.0455, -0.0947,  0.0105,  0.0346, -0.0499,
         0.0062,  0.0033, -0.0160, -0.0582, -0.0833,  0.0028, -0.0459, -0.0609,
        -0.0185,  0.0056,  0.0244,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0719, -0.0490, -0.2326,  0.0121, -0.0684, -0.0057, -0.0013,  0.0221,
        -0.0008, -0.0100, -0.0302,  0.0083, -0.0104, -0.0041, -0.0121, -0.0297,
         0.0245,  0.0018, -0.0925, -0.0105, -0.0362,  0.0213, -0.0233, -0.0718,
        -0.0425, -0.0167, -0.0655, -0.0040, -0.0206,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0368, -0.3344, -0.0908, -0.0668, -0.0218, -0.0550, -0.0045, -0.0046,
        -0.0075, -0.0281, -0.0249, -0.0191,  0.0068,  0.0128,  0.0083, -0.0407,
        -0.0066, -0.0070, -0.0488,  0.0154,  0.0084, -0.0127,  0.0154,  0.0061,
         0.1165,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0193, -0.0077,  0.0068,  0.0034, -0.2495, -0.0417, -0.0335, -0.1345,
        -0.1438, -0.0304, -0.0096, -0.0059,  0.0113,  0.0268, -0.0839, -0.1050,
         0.0168, -0.0302, -0.0039,  0.0133, -0.0187,  0.0003,  0.0038,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.3841e-02, -4.0866e-03,  6.3308e-03, -6.5025e-05, -6.3411e-03,
         1.5526e-02,  1.4170e-02,  5.9978e-03,  6.3068e-03, -2.7443e-03,
        -5.8881e-03, -3.8587e-03, -9.9613e-02, -6.8180e-02, -1.0651e-02,
        -1.6827e-02,  1.8557e-02, -5.9169e-02, -5.2635e-03, -4.9004e-03,
        -8.1837e-03, -1.9509e-02,  8.6458e-03, -2.5724e-02, -2.8334e-02,
        -2.3400e-02, -1.8516e-01, -1.0436e-02, -1.2501e-01, -1.0270e-02,
        -5.0285e-02, -4.2675e-03, -6.7089e-03, -5.6076e-03, -6.3181e-02,
        -1.3893e-02,  8.0733e-03,  3.3336e-03,  1.6584e-03,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0813, -0.0125,  0.0436,  0.1024,  0.3213,  0.0219,  0.0058,  0.1760,
         0.1819, -0.0006,  0.0169,  0.0359,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0542, -0.0686, -0.2507,  0.0167,  0.0036, -0.0693, -0.0390, -0.0051,
         0.0068, -0.0202,  0.0059, -0.0051, -0.0085,  0.0020, -0.0457,  0.0036,
        -0.0312, -0.0586, -0.0058, -0.0079, -0.0238, -0.0232, -0.0008, -0.0080,
         0.0006, -0.0045, -0.0136, -0.0188, -0.0127, -0.0102,  0.0248,  0.0173,
        -0.0356,  0.0012,  0.0099, -0.0027, -0.0376, -0.0046, -0.0006,  0.0005,
         0.0079,  0.0097, -0.0005,  0.0021,  0.0203,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0140, -0.4883, -0.0562, -0.1306, -0.0382,  0.0213,  0.0135, -0.0201,
        -0.1211, -0.0287,  0.0093,  0.0589,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0484, -0.4473, -0.0775, -0.1601, -0.0358, -0.0747, -0.0577,  0.0985,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.4290e-02, -1.9445e-01, -3.8206e-02, -8.4933e-03, -4.2575e-02,
        -8.9346e-03, -7.9711e-03, -1.0400e-02, -3.4488e-02, -5.4421e-02,
        -2.2692e-02, -1.1112e-02, -2.8452e-03, -1.1847e-03,  3.4029e-03,
        -8.9883e-04,  1.8504e-02,  1.2350e-03,  7.4700e-03, -2.2607e-03,
         3.2044e-03,  5.7672e-03, -7.9637e-03, -3.7214e-05, -2.9168e-03,
        -1.8601e-03, -2.2237e-03, -6.9081e-03, -1.9187e-02, -2.4990e-03,
        -3.3264e-03, -3.6305e-03,  4.8110e-05,  1.4497e-03, -3.4856e-03,
         6.5160e-04, -4.4643e-03, -3.9659e-03,  1.8771e-04, -2.4290e-03,
        -5.5539e-03, -2.2484e-03, -3.2821e-03, -1.0910e-02, -7.2964e-03,
         3.0499e-02, -7.5095e-02, -1.0173e-02, -6.2215e-02, -8.5333e-02,
        -1.1680e-02, -1.9921e-03, -3.5270e-02,  5.0925e-03, -1.6132e-02,
        -2.4655e-03, -6.1968e-03,  7.0061e-04,  5.1305e-03,  1.8036e-02,
         1.2659e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0026, -0.0055,  0.0069, -0.0469, -0.1157,  0.0096, -0.0354,  0.0305,
         0.0129,  0.0059, -0.0171, -0.1104,  0.0010, -0.0648, -0.0962,  0.0052,
        -0.0188,  0.0076, -0.0407, -0.0405, -0.0981, -0.0487, -0.0125, -0.0114,
        -0.0346, -0.0348, -0.0161, -0.0157, -0.0145, -0.0169,  0.0006,  0.0011,
        -0.0207,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0282, -0.0773, -0.1400, -0.0947,  0.0004,  0.0206, -0.0955,  0.0159,
         0.0004, -0.0099, -0.0014, -0.0149, -0.0684, -0.0123,  0.0259, -0.0157,
        -0.0098, -0.0551, -0.1613,  0.0172,  0.0502,  0.0327, -0.0515, -0.0003,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0511, -0.1819, -0.1305, -0.0073, -0.0082, -0.0942,  0.0073, -0.0518,
         0.0047, -0.0415, -0.0871, -0.0755, -0.1073,  0.0043,  0.0089, -0.0042,
         0.0078, -0.0111,  0.0189,  0.0034, -0.0465, -0.0110, -0.0086, -0.0271,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0217, -0.0226, -0.0444, -0.0966, -0.0105,  0.0224, -0.0039, -0.0553,
        -0.0081, -0.0030,  0.0086, -0.0076, -0.0104, -0.0066, -0.0578, -0.0539,
        -0.0652,  0.0039, -0.0029, -0.0424, -0.1258, -0.0036, -0.0671, -0.0850,
        -0.0029,  0.0050, -0.0398, -0.0018,  0.0140, -0.0299, -0.0195, -0.0121,
         0.0198,  0.0259,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0149,  0.0136,  0.0044,  0.0110, -0.1352, -0.0089, -0.0102, -0.0152,
         0.0029,  0.0062,  0.0106,  0.0285, -0.0120, -0.1890,  0.0129,  0.0415,
        -0.0858,  0.0228, -0.0793, -0.0049, -0.0477, -0.1438,  0.0362, -0.0084,
        -0.0539,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0737,  0.0212,  0.0052,  0.0045,  0.0150,  0.0408, -0.1286, -0.0695,
        -0.1055, -0.3965, -0.0238, -0.0189, -0.0490, -0.0059,  0.0017, -0.0401,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0258, -0.3091,  0.0018, -0.0933, -0.0179, -0.0583, -0.1806,  0.0322,
         0.0054, -0.0106,  0.0128,  0.0138, -0.0280, -0.0051,  0.0250,  0.0055,
        -0.0392, -0.0663, -0.0235, -0.0188, -0.0027,  0.0196,  0.0046,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0241, -0.0020,  0.0076, -0.1996, -0.3824,  0.0759,  0.0406, -0.0142,
        -0.0264,  0.0222, -0.0137,  0.0390,  0.0138, -0.1384,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0347, -0.0798, -0.0741, -0.1146, -0.0004, -0.0238, -0.0591,  0.0178,
        -0.0258, -0.0060, -0.0968, -0.1024, -0.0029, -0.0020, -0.0026, -0.0275,
        -0.0430, -0.0949, -0.0128, -0.1071, -0.0248,  0.0011,  0.0202,  0.0258,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0270,  0.1262, -0.1210, -0.0413, -0.2510, -0.0322, -0.3346, -0.0185,
         0.0167, -0.0210, -0.0105,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0369,  0.0290,  0.0146,  0.0088,  0.0418, -0.1260, -0.1670, -0.0328,
        -0.1086, -0.0278, -0.0814,  0.0026, -0.0519,  0.0133, -0.0575, -0.0009,
        -0.0230, -0.1093, -0.0351, -0.0220, -0.0095,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0686, -0.0159, -0.0476, -0.0180, -0.0573, -0.0097, -0.0254, -0.3300,
        -0.0927, -0.1771, -0.0583, -0.0203, -0.0170, -0.0623,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0776,  0.0053, -0.0011, -0.0137, -0.0307,  0.0128,  0.0046, -0.0976,
        -0.0302, -0.0026,  0.0850,  0.1642,  0.2776,  0.0286,  0.0022,  0.0479,
        -0.0464,  0.0372,  0.0150, -0.0040,  0.0157,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0025, -0.3393,  0.0058, -0.1033,  0.0413, -0.1027, -0.0212, -0.1562,
        -0.0124, -0.1144, -0.0311, -0.0265,  0.0432,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0558, -0.1697, -0.0032, -0.0067, -0.0025, -0.0219, -0.0471, -0.0482,
        -0.0236,  0.0105,  0.0236,  0.0021, -0.0030,  0.0077,  0.0067,  0.0014,
         0.0087,  0.0028,  0.0092,  0.0157, -0.0081,  0.0093,  0.0116,  0.0078,
        -0.0594,  0.0139, -0.0720, -0.0647, -0.0899, -0.0047, -0.0389,  0.0191,
         0.0092,  0.0129,  0.0125,  0.0078, -0.0133,  0.0008,  0.0032, -0.0115,
        -0.0141, -0.0008, -0.0027,  0.0284,  0.0131], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.7769e-02, -3.2415e-02, -6.7015e-02, -3.8717e-03,  1.3739e-02,
        -1.1547e-02,  7.9618e-03, -2.8843e-02,  2.1414e-03, -6.0707e-03,
        -8.2599e-02,  1.5842e-02, -5.4760e-03, -3.6491e-04, -5.8613e-02,
        -3.2703e-02, -1.4077e-01, -1.0110e-02, -8.1896e-03,  1.6306e-02,
        -1.4966e-02, -1.0268e-01, -5.8708e-03, -8.2678e-02, -9.8196e-03,
         1.4650e-02, -5.2064e-02, -2.7945e-02, -5.3728e-02, -7.0902e-03,
         6.7107e-03,  7.0863e-05, -4.4246e-02,  1.5131e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0262,  0.0520, -0.0214, -0.0876, -0.0073,  0.0017,  0.0118, -0.0073,
         0.0311, -0.0114, -0.0347, -0.0003,  0.0112, -0.0126,  0.0193,  0.0079,
        -0.0029,  0.0176, -0.0388, -0.2090, -0.2435, -0.0316,  0.0075, -0.0960,
        -0.0078,  0.0015,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0149, -0.4219, -0.0119, -0.0815,  0.0263,  0.0098, -0.0175, -0.0507,
        -0.0156,  0.0140, -0.0021,  0.0041, -0.0063, -0.0046, -0.0332, -0.0490,
        -0.1137,  0.0028, -0.0187, -0.0071, -0.0382, -0.0233, -0.0328,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0409, -0.1876, -0.0296, -0.0456, -0.0110,  0.0020, -0.0030, -0.0250,
        -0.0589, -0.0099,  0.0039,  0.0051,  0.0015, -0.0111, -0.1341,  0.0217,
        -0.0180, -0.0345, -0.0647,  0.0033,  0.0020, -0.0014,  0.0081,  0.0038,
         0.0027, -0.0018, -0.0463, -0.0924, -0.0107, -0.0173,  0.0085, -0.0117,
        -0.0067,  0.0087,  0.0025, -0.0047, -0.0254,  0.0184, -0.0150,  0.0004,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.6466e-02, -3.2181e-02,  4.9818e-03, -3.2668e-02, -6.8988e-02,
        -3.6136e-05, -3.3009e-02, -9.3423e-02,  1.7091e-02, -1.1038e-02,
        -5.6300e-03, -1.0807e-02, -3.9194e-02, -1.0793e-02, -2.5777e-02,
        -5.9353e-02, -4.7281e-02, -8.0235e-03, -2.6344e-03, -4.0873e-02,
        -1.4115e-02, -7.8992e-02, -5.4884e-02,  3.7717e-03, -1.1400e-02,
        -1.1280e-02, -7.9267e-03, -5.2908e-02, -1.6213e-03, -1.0472e-02,
        -4.7166e-02, -3.0535e-03,  3.7125e-02, -8.5039e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #100: [tensor([-5.4004e-02, -3.9483e-02, -2.3336e-02, -1.8820e-02, -7.3976e-02,
         1.1091e-03,  2.1841e-02, -2.5256e-04,  5.4264e-03,  3.9803e-03,
        -5.5138e-03, -9.2247e-03, -4.8019e-03, -1.2848e-02, -1.4255e-01,
         1.7160e-02,  5.8738e-03, -9.7395e-03, -5.2151e-02, -1.1863e-03,
         3.2403e-03, -5.2142e-03, -6.3947e-03,  5.5454e-05, -7.0874e-02,
        -9.6482e-02,  1.4782e-03, -5.3623e-03, -4.6557e-03,  6.8605e-03,
        -3.5081e-03,  9.5082e-03,  1.8824e-03, -6.5581e-03, -4.4246e-02,
         3.3026e-03, -1.0901e-02, -3.3260e-02, -8.4191e-03, -2.8858e-02,
        -3.1290e-02,  4.5208e-03, -3.0821e-02, -1.7040e-02, -2.7914e-03,
        -5.9550e-03,  2.8752e-03,  2.7404e-03, -1.5804e-02, -1.7514e-02,
         3.1370e-03, -1.1173e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0415,  0.0148,  0.0096,  0.0113, -0.0903, -0.0098,  0.0070,  0.0255,
         0.0065,  0.0019,  0.0028,  0.0073, -0.0210, -0.1633, -0.0378, -0.0626,
        -0.1220, -0.0688, -0.0798,  0.0113, -0.0369, -0.0082, -0.0285, -0.0716,
         0.0051, -0.0079,  0.0115,  0.0066, -0.0126,  0.0056, -0.0106,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0689, -0.2101,  0.0205, -0.0112,  0.0046,  0.0212, -0.0284, -0.0079,
        -0.0193,  0.0168, -0.0172, -0.0944, -0.0142,  0.0120, -0.0467, -0.0660,
        -0.0054, -0.0542, -0.0945, -0.0857, -0.0311, -0.0199, -0.0219,  0.0165,
         0.0113,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0399, -0.0096,  0.0368,  0.0353, -0.0435, -0.2597, -0.4871,  0.0018,
        -0.0229, -0.0074,  0.0401,  0.0158,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0101, -0.2804, -0.0594, -0.0639, -0.1021, -0.0059, -0.0136, -0.0045,
        -0.0034, -0.0101, -0.0081, -0.0425, -0.0521, -0.0872, -0.0362, -0.0485,
        -0.0069, -0.0048, -0.0076,  0.0083, -0.0277, -0.0065, -0.0187, -0.0020,
         0.0245, -0.0368,  0.0145, -0.0058, -0.0082,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0543, -0.3958,  0.0401,  0.0602,  0.0260, -0.0867, -0.2220,  0.0435,
        -0.0159,  0.0210, -0.0257,  0.0088,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0207, -0.3241, -0.1215, -0.1060,  0.0063, -0.0945, -0.1194, -0.0100,
        -0.1602, -0.0248,  0.0007, -0.0118,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0165,  0.0553, -0.1569,  0.0337, -0.1062,  0.0341, -0.0211, -0.0160,
        -0.1518, -0.0187,  0.0093, -0.1665,  0.0028, -0.0183, -0.0117, -0.1153,
        -0.0047,  0.0253, -0.0112, -0.0246,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0273, -0.2294, -0.0348, -0.0531, -0.0548, -0.0011, -0.0335,  0.0023,
        -0.0200,  0.0022, -0.0206, -0.0685, -0.0355, -0.0172, -0.0068,  0.0047,
        -0.0028,  0.0016, -0.0010, -0.0260, -0.0489,  0.0284,  0.0145, -0.0244,
        -0.0015,  0.0194, -0.0005, -0.0320, -0.0231,  0.0046,  0.0117, -0.0048,
        -0.0288, -0.0870, -0.0185, -0.0063, -0.0024,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.8474e-02,  7.5033e-01,  2.2189e-02, -4.7244e-02,  2.0926e-02,
        -2.5983e-02,  5.7594e-04,  1.5905e-02,  3.8373e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0099,  0.0493,  0.0059, -0.1388, -0.0078, -0.0121, -0.1186, -0.1104,
        -0.0182,  0.0102,  0.0530,  0.0064, -0.0889, -0.0845,  0.0174,  0.0393,
        -0.0035,  0.0416, -0.0423, -0.0719, -0.0056,  0.0137, -0.0061,  0.0157,
        -0.0074, -0.0045, -0.0167,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0285, -0.3037,  0.0316, -0.0695,  0.0045, -0.0456, -0.0065,  0.0047,
         0.0035,  0.0115,  0.0138, -0.0771, -0.0103,  0.0019,  0.0014,  0.0022,
        -0.0170,  0.0104, -0.0015,  0.0173,  0.0137, -0.0539, -0.0856,  0.0126,
        -0.0059, -0.0095, -0.0062, -0.0367, -0.0608,  0.0058,  0.0007, -0.0104,
        -0.0073, -0.0286,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-4.0873e-02, -2.3310e-03, -2.5357e-03,  3.8811e-03, -4.7536e-02,
        -2.3136e-02,  8.4289e-03,  5.4582e-03,  2.7531e-05,  3.8952e-03,
        -8.9444e-03, -5.4227e-02,  9.0024e-03,  1.1640e-02,  3.3391e-02,
        -7.2039e-02, -1.2936e-01,  1.4090e-02, -3.4443e-02,  7.9836e-04,
         8.1152e-03, -2.4776e-03, -2.1189e-02, -2.0331e-03,  1.7185e-02,
         1.8649e-03, -1.1116e-03,  6.2696e-03, -1.3875e-03,  3.6062e-03,
         6.2375e-03, -9.2831e-02, -9.4706e-02, -2.0178e-02, -3.7378e-02,
        -4.7866e-02, -4.4224e-02, -8.1049e-03,  3.9257e-03, -1.1987e-03,
        -1.3775e-02, -3.3466e-02, -5.8807e-03,  4.5896e-03,  1.4366e-02,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0871,  0.0074, -0.0457,  0.0285, -0.0644, -0.0545, -0.1871,  0.0189,
         0.0612, -0.1338,  0.0907, -0.0270, -0.0804, -0.0133, -0.0330, -0.0669,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.0878e-02, -2.4113e-02,  6.7196e-03,  4.8906e-03,  5.0571e-03,
         2.1757e-02, -4.8056e-03,  5.5892e-02,  1.7873e-01,  1.7600e-01,
         6.4553e-02,  1.0993e-01, -2.5284e-02,  8.0585e-02,  8.6801e-02,
        -7.5334e-03,  9.7834e-03,  7.8225e-03, -3.8248e-03,  1.1740e-02,
        -2.5245e-03,  9.6475e-04,  1.7908e-03, -2.2175e-02, -2.9530e-03,
         1.3941e-04, -4.2757e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0348e-02,  3.6146e-03, -5.9293e-02, -1.2465e-02, -3.8615e-02,
         4.1131e-03, -2.2386e-02, -9.5895e-03, -2.0619e-02, -7.8233e-04,
        -2.3482e-03, -3.1184e-03, -2.5228e-02, -8.1480e-03, -4.1180e-02,
        -9.5963e-03, -3.2837e-02,  9.1041e-05,  3.3728e-03, -5.1313e-02,
        -3.4751e-04, -3.0123e-02, -1.0316e-02, -7.1127e-02,  1.8255e-02,
        -1.3523e-03, -1.5408e-02, -7.5315e-02,  5.3790e-05, -4.5903e-02,
         2.7388e-03, -3.6167e-02, -5.4725e-02,  5.5710e-03, -1.8504e-02,
        -5.3180e-02, -3.1898e-02, -6.3708e-02,  3.2535e-04,  4.1850e-03,
        -1.9487e-02, -3.4595e-02, -6.5830e-03, -1.2460e-02,  1.4748e-02,
        -1.3865e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0297,  0.0375, -0.0240,  0.0267, -0.0083,  0.0104, -0.0125, -0.2098,
         0.0048, -0.0249, -0.0281,  0.0323,  0.0086, -0.0140, -0.0189,  0.0060,
        -0.0078, -0.0282, -0.0947, -0.1972,  0.0012, -0.0004, -0.0075, -0.0174,
         0.0031, -0.0078, -0.0639, -0.0324,  0.0004,  0.0414,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0379,  0.0256,  0.0262,  0.0167, -0.0407,  0.0356,  0.1142,  0.5636,
         0.1145,  0.0095, -0.0156,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0031, -0.3475, -0.0145,  0.0304, -0.0772, -0.0065, -0.0654, -0.1205,
         0.0034, -0.0061, -0.0137,  0.0025, -0.0044, -0.0186,  0.0331, -0.1198,
        -0.0455, -0.0233,  0.0084, -0.0564,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1783, -0.3530, -0.1368, -0.0873,  0.0924, -0.0504, -0.0703, -0.0108,
        -0.0146, -0.0026, -0.0035,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0024, -0.2955, -0.1173, -0.1193, -0.0460, -0.0497, -0.0144, -0.0518,
        -0.0453,  0.0102, -0.0096, -0.0013, -0.0172, -0.0236, -0.0080, -0.0429,
         0.0183, -0.0036, -0.0069,  0.0051,  0.1114,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0682, -0.2949, -0.1102, -0.0220,  0.0303, -0.2370, -0.0269, -0.0213,
        -0.0348, -0.0027,  0.0021, -0.0484, -0.0168, -0.0165, -0.0164,  0.0009,
         0.0136,  0.0370,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0314, -0.0985, -0.1768, -0.0124, -0.0140,  0.0107, -0.0033,  0.0063,
         0.0004, -0.0035, -0.0005,  0.0028, -0.0156, -0.0563, -0.0434, -0.0182,
        -0.0254, -0.0410, -0.0311, -0.0066, -0.0076, -0.0365, -0.0859, -0.0088,
        -0.0351, -0.0015, -0.0530, -0.0196, -0.0086, -0.0675,  0.0004, -0.0052,
        -0.0086,  0.0037, -0.0135,  0.0013,  0.0075,  0.0027,  0.0347,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0559, -0.3211, -0.0142, -0.0654, -0.0078, -0.0098, -0.0054, -0.0175,
        -0.0148, -0.0131, -0.0232, -0.0585, -0.0098, -0.0086, -0.0310,  0.0024,
        -0.0373, -0.0057,  0.0214, -0.0207, -0.0154, -0.0054,  0.0045, -0.0037,
        -0.0069, -0.0010, -0.0296, -0.0096, -0.0426, -0.0700, -0.0015, -0.0046,
         0.0169,  0.0447,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0252,  0.0703, -0.1982, -0.1866, -0.0193, -0.2429,  0.1314, -0.0249,
         0.1014,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.2209e-03, -2.3232e-01, -8.7582e-03, -4.6155e-02, -1.6985e-02,
        -6.4131e-02,  1.7751e-04, -2.7601e-02, -1.6721e-02,  1.2109e-03,
        -2.7586e-02, -1.7544e-02,  4.4318e-03, -1.3880e-02,  5.9073e-03,
        -7.9724e-02, -1.6461e-02, -1.0564e-01, -9.3079e-03, -1.5427e-02,
         5.3806e-03,  1.1845e-02, -5.4080e-03,  1.4486e-02, -2.1397e-02,
         6.7535e-03, -8.2969e-03,  8.0277e-03,  7.4378e-04,  4.7826e-03,
         1.6100e-03, -3.8994e-02, -2.8236e-03, -2.8359e-02,  7.4917e-03,
        -4.6345e-03,  9.9178e-03, -1.2678e-02,  2.7151e-02, -6.1030e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.7810e-02, -8.2458e-02, -1.5857e-01, -5.4530e-03, -4.3753e-02,
        -1.5312e-02, -3.6515e-03, -2.3185e-02, -5.4585e-02, -7.7211e-02,
        -8.6228e-04, -3.7751e-04, -1.4146e-03, -1.9877e-03, -1.5323e-02,
         9.0624e-03,  6.5857e-03,  4.0935e-03, -5.2636e-02,  1.7171e-02,
        -3.5880e-02, -1.0603e-02,  1.2837e-02, -6.3012e-03,  6.0971e-03,
        -1.0750e-02, -2.6735e-02, -4.7728e-06, -4.2544e-02, -7.3231e-02,
        -1.5919e-03, -1.7905e-03, -1.1073e-03, -1.3404e-03,  3.9591e-03,
        -2.3510e-03, -3.9991e-02, -5.6153e-02,  3.3000e-03,  1.2466e-04,
         1.4037e-02, -1.2720e-03,  7.4641e-03,  6.2994e-03,  2.5528e-04,
         2.6120e-03, -3.3766e-03,  3.6940e-03, -6.4913e-04,  8.2819e-04,
         7.9214e-03, -1.3400e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0040, -0.2381, -0.0731, -0.0603, -0.0229, -0.0103, -0.0920, -0.0747,
        -0.0478, -0.0060,  0.0413, -0.0271, -0.0666, -0.0178, -0.0221, -0.0459,
        -0.0493,  0.0021, -0.0369,  0.0116,  0.0029,  0.0471,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0244,  0.4499, -0.0685, -0.0081, -0.0070, -0.0189,  0.0165,  0.0790,
        -0.0155,  0.0222,  0.0310,  0.0404,  0.0205,  0.0109,  0.0113,  0.1101,
         0.0227, -0.0116,  0.0007, -0.0288, -0.0020,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0075, -0.2184, -0.0358, -0.0145, -0.0149,  0.0042, -0.0310, -0.0003,
        -0.0317,  0.0035,  0.0046, -0.0024, -0.0022, -0.0010,  0.0055, -0.0060,
         0.0012,  0.0019, -0.0324, -0.0016,  0.0061,  0.0014,  0.0014,  0.0114,
         0.0030, -0.0017, -0.0101, -0.0328, -0.0058, -0.0298, -0.0953, -0.0491,
        -0.0030, -0.0352, -0.0455, -0.0112,  0.0012,  0.0132,  0.0041, -0.0069,
        -0.0668,  0.0014,  0.0099,  0.0020, -0.0379,  0.0023, -0.0282, -0.0116,
        -0.0053,  0.0116, -0.0211,  0.0035, -0.0025,  0.0066,  0.0004],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0081,  0.0043,  0.0045, -0.0062, -0.0082, -0.0270, -0.0659, -0.0203,
        -0.0156,  0.0012, -0.0076,  0.0012,  0.0020, -0.0283, -0.0563, -0.0107,
        -0.0101, -0.0282, -0.0135, -0.0784, -0.0093, -0.0075, -0.0320,  0.0093,
         0.0189, -0.0051,  0.0060, -0.0886, -0.0004,  0.0064,  0.0137,  0.0199,
        -0.0176, -0.1191, -0.0112, -0.0269, -0.0005, -0.0453, -0.0528, -0.0687,
         0.0008, -0.0056,  0.0038, -0.0042, -0.0288,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0041, -0.2481, -0.0778, -0.1127,  0.0260, -0.0144, -0.0088, -0.0094,
        -0.0482, -0.0107,  0.0129, -0.0232,  0.0383, -0.0586, -0.0413, -0.0282,
        -0.0115, -0.0737,  0.0008, -0.0242, -0.0781,  0.0340,  0.0120, -0.0029,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0195,  0.0195, -0.0028,  0.0198, -0.0114,  0.0037, -0.0514, -0.0393,
         0.0100,  0.0051,  0.0025,  0.0038, -0.0076, -0.0502,  0.0103,  0.0249,
        -0.0391, -0.0719, -0.0689, -0.0909, -0.0298, -0.0045, -0.0442, -0.0602,
        -0.0146, -0.0404, -0.0564,  0.0047, -0.0286, -0.0513, -0.0115, -0.0329,
        -0.0681,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0272, -0.4460, -0.0326, -0.0579,  0.0026, -0.0090,  0.0016, -0.0334,
        -0.1323, -0.0155,  0.0211, -0.0059,  0.0204, -0.0701,  0.0199, -0.0483,
        -0.0444,  0.0092,  0.0026,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.1975e-02, -7.5179e-03,  1.7914e-02,  7.1740e-03, -1.1990e-01,
        -1.7635e-02, -1.4706e-01, -6.3709e-02, -1.6282e-01, -2.1259e-02,
         9.0565e-03, -6.1314e-03, -6.1144e-03, -1.7663e-03,  1.0179e-03,
         2.2575e-03, -5.3788e-05, -6.6540e-02, -3.3995e-02, -2.1669e-02,
        -6.9905e-02, -1.1296e-01, -1.1892e-02,  1.1761e-02, -2.7917e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0214, -0.2837, -0.1057,  0.0071,  0.0006, -0.2113, -0.1393,  0.0427,
        -0.0378, -0.0481,  0.0060,  0.0011, -0.0357,  0.0067, -0.0085, -0.0012,
         0.0432,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.0037,  0.0209, -0.0186,  0.0019, -0.0131,  0.0080, -0.0050,  0.0086,
         0.0042, -0.0703, -0.0195,  0.0071,  0.0200,  0.0316,  0.0102,  0.0218,
        -0.0888, -0.1444,  0.0146, -0.0424, -0.0228, -0.1210, -0.2026, -0.0185,
        -0.0075,  0.0117,  0.0127,  0.0020, -0.0111, -0.0108,  0.0003,  0.0245,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0301, -0.1269, -0.1811, -0.0050, -0.0070, -0.0027, -0.0415, -0.0952,
        -0.0123,  0.0099,  0.0134,  0.0258,  0.0045, -0.0051, -0.0197, -0.0200,
        -0.0154, -0.0027, -0.0623, -0.0072, -0.0379, -0.0092, -0.0854, -0.1195,
        -0.0083, -0.0044, -0.0124, -0.0348,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0533,  0.0931,  0.0042, -0.0277,  0.0156,  0.1042,  0.1681,  0.0607,
         0.0240,  0.0610, -0.0039,  0.0793,  0.0051,  0.0167,  0.0229,  0.0711,
         0.0014,  0.0450, -0.0003, -0.0080,  0.0233, -0.0051,  0.0035,  0.0066,
         0.0577, -0.0039, -0.0139,  0.0097,  0.0111,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.7614e-02, -8.1018e-02, -3.3224e-02, -6.9980e-02, -1.4767e-01,
        -7.3939e-03,  8.2792e-03, -1.5817e-03, -1.5892e-02,  7.8476e-03,
         3.2640e-02,  1.7957e-03, -7.9990e-03, -1.5992e-01,  6.0070e-02,
        -3.7358e-02,  4.8899e-03, -3.2332e-02, -9.5364e-03, -3.5519e-03,
        -3.6596e-05,  7.9135e-03, -1.5251e-02, -3.8366e-03, -3.6842e-02,
        -4.5491e-02, -3.8780e-02, -1.3571e-03, -4.5263e-03, -4.5424e-03,
        -3.6077e-03,  3.6238e-02, -5.0989e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0264,  0.3006,  0.0069,  0.0805, -0.0041,  0.0431, -0.0074,  0.0343,
         0.0231,  0.0711,  0.0781,  0.0124, -0.0037, -0.0061,  0.0037, -0.0122,
         0.0365, -0.0072,  0.0960,  0.0355,  0.0785,  0.0026, -0.0168,  0.0133,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0018, -0.2179, -0.1471, -0.0047, -0.0648, -0.0300, -0.0362, -0.0711,
         0.0175, -0.0307,  0.0035, -0.0443, -0.0637, -0.0151, -0.0063, -0.0062,
        -0.0143, -0.0303, -0.0062, -0.1200, -0.0109,  0.0096, -0.0478,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0339, -0.2245, -0.0007,  0.0017, -0.0243, -0.0183,  0.0119, -0.0035,
        -0.0113, -0.0060,  0.0087, -0.0039, -0.0147,  0.0029,  0.0004,  0.0030,
        -0.0863, -0.0537, -0.0301, -0.0080, -0.0043, -0.0331, -0.0077, -0.0143,
        -0.0618, -0.0217, -0.0010, -0.0066, -0.0189, -0.0508, -0.0630, -0.0098,
        -0.0426, -0.0093, -0.0187, -0.0633, -0.0102,  0.0133, -0.0017,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0053, -0.0669, -0.0007, -0.0348, -0.0928, -0.0055,  0.0055,  0.0087,
         0.0006,  0.0164,  0.0046,  0.0094, -0.0112,  0.0088,  0.0142, -0.0569,
        -0.0249,  0.0014,  0.0077,  0.0051, -0.0609, -0.0090, -0.0077, -0.0013,
        -0.0576, -0.1436, -0.0089, -0.0233, -0.0036, -0.0794, -0.0318, -0.0205,
        -0.0020, -0.0099, -0.0102, -0.0131, -0.0181, -0.0763,  0.0100,  0.0073,
        -0.0123, -0.0036, -0.0083,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0113, -0.2112, -0.0103, -0.0017,  0.0161, -0.0008, -0.0175,  0.0041,
        -0.0657,  0.0007,  0.0075, -0.0035, -0.0292,  0.0014, -0.0027, -0.0016,
        -0.0320, -0.0703,  0.0007, -0.0115,  0.0061,  0.0044, -0.0251, -0.0002,
         0.0031, -0.0042,  0.0128,  0.0021,  0.0066, -0.0033,  0.0114, -0.0545,
        -0.1064, -0.0195, -0.0036,  0.0054,  0.0149, -0.0009, -0.0377, -0.0041,
        -0.0375, -0.0480, -0.0070,  0.0118, -0.0025, -0.0123, -0.0019,  0.0008,
         0.0022,  0.0066,  0.0121,  0.0038,  0.0014,  0.0036, -0.0031, -0.0142,
        -0.0048], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0342,  0.4082,  0.0475,  0.0410,  0.0378, -0.0267,  0.0875,  0.1809,
        -0.0068,  0.0101,  0.0526, -0.0083, -0.0006, -0.0578,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0732, -0.1447, -0.0258,  0.0252,  0.0116, -0.0391, -0.0178,  0.0087,
        -0.0227, -0.0031, -0.0432, -0.1334,  0.0156, -0.0114, -0.0304, -0.0149,
        -0.0406, -0.0166,  0.0123, -0.0742, -0.1005, -0.0077, -0.0265, -0.0116,
         0.0208,  0.0151,  0.0030, -0.0074,  0.0146, -0.0282,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1283, -0.2005,  0.0480, -0.3648, -0.0690,  0.0840, -0.0126,  0.0929,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 9.8467e-03, -2.0419e-01, -2.3310e-02, -2.0197e-02, -4.4228e-02,
        -5.4529e-02,  3.0257e-02, -5.9194e-04, -1.2017e-02, -1.5138e-02,
         5.8587e-05,  6.2551e-03, -2.6075e-02, -8.1310e-02, -7.7192e-03,
        -1.4481e-02, -2.1016e-03, -3.6134e-03, -2.4832e-03, -2.6650e-03,
         2.3233e-03, -1.2908e-03, -1.1725e-02, -2.0559e-02, -4.2578e-02,
        -1.5625e-02, -8.9488e-02, -1.7476e-02, -5.5915e-02, -9.3053e-02,
        -1.9679e-02, -2.1753e-02, -4.7469e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0287, -0.1664, -0.0819, -0.0311, -0.0178, -0.0294, -0.0277,  0.0024,
         0.0111, -0.0415,  0.0015, -0.0070, -0.0057,  0.0159,  0.0004,  0.0046,
         0.0027,  0.0024, -0.0137,  0.0056, -0.0061,  0.0025, -0.0012,  0.0023,
         0.0096, -0.0011, -0.0081, -0.0423,  0.0010,  0.0020, -0.0003, -0.0053,
        -0.0291, -0.0544,  0.0012, -0.0047, -0.0281, -0.0035,  0.0006,  0.0088,
         0.0010, -0.0104, -0.0048, -0.0534, -0.0059, -0.0065, -0.0036, -0.0024,
         0.0020, -0.0009, -0.0105, -0.0155, -0.0091, -0.0176, -0.0012,  0.0026,
        -0.0038,  0.0045,  0.0005, -0.0389, -0.0032, -0.0056, -0.0343,  0.0002,
         0.0020, -0.0059, -0.0083,  0.0043,  0.0124, -0.0013, -0.0179],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0096, -0.0264, -0.0328, -0.0111, -0.0645, -0.0197, -0.1241, -0.0144,
         0.0029,  0.0115,  0.0028,  0.0091,  0.0027,  0.0019,  0.0064, -0.0059,
        -0.0019,  0.0168, -0.0166, -0.1233, -0.0308,  0.0139,  0.0105, -0.0564,
        -0.0060, -0.0386, -0.0069,  0.0155,  0.0115, -0.0117,  0.0028, -0.0086,
        -0.0671, -0.0680, -0.0294, -0.0039, -0.0675, -0.0379, -0.0062,  0.0002,
        -0.0014,  0.0009,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0024, -0.2393, -0.0088, -0.0172,  0.0011,  0.0089,  0.0113,  0.0057,
        -0.0186, -0.0564, -0.0325, -0.0310, -0.0444, -0.0018, -0.0015,  0.0087,
        -0.0075, -0.0537, -0.1012, -0.0166, -0.0216, -0.0116,  0.0019, -0.0033,
        -0.0117, -0.0520, -0.0126, -0.0079, -0.1089, -0.0092,  0.0102, -0.0046,
        -0.0011, -0.0181, -0.0076, -0.0298,  0.0196,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1113, -0.4681, -0.0557, -0.0534,  0.0064,  0.0298, -0.0196, -0.0132,
         0.0049,  0.0229, -0.0283,  0.0292, -0.0548, -0.0357,  0.0665,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0107,  0.0559,  0.0060, -0.0202,  0.1736,  0.1975,  0.0041,  0.0757,
         0.1462,  0.0244, -0.0109, -0.0142,  0.0586, -0.0121, -0.0487,  0.0244,
         0.0137,  0.0268, -0.0023, -0.0738,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0050,  0.0115, -0.0175,  0.0180, -0.1387, -0.0243,  0.0124, -0.0137,
        -0.0009, -0.0126, -0.0397, -0.0624, -0.0146, -0.0092, -0.0441, -0.0218,
        -0.0010, -0.0284,  0.0174, -0.0346, -0.0090, -0.0019,  0.0039,  0.0046,
        -0.0075, -0.0454, -0.0571, -0.0097,  0.0056, -0.0507, -0.0663, -0.0175,
         0.0012, -0.0063, -0.0296, -0.0521,  0.0132, -0.0295, -0.0088, -0.0167,
        -0.0151, -0.0125, -0.0078,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0485, -0.0141, -0.0091, -0.0389, -0.0635, -0.0244, -0.0507, -0.0861,
        -0.0289, -0.0113,  0.0138,  0.0106,  0.0034, -0.0504, -0.0660, -0.0082,
         0.0015,  0.0094, -0.0052, -0.0059, -0.0517, -0.0053,  0.0126,  0.0012,
        -0.0144, -0.0079,  0.0043,  0.0052, -0.0675, -0.0005, -0.0092, -0.0432,
        -0.0184, -0.0554, -0.0808, -0.0436,  0.0007, -0.0280,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.8211e-02, -2.4844e-01, -1.0023e-01, -1.1738e-01,  2.8280e-02,
        -7.5649e-02, -8.8406e-03,  1.1415e-05, -4.7947e-02, -1.1509e-01,
        -9.7251e-03,  2.8085e-03, -7.4570e-02,  7.9898e-04, -6.5973e-03,
         2.8163e-03, -3.7383e-03, -4.5108e-02, -4.9833e-03,  2.3455e-02,
         4.4650e-02, -2.0670e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0029, -0.1618, -0.0232,  0.0123,  0.0213, -0.0130, -0.0017,  0.0121,
        -0.0033, -0.1031, -0.0055,  0.0077, -0.0067,  0.0070, -0.0371, -0.0341,
        -0.0921, -0.0091, -0.0042, -0.0660,  0.0068, -0.0125, -0.0466, -0.0116,
        -0.0294, -0.0037, -0.0038,  0.0183, -0.0219, -0.0546, -0.0087, -0.0189,
        -0.0038, -0.0356, -0.0360, -0.0350,  0.0190, -0.0008,  0.0029, -0.0061,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.4867e-02, -3.0007e-03, -1.0028e-02,  2.0582e-02, -2.1324e-02,
        -2.7338e-02, -1.5917e-02, -4.0835e-03, -5.2093e-02, -1.5291e-02,
        -1.0775e-02, -1.2794e-02, -1.4871e-02, -1.1774e-01, -9.4343e-02,
        -3.5310e-03, -1.2606e-02,  2.0625e-03, -7.7508e-02,  2.6564e-05,
         2.4770e-02,  2.7997e-03,  9.1319e-03, -8.2791e-03,  2.8081e-03,
        -7.6964e-02, -1.1740e-02, -6.9870e-02, -1.2568e-01,  1.3214e-02,
        -1.1355e-02, -3.2612e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.3625e-02, -3.6347e-01,  8.3910e-03,  1.8247e-02, -5.6554e-03,
        -8.0069e-03, -5.3899e-04, -3.3584e-02, -8.0307e-02, -2.5775e-02,
        -5.1754e-03, -3.2561e-02,  3.5964e-05,  2.6241e-03, -1.8470e-03,
        -2.1707e-02, -5.0811e-02, -9.9982e-03,  1.1356e-03, -2.1618e-03,
        -9.5777e-03,  6.9568e-03, -7.4820e-03, -7.9135e-02, -1.7158e-02,
         1.3693e-02, -6.0150e-02, -8.8207e-03,  9.9194e-03, -3.0538e-02,
         1.0053e-02, -1.3811e-02,  1.8190e-03,  1.8143e-02,  7.0881e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0621, -0.0463, -0.1363, -0.1208, -0.0488, -0.0189, -0.1186, -0.1319,
        -0.0133, -0.1133, -0.0075, -0.0329, -0.0027, -0.0452, -0.0161, -0.0130,
        -0.0389, -0.0079, -0.0070,  0.0074,  0.0109,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0529, -0.4256,  0.0060, -0.0670, -0.0893, -0.0523,  0.0054, -0.0485,
        -0.0729,  0.0057, -0.0033, -0.0124,  0.0045,  0.0073, -0.0391, -0.0078,
        -0.0529,  0.0268, -0.0204,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0390, -0.0156, -0.1487,  0.0400, -0.0412, -0.0210,  0.0060, -0.0175,
         0.0083, -0.0679, -0.1511, -0.0561, -0.0631, -0.2000,  0.0187, -0.0286,
         0.0048, -0.0359,  0.0159, -0.0206,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0272, -0.3920,  0.0505, -0.0877,  0.0309, -0.0033, -0.0498, -0.0275,
        -0.0030,  0.0100, -0.0093, -0.0076,  0.0057, -0.0392,  0.0348, -0.0332,
        -0.0554, -0.0085, -0.0183, -0.0436,  0.0130,  0.0174, -0.0321,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.5753e-04, -2.1785e-01, -3.9378e-02, -1.2985e-02, -1.4498e-02,
        -3.3635e-02,  2.3249e-02, -5.5370e-02, -8.1688e-02, -1.1670e-01,
        -2.8568e-02, -9.2285e-02, -1.5960e-02, -7.9154e-02, -7.5976e-03,
         2.2917e-03,  2.4017e-02,  1.1185e-02, -1.0167e-02,  1.3674e-02,
        -4.1649e-02, -1.7591e-02,  1.1452e-02, -1.0999e-02,  3.7894e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0331,  0.3788,  0.0226, -0.0040, -0.0048,  0.0274,  0.0086,  0.0102,
        -0.0176,  0.0444,  0.0670, -0.0013,  0.0014, -0.0107, -0.0029,  0.0263,
         0.0129, -0.0023, -0.0025, -0.0004,  0.0472,  0.0816,  0.0293,  0.0210,
         0.0159,  0.0135,  0.0213,  0.0124,  0.0025,  0.0164,  0.0082,  0.0009,
         0.0189,  0.0146,  0.0173,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0133, -0.0172,  0.0112, -0.1205, -0.0181, -0.0050, -0.0915, -0.0302,
        -0.1278,  0.0042, -0.0072,  0.0029, -0.0350, -0.0475, -0.0717, -0.1506,
        -0.0307, -0.1099, -0.0176, -0.0223, -0.0320,  0.0337,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0468,  0.0247, -0.0324,  0.0175,  0.0520,  0.0731, -0.0079,  0.0040,
        -0.0254,  0.2758,  0.0819,  0.0171, -0.0049,  0.0343,  0.0181,  0.1240,
         0.0448,  0.0429, -0.0345, -0.0162, -0.0219,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0314, -0.4812, -0.0555, -0.0223, -0.0010,  0.0176, -0.0483, -0.1070,
        -0.0026,  0.0210,  0.0072,  0.0125,  0.0049, -0.0969,  0.0078, -0.0549,
         0.0130,  0.0055, -0.0093,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0061,  0.0427, -0.0012,  0.0011, -0.0058, -0.0907,  0.0067, -0.0287,
         0.0006, -0.0100, -0.0024, -0.0247, -0.0080, -0.0274,  0.0090,  0.0025,
        -0.0570, -0.0349,  0.0110, -0.0307, -0.0137, -0.0112, -0.0397,  0.0086,
        -0.0475, -0.0119, -0.0216, -0.0261,  0.0103, -0.1277,  0.0051, -0.0425,
        -0.0447, -0.0855, -0.0029, -0.0096, -0.0038, -0.0081, -0.0063, -0.0355,
         0.0016,  0.0082,  0.0011, -0.0111, -0.0010,  0.0028,  0.0105],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0029,  0.2428,  0.1085,  0.0028, -0.0103, -0.0117,  0.0099,  0.0017,
         0.0366, -0.0090,  0.0025,  0.0002, -0.0006, -0.0137, -0.0037, -0.0109,
         0.0692,  0.0034, -0.0020,  0.0125,  0.0021, -0.0052,  0.1279,  0.1503,
        -0.0249, -0.0142,  0.0399,  0.0018, -0.0112, -0.0053, -0.0059, -0.0150,
         0.0090, -0.0322,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.1837e-02, -3.4810e-01, -1.7671e-02,  1.4638e-02, -2.2821e-02,
        -2.6222e-02,  6.9065e-03, -2.2017e-03, -6.9335e-03, -1.7524e-02,
         9.4621e-03, -1.2848e-03, -4.4470e-02, -6.9673e-02,  3.2667e-03,
        -4.4626e-03, -3.6372e-03,  3.7436e-05,  2.5517e-02, -9.1575e-02,
        -1.9864e-02, -2.2764e-02, -2.6057e-02, -1.0979e-02, -8.6123e-03,
         8.0233e-03,  6.8425e-03, -1.6022e-02, -2.1047e-03, -5.2961e-02,
         2.5321e-03,  3.7000e-03, -3.3883e-02,  3.7420e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0102,  0.4081,  0.0195,  0.0147,  0.0080,  0.0946,  0.0100,  0.0516,
         0.0260,  0.0983,  0.0159, -0.0100,  0.0203,  0.1261,  0.0262,  0.0237,
        -0.0276,  0.0093,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0700, -0.0051,  0.0126,  0.0163,  0.0048,  0.0041,  0.0115,  0.0096,
        -0.1707, -0.3756,  0.0078, -0.0580, -0.0234, -0.0060,  0.0046,  0.0080,
        -0.0177,  0.0058,  0.0290, -0.0640,  0.0063, -0.0165, -0.0022,  0.0159,
        -0.0024,  0.0154, -0.0367,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0642, -0.0176,  0.0050, -0.0005,  0.0853,  0.0115,  0.0067, -0.0069,
        -0.0934,  0.0122, -0.0071, -0.0616, -0.1485,  0.0033,  0.0137, -0.0033,
        -0.0050, -0.0282, -0.0226, -0.2289, -0.0085, -0.0098, -0.0505, -0.0188,
        -0.0563, -0.0304,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0795, -0.2824, -0.0023, -0.0831, -0.0825,  0.0224,  0.0017, -0.0277,
        -0.0715, -0.0254, -0.0670, -0.0526,  0.0146,  0.0063, -0.0089, -0.0596,
        -0.0305, -0.0188,  0.0050, -0.0117,  0.0464,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0354, -0.2487, -0.1981, -0.1524, -0.0276,  0.0008, -0.1731, -0.0006,
        -0.0364,  0.0252,  0.0056, -0.0454,  0.0378,  0.0055,  0.0073,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0327, -0.0020,  0.0063,  0.0177, -0.0716,  0.0164,  0.0426,  0.0035,
         0.0276,  0.0503, -0.0088,  0.0023, -0.4348, -0.1117,  0.1085, -0.0382,
        -0.0138,  0.0112,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 9.1047e-02, -2.8707e-01, -9.0061e-02, -1.7306e-01,  4.5173e-03,
         1.5894e-02,  1.6083e-02, -2.6146e-02, -2.3772e-02, -5.7544e-02,
        -5.8268e-02, -3.9676e-02,  1.2816e-04, -6.5339e-02, -1.0228e-02,
         1.8962e-03, -1.2300e-03,  3.3654e-02,  4.3893e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.2754e-02, -1.7996e-01, -9.4839e-02, -2.2405e-02, -6.2938e-02,
        -3.0162e-02,  2.6078e-02, -1.2788e-02, -1.3952e-02,  3.5391e-04,
         2.0883e-03, -1.0080e-03, -1.3600e-02, -1.9984e-03,  6.5240e-03,
         1.0200e-02, -6.0880e-02,  4.0166e-03, -8.4861e-04,  3.3186e-03,
        -3.4631e-02, -4.3954e-03, -3.8987e-02, -7.5523e-03, -2.5202e-02,
        -2.6813e-03, -4.7199e-02,  1.5552e-02, -1.0433e-02, -3.3557e-02,
         4.3445e-05,  4.3111e-03, -3.0789e-02, -1.0871e-02, -3.6935e-02,
         1.0618e-03, -3.2904e-02,  9.4582e-03, -1.0291e-02, -5.1923e-03,
        -1.7720e-03, -8.7690e-04,  8.8639e-03, -5.7261e-03], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0094, -0.2119, -0.0183, -0.0721, -0.0736, -0.0202, -0.0415, -0.0061,
        -0.0813, -0.0084, -0.0381,  0.0257,  0.0003,  0.0004, -0.0178,  0.0043,
         0.0094, -0.0570,  0.0185, -0.0059, -0.0299, -0.0018, -0.0121, -0.0045,
         0.0053,  0.0063,  0.0148, -0.0385,  0.0039,  0.0167, -0.0041, -0.0617,
         0.0079, -0.0438,  0.0284,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0398, -0.4384, -0.0329, -0.1099, -0.0136, -0.0136, -0.0785, -0.0415,
        -0.0148, -0.0009,  0.0101, -0.0279, -0.0785, -0.0161,  0.0036, -0.0381,
         0.0061, -0.0032,  0.0153,  0.0026,  0.0082, -0.0038, -0.0008,  0.0019,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.8218e-02,  2.4187e-01,  8.6984e-04,  3.1550e-02,  1.6799e-03,
        -2.5160e-03, -1.3589e-02,  2.6830e-02,  8.2809e-02,  1.8531e-02,
         3.3225e-02, -3.6128e-04, -8.7096e-03,  1.7590e-02,  4.2338e-02,
         2.0369e-03, -3.1668e-03,  3.7295e-03,  5.0250e-02,  2.1157e-04,
         2.0289e-03,  3.8745e-02, -9.7287e-03,  3.2858e-02, -4.0984e-03,
         4.0974e-02,  3.2884e-02,  2.6102e-02, -3.9440e-04,  6.1741e-03,
         2.9541e-02, -7.2177e-03, -3.3374e-02,  1.0580e-01,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0701, -0.0066, -0.0183,  0.0082,  0.0023,  0.0279, -0.1178, -0.0204,
         0.0420,  0.0062, -0.0002,  0.0130, -0.0994, -0.0127, -0.1033, -0.1667,
        -0.0154, -0.0133, -0.0041,  0.0361,  0.0098, -0.0013, -0.0313, -0.0586,
        -0.0226, -0.0110, -0.0222, -0.0100,  0.0313,  0.0178,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0892,  0.0036, -0.0697, -0.2360, -0.0546, -0.0472, -0.0446, -0.0429,
        -0.1203, -0.2107, -0.0045, -0.0364, -0.0405,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0788,  0.0098,  0.0409,  0.0041, -0.0094, -0.0517,  0.0206, -0.1913,
        -0.2443, -0.0031,  0.0059, -0.1335, -0.0130, -0.0164, -0.1171,  0.0246,
         0.0316,  0.0039,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1029, -0.2342, -0.4058, -0.0895, -0.0471, -0.0406, -0.0294, -0.0504,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.5121e-02, -2.3588e-01, -1.6370e-02, -1.6113e-02, -3.3897e-02,
        -2.9050e-02, -2.5246e-02,  2.4589e-03, -6.8685e-03, -1.1085e-02,
         5.7946e-03,  2.6104e-02, -1.6495e-02, -9.4408e-03, -5.1846e-04,
        -7.9979e-03,  5.7090e-03,  3.8070e-03,  6.6086e-03,  6.8639e-04,
         9.3583e-03, -1.8016e-02, -3.7830e-02, -3.2192e-02, -1.9539e-03,
        -1.8747e-02, -7.8257e-03, -4.6331e-03, -6.7731e-02, -3.8224e-02,
        -5.2312e-04, -4.7288e-02, -7.0505e-03, -9.0552e-02,  1.7112e-03,
        -2.9795e-02, -2.7368e-02, -2.6056e-03,  9.1960e-03,  2.6856e-03,
        -4.8569e-03, -1.0032e-02,  1.7558e-04,  6.4890e-03,  1.5763e-02,
        -1.2349e-02,  9.8030e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0530, -0.2634, -0.0374, -0.1927, -0.0623,  0.1549,  0.0486, -0.0102,
         0.0126, -0.1267, -0.0382,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0505,  0.2932,  0.0906,  0.0791,  0.0203, -0.0070,  0.0266,  0.1363,
        -0.0105,  0.0413,  0.0124, -0.0351,  0.0635, -0.0017, -0.0325,  0.0209,
        -0.0080, -0.0704,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0028, -0.0970, -0.2843, -0.0163,  0.0389,  0.0146, -0.2268,  0.0153,
        -0.0528, -0.0122,  0.0286, -0.0065,  0.0096,  0.0594,  0.1349,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.0785e-02, -2.6159e-01,  1.2650e-02, -6.7227e-02,  6.0950e-03,
        -9.6204e-02, -1.6134e-01, -4.6564e-02,  1.1033e-04, -8.3879e-02,
        -8.9558e-02,  1.8941e-03, -6.4564e-03,  1.2225e-02,  7.3428e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0833,  0.0101,  0.0099,  0.0262, -0.0108, -0.0059, -0.0010,  0.0152,
        -0.0374, -0.0257, -0.0222, -0.1044, -0.1304, -0.0209, -0.0023,  0.0119,
        -0.0475,  0.1044, -0.0179,  0.0132, -0.0909, -0.1875, -0.0019,  0.0140,
        -0.0004, -0.0047,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0758, -0.1077, -0.0310,  0.0377,  0.0462, -0.0388, -0.0096, -0.2238,
        -0.2709, -0.0387, -0.0185,  0.0502,  0.0268,  0.0077, -0.0166,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0213, -0.3473,  0.0223, -0.0452, -0.0586, -0.0398, -0.0656,  0.0168,
        -0.0122, -0.0823,  0.0161,  0.0039, -0.0330, -0.0070, -0.0359,  0.0052,
        -0.0244,  0.0218, -0.0459,  0.0241, -0.0188,  0.0185, -0.0020,  0.0085,
        -0.0100, -0.0133,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0671, -0.0165,  0.0088,  0.0024, -0.0014, -0.0258, -0.0017,  0.0074,
        -0.0758, -0.0767, -0.1515, -0.0129, -0.0138,  0.0330, -0.1574, -0.0912,
         0.0332, -0.0358,  0.0046, -0.0667, -0.0127, -0.0331,  0.0023, -0.0457,
        -0.0041, -0.0184,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-1.2939e-02,  3.1819e-01, -2.3056e-02, -1.4406e-03,  1.2950e-02,
        -1.3718e-03,  6.2066e-02,  7.2547e-02,  2.0675e-03,  3.3274e-02,
         1.5063e-02,  1.3487e-02,  5.1826e-02, -9.7727e-03,  3.1109e-02,
         6.5861e-03, -8.9031e-03, -3.5188e-03,  1.2234e-02, -3.0660e-03,
        -6.0056e-03,  1.9827e-02, -7.7497e-03,  3.6682e-02,  9.9497e-02,
        -1.8481e-02,  2.1756e-02,  1.4278e-03, -2.2183e-04, -6.8044e-03,
         5.1122e-04,  5.9293e-02,  1.2129e-03,  5.8627e-03, -8.3918e-03,
        -1.0810e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0408, -0.0879, -0.0554, -0.0138, -0.0547,  0.0023, -0.0681, -0.0880,
         0.0003,  0.0133, -0.0229, -0.0021, -0.0041, -0.0017, -0.0076, -0.0184,
         0.0117, -0.0043,  0.0199,  0.0049, -0.0107, -0.0450, -0.0091, -0.0191,
        -0.0416, -0.0655,  0.0101, -0.0329, -0.0594, -0.0166, -0.0150, -0.0680,
        -0.0533, -0.0008,  0.0025, -0.0281,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0601,  0.0115,  0.0619, -0.0483,  0.0010, -0.2262, -0.0233, -0.0081,
         0.0113,  0.0135, -0.0345, -0.0834, -0.2171,  0.1223,  0.0175,  0.0075,
         0.0007,  0.0122, -0.0171,  0.0224,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0474, -0.1512, -0.0390, -0.0864, -0.0173,  0.0173, -0.0038, -0.0151,
         0.0035, -0.0482, -0.0136, -0.0048, -0.0065, -0.0084, -0.0695, -0.0510,
        -0.0081, -0.0308,  0.0038,  0.0018, -0.0093, -0.0046,  0.0056,  0.0210,
        -0.0066, -0.0088, -0.0038,  0.0111, -0.0115, -0.0622, -0.0537, -0.0199,
         0.0056, -0.0666, -0.0370, -0.0327, -0.0054,  0.0070,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0232, -0.2499, -0.1596, -0.0750, -0.0132, -0.0053, -0.0159,  0.0163,
         0.0064,  0.0044, -0.0226, -0.0539,  0.0044,  0.0071, -0.0034, -0.0451,
        -0.0049, -0.0395, -0.0005,  0.0071,  0.0030,  0.0093,  0.0063,  0.0026,
         0.0069, -0.0091, -0.0137, -0.0259, -0.0497,  0.0039, -0.0093, -0.0139,
         0.0132,  0.0018, -0.0406, -0.0023, -0.0109, -0.0059, -0.0042, -0.0063,
        -0.0030], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0355, -0.4431,  0.0142, -0.0329,  0.0112, -0.0802, -0.0191, -0.0089,
        -0.0095,  0.0025,  0.0004, -0.0744, -0.0882,  0.0043,  0.0010, -0.0646,
        -0.0043, -0.0645, -0.0011, -0.0096,  0.0143, -0.0163,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.9070e-02, -2.8369e-03,  3.7351e-03,  4.2693e-03,  2.9274e-02,
        -2.8302e-02,  2.4046e-03,  1.3994e-02,  6.7538e-04,  3.3393e-04,
         1.3971e-02, -6.7190e-02,  6.1767e-03, -3.4842e-02,  6.9052e-03,
        -9.8377e-02,  4.2781e-03, -5.7015e-02,  2.9526e-03, -7.5316e-02,
        -1.2173e-01, -8.0571e-03, -1.5734e-02, -7.2076e-02,  2.1218e-03,
        -2.9068e-03, -7.7554e-03, -3.3107e-02, -1.1004e-01,  6.7449e-03,
        -5.4036e-02, -6.5600e-02, -8.3538e-06,  6.3305e-03,  1.1832e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1049, -0.0440, -0.2364, -0.0170, -0.0622, -0.0350,  0.0102, -0.0192,
         0.0098,  0.0084, -0.0442,  0.0079, -0.0070,  0.0207, -0.0025, -0.0348,
         0.0144, -0.0293, -0.0353,  0.0085, -0.0216,  0.0224, -0.0055, -0.0562,
         0.0061,  0.0084, -0.0643, -0.0080,  0.0558,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.8039e-02, -4.1760e-01, -6.4845e-02, -4.5252e-02, -1.4924e-02,
        -6.3944e-02, -1.6398e-02, -1.0483e-02, -4.1018e-02,  4.3112e-03,
        -3.7862e-02, -7.3185e-02,  1.0975e-02, -3.6886e-03, -6.1835e-03,
        -4.1639e-02, -3.3392e-02, -1.0840e-04, -2.1028e-02,  1.8106e-02,
         7.0203e-03, -2.5089e-03,  2.7814e-03,  2.7357e-02,  1.7353e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0139,  0.0074, -0.0277,  0.0079, -0.2379, -0.0036, -0.0178, -0.1290,
        -0.1614,  0.0099, -0.0263,  0.0234, -0.0123,  0.0011, -0.0755, -0.1012,
         0.0152, -0.0387,  0.0068,  0.0095, -0.0068, -0.0208, -0.0457,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0143, -0.0081,  0.0018,  0.0037, -0.0033,  0.0216, -0.0007,  0.0088,
         0.0019, -0.0029,  0.0100, -0.0067, -0.0500, -0.1545,  0.0133, -0.0150,
         0.0032, -0.0420, -0.0044,  0.0131, -0.0220, -0.0039,  0.0158, -0.0172,
        -0.0153,  0.0103, -0.1237,  0.0095, -0.0820, -0.0164, -0.0812, -0.0067,
        -0.0379, -0.0285, -0.0879, -0.0050, -0.0298, -0.0216,  0.0062,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0410,  0.0215,  0.0477,  0.1518,  0.2679, -0.0134, -0.0015,  0.1691,
         0.1816,  0.0220,  0.0337,  0.0489,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-2.4379e-02, -7.6505e-02, -1.9595e-01, -3.0765e-03, -1.0605e-02,
        -6.0865e-03, -2.7010e-02,  2.8125e-02, -8.2959e-03, -4.6713e-02,
        -1.4190e-03,  1.4375e-02, -1.8240e-02,  1.5372e-02, -5.4445e-02,
         1.0026e-02, -2.5576e-02, -4.8476e-02, -7.6938e-04, -6.1665e-03,
        -1.8603e-02, -3.0728e-02,  3.1896e-03, -1.7064e-04, -2.0149e-02,
         3.4512e-03, -8.8943e-04, -3.0270e-02,  5.6913e-02, -3.5869e-02,
         7.9758e-03,  7.0847e-03, -4.7092e-02, -3.4288e-04, -3.9163e-04,
        -1.7978e-02, -3.8146e-02, -2.3303e-03,  3.9708e-03, -4.2325e-03,
         1.9981e-03,  3.0461e-03,  1.7713e-03, -3.0948e-02, -1.0853e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0405,  0.5135,  0.0759,  0.1014, -0.0391, -0.0232,  0.0027,  0.0219,
         0.1283,  0.0248,  0.0176, -0.0113,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0611, -0.6757, -0.0287, -0.0809,  0.0068,  0.0751,  0.0041,  0.0676,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.3014e-02, -1.6576e-01, -3.5111e-02,  2.9084e-02, -5.0075e-02,
        -1.4826e-03, -1.0662e-02, -2.5417e-02, -3.3481e-02, -1.0035e-01,
        -2.4490e-02, -9.2389e-03, -3.2860e-03, -6.9080e-04,  4.4159e-04,
         4.4986e-03, -6.3557e-03, -2.0115e-03,  4.9111e-03, -3.2029e-03,
         5.2827e-04,  7.5879e-03, -8.7749e-03,  6.6822e-03, -6.6800e-03,
        -2.1528e-03,  2.5519e-03, -2.3364e-03, -2.0629e-02,  2.4046e-03,
        -5.2166e-03, -8.9637e-03, -2.8335e-03,  3.6711e-03, -1.3592e-03,
        -3.3699e-05, -1.2152e-03,  1.7530e-03,  3.5709e-04,  6.8350e-03,
         5.3669e-03, -2.1738e-03, -2.8251e-04, -3.9019e-03, -3.7489e-03,
        -9.0768e-03, -7.3529e-02, -1.5430e-02, -6.9831e-02, -7.1299e-02,
        -1.7171e-02, -6.8666e-03, -2.7363e-02, -2.7291e-03, -1.1346e-02,
        -1.0029e-03, -1.8651e-02,  4.9720e-03,  2.9041e-03, -1.2799e-03,
         2.4945e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0278, -0.0019, -0.0051, -0.0581, -0.0736, -0.0043, -0.0793, -0.0090,
         0.0063,  0.0181, -0.0116, -0.1002, -0.0069, -0.0546, -0.1066, -0.0117,
        -0.0151,  0.0037, -0.0225, -0.0241, -0.0736, -0.0697,  0.0131, -0.0181,
        -0.0381, -0.0330, -0.0170, -0.0143, -0.0290, -0.0164,  0.0025, -0.0175,
         0.0172,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0322,  0.0644,  0.1601,  0.1125,  0.0293,  0.0229,  0.0611, -0.0365,
        -0.0081,  0.0049,  0.0159, -0.0079,  0.0240,  0.0099,  0.0026, -0.0225,
         0.0056,  0.0760,  0.2197, -0.0112, -0.0052,  0.0464, -0.0149, -0.0063,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0464, -0.1029, -0.0657, -0.0358, -0.0119, -0.1025,  0.0082, -0.0410,
         0.0031, -0.0714, -0.0872, -0.0778, -0.1228, -0.0040,  0.0119,  0.0129,
         0.0174, -0.0233,  0.0161,  0.0394, -0.0185, -0.0219,  0.0358, -0.0220,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1156,  0.0032, -0.0785, -0.0736,  0.0104,  0.0123, -0.0025, -0.0656,
        -0.0009, -0.0140,  0.0094,  0.0083, -0.0035, -0.0139, -0.0501, -0.0843,
        -0.0508,  0.0038, -0.0054, -0.0290, -0.0641, -0.0032, -0.0347, -0.0930,
         0.0006, -0.0042, -0.0291, -0.0227,  0.0023, -0.0390, -0.0181, -0.0218,
         0.0302, -0.0017,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.4031e-02, -2.2318e-03,  3.3188e-02,  1.1274e-04, -3.8444e-02,
         5.4045e-03, -6.3578e-03,  1.9221e-02, -4.1164e-03, -1.2450e-02,
        -4.5257e-02,  2.6642e-02,  2.5516e-02, -1.6270e-01, -3.5541e-02,
        -8.1663e-03, -1.7873e-01, -7.1334e-03, -5.5469e-02,  5.2709e-03,
        -6.6952e-02, -1.8744e-01,  2.5647e-02,  2.4331e-03, -2.1554e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0450,  0.0213, -0.0364, -0.0201, -0.0020,  0.0464, -0.1326, -0.0609,
        -0.1155, -0.2930,  0.0643, -0.0310, -0.0355, -0.0400, -0.0152, -0.0410,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1332, -0.2705,  0.0189, -0.1091, -0.0311, -0.0722, -0.0774, -0.0181,
         0.0064,  0.0054, -0.0080, -0.0344, -0.0372,  0.0066,  0.0249,  0.0019,
        -0.0238, -0.0115, -0.0479, -0.0044,  0.0182, -0.0219,  0.0169,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0534, -0.0041,  0.0426,  0.2060,  0.4052, -0.0442, -0.0537,  0.0135,
         0.0586, -0.0004, -0.0492, -0.0451, -0.0034,  0.0205,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0352, -0.0796, -0.1028, -0.1709, -0.0381,  0.0045, -0.0514,  0.0021,
        -0.0513, -0.0199, -0.0878, -0.0813, -0.0167,  0.0010, -0.0130,  0.0190,
         0.0072, -0.0963,  0.0141, -0.0774, -0.0190,  0.0017,  0.0029, -0.0069,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0241,  0.0264, -0.1497, -0.0029, -0.1798,  0.1210, -0.3640, -0.0700,
         0.0121, -0.0099,  0.0401,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0605,  0.0097,  0.0832,  0.0093,  0.0234, -0.0302, -0.2120, -0.0222,
        -0.1197, -0.0340, -0.0548, -0.0132, -0.0628,  0.0059, -0.0748, -0.0202,
        -0.0313, -0.1230,  0.0038,  0.0041, -0.0021,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0893,  0.0466,  0.0148, -0.0119, -0.0859,  0.0052, -0.0095, -0.3733,
        -0.0570, -0.1945, -0.0230,  0.0222,  0.0019, -0.0649,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2747,  0.0337,  0.0156,  0.0088,  0.0340, -0.0015,  0.0112,  0.0048,
        -0.0029,  0.0589,  0.0016,  0.0494,  0.3392, -0.0161,  0.0209,  0.0026,
         0.0437, -0.0073, -0.0177,  0.0354, -0.0199,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0438,  0.4337, -0.0141,  0.0467, -0.0701,  0.1051,  0.0282,  0.0799,
         0.0061,  0.0475,  0.0027,  0.0286, -0.0936,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0214, -0.1756, -0.0122,  0.0020, -0.0109, -0.0189, -0.0298, -0.1373,
        -0.0252, -0.0228,  0.0003,  0.0134,  0.0122, -0.0047, -0.0025, -0.0145,
         0.0003, -0.0018,  0.0063,  0.0214, -0.0046,  0.0030, -0.0138, -0.0113,
        -0.0650,  0.0246, -0.0718, -0.0399, -0.0405, -0.0147, -0.0039, -0.0016,
        -0.0027,  0.0006, -0.0057,  0.0038, -0.0857, -0.0030, -0.0017, -0.0117,
        -0.0297, -0.0008, -0.0017,  0.0138, -0.0108], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0531, -0.0376, -0.0538, -0.0250,  0.0104,  0.0193, -0.0085, -0.0452,
        -0.0047,  0.0265, -0.0649, -0.0195, -0.0127, -0.0162, -0.0534,  0.0253,
        -0.1124, -0.0029,  0.0163, -0.0026, -0.0074, -0.1069,  0.0089, -0.0580,
         0.0167,  0.0169,  0.0071, -0.0248, -0.0481, -0.0177, -0.0060, -0.0182,
        -0.0464, -0.0066,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0085,  0.0951,  0.0208, -0.0607, -0.0236, -0.0148,  0.0099, -0.0119,
         0.0893,  0.0025, -0.0104,  0.0041, -0.0070, -0.0052, -0.0073,  0.0254,
        -0.0084,  0.0267,  0.0096, -0.2534, -0.2251,  0.0099, -0.0086,  0.0100,
        -0.0106, -0.0416,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0656, -0.3573, -0.0115, -0.0640,  0.0273, -0.0036, -0.0251, -0.0441,
        -0.0138, -0.0211,  0.0143,  0.0164,  0.0029,  0.0157, -0.0153, -0.1196,
        -0.1034,  0.0011, -0.0233, -0.0036, -0.0193, -0.0099,  0.0217,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0013,  0.1457,  0.0092,  0.0562,  0.0036, -0.0025, -0.0048,  0.0516,
         0.0880,  0.0152, -0.0071, -0.0010, -0.0072,  0.0064,  0.1889, -0.0006,
        -0.0166,  0.0351,  0.0499,  0.0065,  0.0099,  0.0031, -0.0056, -0.0254,
         0.0004, -0.0071,  0.0347,  0.0829,  0.0137,  0.0049,  0.0027, -0.0056,
        -0.0111, -0.0184,  0.0061, -0.0142,  0.0159, -0.0176, -0.0012, -0.0222,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0412, -0.0099,  0.0026, -0.0495, -0.1103, -0.0087, -0.0492, -0.0799,
        -0.0171, -0.0103, -0.0104, -0.0081, -0.0424,  0.0144, -0.0392, -0.0650,
        -0.0530,  0.0049, -0.0024, -0.0679, -0.0182, -0.0590, -0.0214, -0.0172,
         0.0028, -0.0142, -0.0011, -0.0612, -0.0063, -0.0164, -0.0542, -0.0135,
        -0.0061,  0.0221,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0123, -0.0315, -0.0297, -0.0057, -0.0863, -0.0035,  0.0192,  0.0035,
        -0.0050,  0.0117, -0.0014, -0.0088,  0.0428, -0.0033, -0.1549,  0.0240,
         0.0167, -0.0070, -0.0432,  0.0024, -0.0053,  0.0034, -0.0124,  0.0134,
        -0.0408, -0.0738,  0.0052,  0.0054,  0.0194, -0.0018, -0.0056, -0.0107,
         0.0027,  0.0176, -0.0281,  0.0076, -0.0248, -0.0430, -0.0137, -0.0436,
        -0.0260, -0.0158, -0.0406, -0.0058, -0.0038,  0.0009, -0.0013,  0.0036,
        -0.0044, -0.0030, -0.0018,  0.0011], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.5995e-02, -8.9924e-04,  4.8848e-03,  8.8664e-03, -1.4266e-01,
         7.0463e-03,  1.2248e-02,  1.1543e-02,  1.1218e-04,  5.6106e-03,
         1.9430e-02,  1.1471e-02, -2.3060e-02, -1.6010e-01, -3.2236e-02,
        -4.7240e-02, -1.1776e-01, -1.0934e-01, -6.5293e-02, -7.0572e-06,
        -9.7371e-03, -9.8593e-03, -2.1703e-02, -1.0075e-01,  1.3038e-02,
        -5.0237e-03,  1.3356e-03,  1.2176e-04,  1.5949e-02,  1.6099e-02,
         1.0579e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.7156e-02, -1.6622e-01,  1.1557e-02, -3.5473e-02, -6.5489e-05,
        -1.0532e-02, -3.6723e-02, -9.3201e-03, -6.2734e-03,  5.7144e-02,
         6.9996e-04, -6.1449e-02, -2.0427e-02,  9.8256e-03, -5.9689e-02,
        -6.8212e-02, -1.1256e-02, -9.6717e-02, -6.5244e-02, -1.1588e-01,
        -1.6250e-02, -3.0634e-03, -1.3574e-02,  2.5451e-02, -6.1790e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0382, -0.0341, -0.0043,  0.0214,  0.0772,  0.2319,  0.3509,  0.0620,
        -0.0290,  0.0447,  0.0473,  0.0590,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0063, -0.3395, -0.0232, -0.0321, -0.0893, -0.0144,  0.0007,  0.0091,
         0.0020, -0.0021, -0.0096, -0.0116, -0.0764, -0.0942, -0.0035, -0.0796,
        -0.0125, -0.0060, -0.0036, -0.0172, -0.0297, -0.0056, -0.0136, -0.0163,
        -0.0055, -0.0582, -0.0176, -0.0114, -0.0092,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 9.4555e-03,  4.4311e-01, -3.6754e-02, -7.1118e-02, -1.5579e-02,
         1.2053e-01,  1.9771e-01, -6.0719e-02,  3.8689e-03, -2.8482e-02,
         2.8293e-05, -1.2652e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0397, -0.3385,  0.0054, -0.1087,  0.0501, -0.1445, -0.1552, -0.0319,
        -0.0482,  0.0156,  0.0296, -0.0325,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0305,  0.0797, -0.1937, -0.0078, -0.1493,  0.0006,  0.0021,  0.0222,
        -0.1270,  0.0346, -0.0096, -0.1247, -0.0142, -0.0190, -0.0161, -0.0837,
         0.0120,  0.0146,  0.0107, -0.0478,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.9058e-02, -2.4171e-01,  9.2270e-04, -7.4840e-02, -1.4203e-01,
        -2.4462e-03, -2.9536e-02, -2.4453e-04, -1.2912e-02,  1.5619e-04,
        -3.6903e-03, -5.2627e-02, -2.9479e-02, -1.1568e-02, -1.0407e-03,
         1.7271e-04, -2.5164e-02,  7.4810e-03,  5.7942e-03, -7.1457e-03,
        -2.6922e-02,  1.5547e-02, -7.8024e-03, -5.3226e-03,  5.3966e-03,
         5.0609e-03,  2.9668e-02, -3.4250e-02, -4.1649e-02, -5.6581e-04,
         5.6169e-03, -1.4780e-03, -2.4951e-02, -7.4906e-02, -4.4089e-03,
         1.7799e-02, -2.0645e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1400,  0.6748,  0.0229, -0.0605,  0.0236,  0.0130,  0.0043,  0.0538,
        -0.0070,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0375,  0.0138, -0.0285, -0.1710, -0.0391, -0.0031, -0.0134, -0.0741,
         0.0187, -0.0175, -0.0123, -0.0390, -0.0959, -0.1196, -0.0209, -0.0068,
         0.0221, -0.0023, -0.0521, -0.0814,  0.0266,  0.0034, -0.0107,  0.0153,
        -0.0105,  0.0590,  0.0054,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.2750e-02, -2.7502e-01, -2.3037e-02, -5.6042e-02,  3.9947e-03,
        -2.3059e-02,  1.7708e-02,  1.6605e-02, -7.2611e-04,  2.0031e-02,
         6.6595e-03, -4.9684e-02,  5.2257e-03,  9.2411e-03,  3.3968e-03,
         1.1594e-02, -3.3666e-02, -1.4914e-02, -4.6222e-05,  7.9986e-04,
         4.8628e-03, -5.3679e-02, -1.0123e-01,  9.6954e-03, -6.9983e-04,
        -5.6043e-03, -1.3576e-02, -2.8161e-02, -8.4964e-02, -1.5115e-03,
        -8.2466e-03, -4.9379e-03,  1.4631e-02,  1.3993e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0467,  0.0029,  0.0103, -0.0051, -0.0636,  0.0163,  0.0038,  0.0043,
         0.0131,  0.0075, -0.0007, -0.0773,  0.0155, -0.0013,  0.0053, -0.0230,
        -0.0530,  0.0220, -0.0195, -0.0085,  0.0123,  0.0200, -0.0390, -0.0221,
        -0.0024,  0.0073, -0.0028,  0.0018, -0.0044, -0.0111,  0.0103, -0.0648,
        -0.0802,  0.0236, -0.0373, -0.0461, -0.0879, -0.0015, -0.0295, -0.0102,
        -0.0185, -0.0414, -0.0008,  0.0226, -0.0027,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0693,  0.0078,  0.0144,  0.0091, -0.0141, -0.1520, -0.2765, -0.0160,
         0.0414, -0.1923, -0.0015, -0.0222, -0.1025,  0.0285, -0.0008, -0.0514,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1022,  0.0046,  0.0050, -0.0108, -0.0084, -0.0017, -0.0139, -0.0458,
        -0.1500, -0.1305, -0.0428, -0.1848, -0.0045, -0.0532, -0.0891,  0.0014,
        -0.0232, -0.0073,  0.0107, -0.0097, -0.0049,  0.0088, -0.0044, -0.0085,
        -0.0284,  0.0336, -0.0119,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0401,  0.0159, -0.0896, -0.0026, -0.0402, -0.0018, -0.0364, -0.0097,
        -0.0107, -0.0032,  0.0036, -0.0044, -0.0256,  0.0026, -0.0341, -0.0058,
        -0.0200,  0.0024, -0.0028, -0.0177,  0.0158, -0.0229, -0.0128, -0.0681,
        -0.0141, -0.0099, -0.0209, -0.0755,  0.0207, -0.0440, -0.0044, -0.0196,
        -0.0418, -0.0124,  0.0010, -0.0369, -0.0219, -0.0422, -0.0094,  0.0007,
        -0.0301, -0.0321, -0.0058, -0.0038,  0.0261,  0.0380], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0194,  0.0509,  0.0272,  0.0118,  0.0126, -0.0281,  0.0077, -0.1636,
        -0.0108, -0.0103, -0.0076,  0.0136,  0.0124, -0.0104, -0.0023,  0.0036,
        -0.0083, -0.0186, -0.1055, -0.3128, -0.0151, -0.0042, -0.0074, -0.0055,
        -0.0012, -0.0074, -0.0634, -0.0244,  0.0143,  0.0198,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1490,  0.0113,  0.0339, -0.0286, -0.0506,  0.0352,  0.0636,  0.4631,
         0.1219, -0.0075,  0.0353,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0026, -0.4922, -0.0345,  0.0185, -0.0518, -0.0045, -0.0380, -0.0673,
        -0.0028,  0.0156,  0.0114,  0.0075, -0.0163, -0.0044, -0.0313, -0.0800,
        -0.0148, -0.0338, -0.0199,  0.0530,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0953, -0.4097, -0.1465, -0.0902,  0.0696, -0.0293, -0.0860,  0.0262,
        -0.0256,  0.0156, -0.0061,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0596, -0.3438, -0.0997, -0.1750,  0.0115,  0.0205,  0.0019, -0.0302,
        -0.0709,  0.0063,  0.0131, -0.0013,  0.0065,  0.0062,  0.0005, -0.0462,
         0.0155,  0.0119, -0.0105, -0.0116,  0.0573,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 6.6260e-04, -2.1758e-01, -1.9210e-01, -4.4368e-02, -1.4874e-04,
        -2.1592e-01,  7.3756e-02,  2.2526e-02, -5.9710e-02,  3.6456e-02,
        -6.0017e-03, -4.6183e-02,  1.4878e-02, -7.2850e-03, -2.0066e-02,
        -2.4046e-02, -1.5682e-02,  2.6258e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.3444e-03, -7.7671e-02, -2.3178e-01, -5.3087e-03, -1.7610e-02,
         7.4701e-03,  2.4001e-04, -5.9018e-03, -5.4797e-03, -1.1361e-02,
         2.1385e-03, -1.0965e-03, -6.2914e-03, -6.4380e-02, -6.2011e-02,
         4.4304e-03, -2.7955e-02, -5.7028e-02, -1.6028e-02,  3.2075e-03,
        -2.4002e-02, -6.4695e-02, -6.8628e-02, -2.9008e-03, -3.5139e-02,
        -2.1217e-03, -3.8529e-02, -5.8845e-03, -2.1525e-02, -5.8694e-02,
        -7.5429e-03, -1.1879e-02,  1.3173e-02,  8.6895e-04, -1.3551e-02,
        -1.7226e-04,  6.1462e-04,  7.7018e-03, -8.6444e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0402, -0.3955, -0.0137, -0.0587, -0.0028, -0.0019,  0.0264,  0.0199,
        -0.0204, -0.0074, -0.0117, -0.1115,  0.0160,  0.0030,  0.0183, -0.0032,
        -0.0498,  0.0072,  0.0034,  0.0007, -0.0057,  0.0032, -0.0021,  0.0067,
         0.0019, -0.0014, -0.0056,  0.0153, -0.0343, -0.0669,  0.0025, -0.0078,
         0.0112, -0.0237,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.2483,  0.0910, -0.0757, -0.2923, -0.0053, -0.1493, -0.0250, -0.0308,
        -0.0823,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0502, -0.1726, -0.0022, -0.0546, -0.0112, -0.0731,  0.0046, -0.0692,
        -0.0267, -0.0065, -0.0628, -0.0204, -0.0041, -0.0024,  0.0056, -0.0449,
        -0.0193, -0.0796, -0.0020,  0.0077, -0.0097, -0.0075,  0.0058, -0.0007,
        -0.0565,  0.0024, -0.0106, -0.0019,  0.0063,  0.0033, -0.0188, -0.0498,
         0.0085, -0.0507,  0.0077,  0.0031, -0.0003, -0.0045,  0.0200,  0.0123,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.7918e-02, -5.6293e-02, -1.9678e-01, -5.5902e-03, -4.9475e-02,
        -3.1294e-02,  1.6722e-03, -1.8394e-02, -5.4613e-02, -5.3059e-02,
        -2.0705e-03, -2.0137e-03,  5.8791e-04, -9.7387e-04,  5.9403e-03,
         7.0289e-03,  7.1820e-03,  1.2039e-02, -7.4744e-03, -1.1761e-04,
        -3.6004e-02,  6.6255e-03, -9.2506e-04, -1.3693e-02, -8.9815e-03,
        -1.7073e-02, -4.2394e-02,  8.0394e-03, -3.8120e-02, -8.5738e-02,
         5.7417e-03, -1.1651e-02, -2.0402e-03, -1.3818e-03,  3.4001e-03,
         5.4991e-03, -3.1438e-02, -5.8117e-02, -4.7781e-03, -6.6075e-03,
         3.7469e-03,  1.5394e-03, -8.2366e-04,  2.9070e-03,  7.9577e-04,
         6.2176e-03, -3.5871e-03, -1.7285e-04, -8.3722e-03,  4.6154e-03,
        -1.4505e-02,  2.3951e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.2776e-02, -3.6426e-01, -2.4775e-02, -3.2584e-02, -4.5068e-02,
         3.0508e-05, -6.6114e-02, -2.2662e-02, -5.4227e-02, -5.2636e-03,
         2.2983e-02, -1.4468e-03, -8.0786e-02, -4.4592e-03, -3.1102e-02,
        -6.6332e-02, -3.1377e-02, -1.8989e-02, -5.0181e-02, -2.2953e-03,
        -3.3893e-03, -2.8897e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0286,  0.6180,  0.0416,  0.0040, -0.0264,  0.0025,  0.0011,  0.0781,
        -0.0069, -0.0088,  0.0156,  0.0436,  0.0061,  0.0244,  0.0057,  0.0405,
         0.0273, -0.0021, -0.0043, -0.0091, -0.0055,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.9162e-02, -1.4862e-01, -1.8107e-02, -8.6597e-03,  7.5532e-03,
         4.7279e-03, -1.4986e-02,  1.3737e-02, -3.2548e-02,  5.9392e-03,
         2.9187e-03, -5.2678e-03, -2.5569e-03,  6.1955e-03,  1.0016e-03,
        -1.2667e-03,  1.0286e-02,  3.8794e-03, -3.5041e-02, -5.5545e-04,
         4.4634e-03,  3.0958e-04, -4.8296e-04,  2.6143e-04, -4.2651e-05,
        -2.6924e-03, -2.2615e-02, -5.3712e-02, -5.2282e-03, -4.1836e-02,
        -1.1781e-01, -6.0015e-02, -2.3073e-03, -3.4114e-02, -6.7807e-03,
        -4.5962e-03,  7.5405e-03,  1.0596e-02,  7.2382e-03, -4.2174e-04,
        -7.4339e-02,  3.8382e-03, -3.6765e-03, -1.8933e-03, -2.1650e-02,
         5.3245e-04, -3.1747e-02, -9.6555e-03, -9.8658e-03,  1.3348e-02,
        -2.2929e-02, -2.3961e-03,  8.7254e-03, -1.3602e-02, -3.5734e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0016, -0.0051,  0.0025, -0.0194, -0.0134, -0.0331, -0.0829, -0.0176,
         0.0237,  0.0039, -0.0031,  0.0089, -0.0106, -0.0159, -0.0475, -0.0116,
        -0.0299, -0.0292, -0.0027, -0.0324, -0.0302, -0.0219, -0.0236,  0.0026,
         0.0101, -0.0052, -0.0043, -0.0751, -0.0205, -0.0050,  0.0167,  0.0016,
        -0.0193, -0.0991, -0.0031, -0.0211, -0.0087, -0.0356, -0.0369, -0.0412,
         0.0018, -0.0119, -0.0314, -0.0670, -0.0114,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0762, -0.2605, -0.0722, -0.0726,  0.0265, -0.0160, -0.0133,  0.0045,
        -0.0706,  0.0022,  0.0283, -0.0457,  0.0012, -0.0446, -0.0695, -0.0068,
        -0.0193, -0.0308, -0.0058, -0.0341, -0.0413, -0.0134, -0.0161,  0.0286,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0063,  0.0181,  0.0089, -0.0047, -0.0047, -0.0113, -0.0752, -0.0428,
         0.0298, -0.0009,  0.0022, -0.0086, -0.0108, -0.0161,  0.0139,  0.0466,
        -0.0563, -0.0396, -0.0521, -0.0792, -0.0070, -0.0109, -0.0557, -0.0686,
         0.0073, -0.0339, -0.0383, -0.0050, -0.0284, -0.0778, -0.0261,  0.0527,
        -0.0600,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0992,  0.4134,  0.0159,  0.1082,  0.0076,  0.0181, -0.0127,  0.0075,
         0.1034, -0.0035,  0.0185,  0.0104, -0.0194,  0.0717, -0.0010,  0.0400,
         0.0074, -0.0067,  0.0354,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0259,  0.0215,  0.0383, -0.0023, -0.1216, -0.0146, -0.0557, -0.1498,
        -0.1877, -0.0017,  0.0146, -0.0101,  0.0146,  0.0242, -0.0066,  0.0055,
        -0.0127, -0.0690, -0.0429, -0.0156, -0.0527, -0.0880, -0.0136,  0.0064,
        -0.0040,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0103, -0.2107, -0.1249,  0.0205, -0.0269, -0.0968, -0.1832,  0.0412,
         0.0071, -0.0377,  0.0408,  0.0165, -0.0404,  0.0086, -0.0043, -0.0604,
         0.0696,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0709,  0.0007, -0.0328,  0.0048, -0.0020, -0.0015, -0.0039, -0.0038,
        -0.0258, -0.1210, -0.0317, -0.0066,  0.0191, -0.0096, -0.0016,  0.0010,
        -0.1214, -0.1635,  0.0039, -0.0638, -0.0068, -0.0959, -0.1123, -0.0061,
         0.0100,  0.0060, -0.0022,  0.0054,  0.0121, -0.0165, -0.0077,  0.0297,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0285, -0.0979, -0.1290, -0.0173, -0.0053, -0.0157, -0.0320, -0.1097,
        -0.0031,  0.0063,  0.0100,  0.0030, -0.0043,  0.0006, -0.0178,  0.0157,
        -0.0254, -0.0130, -0.0627, -0.0036, -0.0718, -0.0133, -0.0635, -0.1284,
        -0.0079, -0.0095, -0.0880, -0.0167,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0424, -0.0998, -0.0127, -0.0333, -0.0353, -0.1035, -0.1633, -0.0647,
         0.0020, -0.0075, -0.0130, -0.0733, -0.0121, -0.0005, -0.0125, -0.0666,
        -0.0104, -0.0628, -0.0210, -0.0017, -0.0059,  0.0175, -0.0064, -0.0211,
        -0.0385,  0.0229,  0.0187,  0.0230, -0.0075,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0175, -0.1179, -0.0119, -0.0587, -0.1498, -0.0259, -0.0022,  0.0039,
        -0.0097,  0.0254, -0.0023, -0.0212, -0.0546, -0.0870,  0.0345, -0.0572,
         0.0150, -0.1159,  0.0053, -0.0175, -0.0042, -0.0054, -0.0160, -0.0043,
        -0.0207, -0.0358, -0.0438, -0.0072,  0.0037,  0.0016, -0.0024, -0.0070,
         0.0149,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0628, -0.3244,  0.0082, -0.0526, -0.0043, -0.0553,  0.0261, -0.0494,
        -0.0172, -0.0813, -0.0519, -0.0078,  0.0054, -0.0109,  0.0061,  0.0096,
        -0.0244,  0.0152, -0.0937,  0.0134, -0.0468,  0.0031, -0.0295,  0.0008,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0045, -0.2304, -0.1685,  0.0030, -0.0591,  0.0052, -0.0147, -0.0684,
         0.0257, -0.0491, -0.0011, -0.0298, -0.0535, -0.0066,  0.0029, -0.0051,
         0.0130, -0.0462,  0.0077, -0.1413, -0.0133,  0.0200,  0.0307,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0482, -0.2186,  0.0273, -0.0048, -0.0201, -0.0361,  0.0080,  0.0062,
         0.0022, -0.0029,  0.0035, -0.0053, -0.0183,  0.0055,  0.0039, -0.0088,
        -0.0830, -0.0435, -0.0167,  0.0093,  0.0043, -0.0391, -0.0038, -0.0216,
        -0.0559, -0.0183, -0.0095,  0.0010, -0.0041, -0.0467, -0.0387, -0.0156,
        -0.0382, -0.0370, -0.0336, -0.0405, -0.0006,  0.0112, -0.0077,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0201, -0.0669,  0.0037, -0.0330, -0.1325, -0.0066, -0.0056,  0.0051,
         0.0046, -0.0114,  0.0096,  0.0055, -0.0049, -0.0039, -0.0002, -0.0578,
        -0.0400, -0.0053, -0.0053,  0.0141, -0.0607, -0.0021, -0.0079, -0.0100,
        -0.0926, -0.1237, -0.0077, -0.0245,  0.0043, -0.0447, -0.0189,  0.0032,
         0.0053, -0.0043, -0.0043,  0.0048, -0.0157, -0.0628,  0.0088, -0.0061,
        -0.0179,  0.0249, -0.0087,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.7302e-02, -2.9946e-01, -2.4095e-02, -2.1291e-03,  7.0812e-03,
         5.3158e-04, -7.6712e-03, -1.1199e-02, -7.5108e-02, -2.6844e-02,
         1.0894e-02,  1.4565e-03, -1.7751e-02, -5.2355e-03,  7.1427e-03,
        -2.8319e-03, -2.7793e-02, -4.4821e-02, -1.0910e-02, -4.1674e-03,
         3.3556e-03,  4.6681e-04, -7.6488e-03, -3.0819e-03,  1.0238e-02,
        -3.3266e-03,  1.6944e-03, -1.7596e-03,  8.7393e-05, -7.9991e-04,
         6.3908e-03, -2.4221e-02, -3.4285e-02,  1.3452e-02, -2.2036e-03,
         5.7972e-03, -2.6232e-02, -3.2500e-02, -3.0589e-02,  6.6240e-04,
        -2.4224e-02, -4.2948e-02, -3.9253e-03,  1.1293e-02, -4.7329e-03,
        -5.9663e-04, -1.2261e-03,  2.5722e-03,  2.4194e-03, -4.6286e-03,
         1.1367e-02, -2.4804e-03,  4.8147e-03, -5.1416e-03,  8.2089e-03,
        -3.2630e-02,  1.3577e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.0891e-02, -5.4001e-01,  7.2649e-02,  8.7067e-03, -7.0539e-03,
        -2.9537e-04, -4.5391e-02, -7.7671e-02, -2.4340e-02,  1.3987e-03,
        -9.2848e-02,  1.8009e-02, -6.0586e-04,  2.0132e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0916, -0.2271, -0.0416,  0.0276, -0.0135, -0.0295, -0.0091,  0.0109,
        -0.0271, -0.0218, -0.0422, -0.1242,  0.0115, -0.0189, -0.0383,  0.0076,
        -0.0244, -0.0359, -0.0027, -0.0366, -0.0602, -0.0038, -0.0200,  0.0040,
         0.0155,  0.0186,  0.0049,  0.0123, -0.0144,  0.0040,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0265, -0.3038, -0.0263, -0.2538, -0.2127,  0.0885,  0.0216, -0.0667,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0202, -0.3405, -0.0441,  0.0014, -0.0233, -0.0865,  0.0106, -0.0051,
         0.0136,  0.0070,  0.0111, -0.0095, -0.0591, -0.0596, -0.0100,  0.0026,
        -0.0022,  0.0058, -0.0046,  0.0101, -0.0037,  0.0011,  0.0035, -0.0206,
        -0.0608, -0.0090, -0.0159,  0.0025, -0.0439, -0.0893,  0.0032,  0.0083,
        -0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7949e-02, -2.2764e-01, -1.0434e-01, -3.2096e-02,  5.9690e-04,
        -1.7271e-02, -2.2070e-02, -1.9922e-02,  9.3044e-03, -3.0201e-02,
         4.1470e-03, -4.9613e-03, -1.3736e-02,  1.0106e-02,  1.2713e-02,
        -3.7461e-03,  7.2443e-03,  1.2878e-03, -1.0802e-02,  2.4812e-03,
         2.1698e-03,  1.0409e-03, -4.8174e-04,  1.4532e-03,  1.3592e-02,
        -5.9775e-03, -1.3092e-02, -4.5897e-02,  2.9624e-03, -3.4973e-03,
         2.0620e-03, -5.9721e-03, -2.4031e-02, -4.1703e-02,  9.5838e-03,
         5.8605e-03, -9.6766e-03,  1.0138e-02, -5.1950e-06,  4.0306e-03,
        -8.3486e-06, -4.2304e-04, -9.3622e-03, -3.3717e-02, -3.6511e-03,
        -2.7780e-03,  4.2768e-03,  8.4939e-04,  3.5315e-03, -1.5043e-05,
         5.3437e-03, -2.1569e-02, -4.1933e-03, -9.7657e-03, -3.7520e-03,
         1.8659e-04,  1.0820e-03, -4.5009e-04,  3.9206e-03, -2.5744e-02,
        -2.5384e-03, -7.6772e-03, -4.9586e-02,  7.5750e-03,  1.8993e-03,
        -9.0107e-05, -6.3092e-03,  8.6125e-03,  4.6219e-03, -5.3307e-03,
         1.5292e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.4448e-02,  1.6347e-02, -7.6806e-03, -4.7392e-03, -4.7375e-02,
        -1.5483e-02, -7.2892e-02, -1.3918e-02,  5.4831e-04, -6.0729e-04,
         1.3358e-04,  6.8691e-05,  5.5783e-03, -2.6096e-03,  1.3263e-02,
         5.6612e-03,  4.9510e-03,  9.3792e-03,  2.0771e-02, -9.4081e-02,
        -2.2402e-03,  8.7275e-03,  3.4518e-02, -8.3060e-02,  1.5971e-02,
        -5.9256e-02, -1.1238e-02,  2.0273e-02,  1.0682e-03, -1.0276e-03,
         1.4109e-02, -1.6981e-02, -6.5993e-02, -8.8433e-02, -2.0748e-02,
        -1.4285e-02, -7.9646e-02, -8.9133e-03,  6.2816e-03,  4.7278e-03,
        -6.5258e-03,  6.5443e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.2937e-03,  2.6061e-01, -1.0608e-02,  1.8460e-02,  4.5012e-03,
        -7.7657e-03, -1.4404e-02, -5.7600e-03,  2.8787e-02,  6.1700e-02,
         1.1512e-02,  2.2456e-02,  3.0905e-02, -1.6732e-05, -6.6088e-03,
         7.0419e-03,  4.1644e-03,  5.7135e-02,  5.3888e-02,  4.4134e-03,
         1.5648e-02,  1.9327e-02, -2.9938e-03, -6.5671e-03,  2.7140e-02,
         6.6390e-02, -7.8541e-03,  2.2622e-03,  7.5115e-02,  7.1567e-03,
         3.3493e-04,  1.3845e-02, -1.4832e-03,  1.9620e-03,  1.6677e-02,
        -8.2439e-02,  3.9770e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1175,  0.4993,  0.0283,  0.1216, -0.0302, -0.0202, -0.0130, -0.0357,
        -0.0106, -0.0107,  0.0296, -0.0263,  0.0234,  0.0070,  0.0266,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1548, -0.0607,  0.0054, -0.0162, -0.0757, -0.1318,  0.0042, -0.0671,
        -0.1616, -0.0306, -0.0179, -0.0126, -0.0339, -0.0030,  0.0440,  0.0244,
        -0.0272, -0.0300, -0.0328,  0.0661,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0266,  0.0050,  0.0040,  0.0064, -0.1257, -0.0214, -0.0003, -0.0126,
        -0.0051, -0.0191, -0.0572, -0.0722, -0.0145, -0.0108, -0.0397, -0.0415,
        -0.0090, -0.0196,  0.0052, -0.0395, -0.0033,  0.0038, -0.0071, -0.0031,
         0.0031, -0.0664, -0.0751,  0.0128, -0.0029, -0.0317, -0.0722, -0.0146,
         0.0023, -0.0048, -0.0316, -0.0422,  0.0099, -0.0334,  0.0004, -0.0031,
        -0.0092,  0.0036, -0.0283,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.4090e-02, -1.2454e-02, -3.9086e-02, -8.3211e-03, -6.2000e-02,
        -3.0547e-02, -4.6549e-02, -9.7297e-02, -2.3229e-02, -1.6479e-03,
         1.6780e-03,  3.4107e-03,  1.0427e-02, -5.6751e-02, -1.0993e-01,
        -9.1623e-04,  2.5449e-03,  1.7994e-02, -8.7816e-05, -2.7928e-02,
        -5.9530e-02, -9.6495e-03,  4.2282e-03,  2.1687e-03, -2.2127e-02,
        -1.4199e-02,  5.7231e-03,  3.7217e-03, -1.0596e-01, -3.6820e-03,
        -2.4400e-03, -5.0773e-02, -7.8527e-03, -5.0602e-02, -5.6783e-02,
        -4.5341e-03, -9.3024e-03,  9.8297e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0341, -0.3273, -0.0782, -0.1139,  0.0161, -0.0675, -0.0158,  0.0221,
        -0.0328, -0.0549, -0.0159,  0.0035, -0.0365,  0.0060,  0.0218, -0.0265,
         0.0264, -0.0686, -0.0039,  0.0114,  0.0097, -0.0073,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0059, -0.1567, -0.0033,  0.0190, -0.0115, -0.0340, -0.0158,  0.0246,
         0.0079, -0.0940, -0.0056,  0.0125, -0.0160,  0.0100, -0.0248, -0.0510,
        -0.0940, -0.0317, -0.0021, -0.0560, -0.0016,  0.0117, -0.0443, -0.0061,
        -0.0135, -0.0013, -0.0009,  0.0103, -0.0008, -0.0305,  0.0036, -0.0126,
        -0.0026, -0.0135, -0.0302, -0.0694,  0.0221, -0.0020, -0.0108, -0.0357,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0036, -0.0087,  0.0075,  0.0123, -0.0085, -0.0545, -0.0106,  0.0065,
        -0.0616,  0.0003,  0.0050, -0.0072, -0.0225, -0.1512, -0.1411,  0.0135,
        -0.0020,  0.0174, -0.1228,  0.0152,  0.0012, -0.0112, -0.0063, -0.0026,
        -0.0102, -0.0750, -0.0078, -0.0656, -0.0961,  0.0058, -0.0314, -0.0145,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0463, -0.3627, -0.0296, -0.0057, -0.0059, -0.0072, -0.0228, -0.0682,
        -0.0847,  0.0015, -0.0016, -0.0230, -0.0182,  0.0049, -0.0035, -0.0202,
        -0.0268, -0.0200,  0.0009,  0.0034, -0.0148,  0.0070, -0.0027, -0.0513,
        -0.0063,  0.0342, -0.0371, -0.0107,  0.0073, -0.0051,  0.0033, -0.0103,
        -0.0043,  0.0197, -0.0289,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0079, -0.0583, -0.1185, -0.1251, -0.0078, -0.0185, -0.1550, -0.1238,
        -0.0407, -0.1567,  0.0029, -0.0246,  0.0095, -0.0375, -0.0161, -0.0115,
        -0.0253, -0.0080,  0.0063,  0.0236,  0.0224,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0040, -0.2944, -0.0113, -0.0804, -0.1427, -0.0442, -0.0236, -0.0648,
        -0.0989,  0.0195, -0.0126,  0.0127, -0.0092, -0.0237, -0.0439, -0.0025,
        -0.0567,  0.0317,  0.0233,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.9455e-02, -2.3637e-02, -1.5591e-01, -1.9576e-02, -7.6779e-02,
        -1.4224e-02,  1.4281e-03,  2.2027e-02,  5.1664e-06, -9.2305e-02,
        -1.3718e-01, -3.8496e-02, -6.0762e-02, -1.3278e-01, -8.9981e-03,
        -4.2205e-02,  5.7348e-03, -6.6732e-02,  5.6477e-02, -5.2886e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0020, -0.3654,  0.0157, -0.0689, -0.0015, -0.0038,  0.0137, -0.0394,
        -0.0156, -0.0219,  0.0136, -0.0167,  0.0221, -0.0896, -0.0317, -0.0415,
        -0.0790, -0.0078, -0.0182, -0.0496,  0.0077,  0.0383,  0.0361,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0232, -0.2335, -0.1084, -0.0336,  0.0119, -0.0522, -0.0055, -0.0434,
        -0.0901, -0.0997, -0.0194, -0.0551, -0.0031, -0.0603,  0.0085, -0.0008,
         0.0037, -0.0113, -0.0118,  0.0096, -0.0220, -0.0094,  0.0170, -0.0430,
         0.0235,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.3595e-02, -3.3885e-01, -4.4173e-02, -1.6780e-02, -7.9425e-03,
        -1.2145e-03,  5.3498e-05, -6.7831e-04,  2.8201e-02,  7.5946e-03,
        -7.2931e-02,  4.6993e-03, -1.0594e-02, -5.2852e-03, -1.2361e-02,
        -5.8534e-03,  9.6805e-04, -7.7097e-04,  9.2500e-03, -2.3897e-02,
        -1.0847e-01, -4.0599e-02, -4.9916e-02, -1.7003e-02, -1.6835e-02,
        -1.4879e-02, -7.4551e-03, -2.6963e-03, -1.3404e-02, -3.6984e-02,
        -2.0871e-02,  5.8004e-03, -1.4007e-02, -1.1096e-02,  4.2825e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0452, -0.0006, -0.0110, -0.0926, -0.0227,  0.0122, -0.0822,  0.0189,
        -0.1124, -0.0011, -0.0127, -0.0364, -0.0005, -0.1036, -0.1121, -0.1471,
        -0.0195, -0.0967,  0.0145,  0.0081, -0.0173,  0.0327,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0930, -0.0204, -0.0013,  0.0309,  0.0167,  0.0591, -0.0218,  0.0231,
         0.0436,  0.3801, -0.0192,  0.0445, -0.0345, -0.0155,  0.0096,  0.0228,
         0.0554,  0.0118, -0.0061,  0.0299,  0.0606,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0567, -0.3800, -0.0395, -0.0259, -0.0213,  0.0030, -0.0471, -0.0908,
         0.0290,  0.0078, -0.0203, -0.0045, -0.0059, -0.0893,  0.0061, -0.0739,
         0.0386,  0.0233, -0.0368,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.0242e-02,  2.3672e-02,  2.2258e-04,  1.0927e-02, -2.1267e-03,
        -9.7062e-02,  9.0751e-03, -1.3412e-02, -1.7379e-03, -9.5057e-03,
        -7.8536e-03, -3.1137e-02, -1.3738e-02, -2.5669e-02, -2.1703e-03,
         8.5690e-04, -3.0372e-02, -7.0596e-02, -1.3988e-02, -3.1010e-02,
         9.7251e-05, -2.6829e-02, -3.7454e-02, -2.0398e-03, -7.3358e-02,
        -8.9271e-03, -3.6958e-03, -3.8740e-02, -1.6531e-02, -1.0213e-01,
         5.7310e-03, -9.5756e-03, -4.9197e-02, -8.9332e-02, -3.1763e-03,
        -6.7968e-03, -9.9468e-03,  1.0438e-02, -3.8868e-03, -3.5109e-02,
        -3.5949e-03,  9.9938e-03, -1.5994e-03, -4.7569e-03,  4.1878e-03,
        -3.0747e-03,  1.4432e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0247, -0.2431, -0.0945, -0.0054,  0.0078, -0.0016, -0.0044,  0.0147,
        -0.0425,  0.0233,  0.0086,  0.0062,  0.0029, -0.0703,  0.0185,  0.0116,
        -0.0359,  0.0153, -0.0028,  0.0028,  0.0017, -0.0540, -0.0969, -0.0615,
         0.0046,  0.0148, -0.0247, -0.0032, -0.0044,  0.0110,  0.0078,  0.0099,
        -0.0363,  0.0322,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.9185e-02, -3.4777e-01, -1.5298e-04,  1.3805e-02, -2.2345e-02,
        -4.9921e-02, -2.0746e-02, -4.3633e-03, -6.8732e-04, -3.7051e-02,
         3.9449e-03,  2.4489e-03, -6.9678e-02, -1.0342e-01, -4.5488e-03,
        -7.5789e-04, -1.7339e-02, -2.5860e-02, -3.6609e-03, -7.6372e-02,
         9.8981e-04, -2.9672e-03, -1.4062e-03,  7.8514e-03,  1.6129e-03,
         5.2905e-03, -1.8767e-03, -1.5879e-02, -4.7252e-03, -2.2294e-02,
         3.0053e-03, -1.0241e-02,  1.3788e-02,  2.4011e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-2.9169e-02, -3.5802e-01, -5.3408e-03, -1.9811e-02, -3.4593e-02,
        -9.2193e-02,  7.6769e-03,  7.9738e-03,  3.0323e-04, -9.9097e-02,
         2.6703e-02, -3.8606e-02, -5.7171e-02, -1.3827e-01, -3.7390e-02,
         4.2734e-03,  2.5240e-02, -1.8174e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0330, -0.0124, -0.0105, -0.0119, -0.0105, -0.0071,  0.0052, -0.0164,
        -0.1900, -0.3042, -0.0579, -0.0220, -0.0141, -0.0166, -0.0035, -0.0305,
        -0.0033,  0.0131, -0.0062, -0.1523,  0.0042,  0.0135, -0.0264,  0.0054,
        -0.0056, -0.0038, -0.0203,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0277, -0.0061, -0.0145,  0.0079,  0.0084,  0.0071,  0.0022, -0.0289,
        -0.1244,  0.0330,  0.0151, -0.0391, -0.1654, -0.0458,  0.0022, -0.0079,
        -0.0005, -0.0151, -0.0117, -0.2778, -0.0159, -0.0021, -0.0695, -0.0212,
        -0.0161, -0.0343,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0181, -0.2889, -0.0481, -0.0776, -0.1032, -0.0241,  0.0025, -0.0029,
        -0.0466, -0.0232, -0.0514, -0.0342,  0.0315,  0.0067, -0.0117, -0.0764,
        -0.0367, -0.0386, -0.0077, -0.0097,  0.0604,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0498, -0.2282, -0.1790, -0.2224, -0.0552, -0.0096, -0.0864,  0.0018,
        -0.0254,  0.0250,  0.0171, -0.0759,  0.0104, -0.0044,  0.0094,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0608,  0.0312, -0.0042, -0.0148, -0.0161, -0.0038, -0.0224,  0.0468,
        -0.0308, -0.0133, -0.0458,  0.0124,  0.4549,  0.1392,  0.0180,  0.0267,
        -0.0177,  0.0412,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0791, -0.2204, -0.0933, -0.1505, -0.0197, -0.0069,  0.0209, -0.0378,
        -0.0137, -0.0634, -0.0840, -0.0367,  0.0305, -0.0369, -0.0079,  0.0114,
         0.0090,  0.0136,  0.0645,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.7955e-02, -1.2791e-01, -1.2034e-01,  9.3490e-03, -5.7377e-02,
        -2.6162e-02,  3.2196e-02, -1.1368e-02, -1.7560e-02,  2.0219e-04,
         1.5525e-03, -6.6566e-03, -2.3183e-02,  6.2641e-03,  1.4115e-03,
         1.3613e-02, -6.4797e-02, -2.2615e-03,  8.0147e-03,  1.1974e-02,
        -5.2135e-02,  1.1666e-02, -3.8693e-02, -8.4429e-03, -5.9611e-02,
         5.1813e-03, -3.3309e-02,  1.4102e-02, -5.3353e-03, -1.6762e-02,
        -3.1286e-05, -2.0926e-03, -3.5617e-02,  2.9531e-03, -3.4703e-02,
         1.4948e-03, -5.6258e-02, -9.6750e-03, -9.5601e-03, -4.5224e-04,
         1.2342e-03,  2.5309e-04,  1.4485e-02,  5.8041e-03], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0280, -0.2758, -0.0375, -0.0898, -0.1106,  0.0057, -0.0235, -0.0050,
        -0.0942, -0.0098, -0.0159,  0.0040,  0.0158,  0.0010,  0.0007,  0.0009,
         0.0139, -0.0329, -0.0282, -0.0409, -0.0316,  0.0025, -0.0008, -0.0040,
         0.0036, -0.0059,  0.0026, -0.0492, -0.0093,  0.0062, -0.0042, -0.0291,
        -0.0098, -0.0049, -0.0025,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.6959e-02, -3.1287e-01, -1.8006e-02, -7.2413e-02, -1.2575e-02,
        -1.7832e-02, -9.1967e-02, -6.0952e-02, -3.2093e-03, -2.0495e-03,
        -4.4926e-04, -3.6676e-02, -1.2223e-01,  4.4800e-03,  2.3036e-04,
        -6.3707e-02,  6.4160e-03, -5.5343e-03,  1.1539e-03, -1.9784e-03,
         7.4621e-03,  5.9496e-03,  1.2913e-02,  5.1988e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0671, -0.2301, -0.0103, -0.0527, -0.0287,  0.0010,  0.0242, -0.0009,
        -0.0898, -0.0158, -0.0315,  0.0159, -0.0100, -0.0292, -0.0624, -0.0100,
         0.0141,  0.0177, -0.0539, -0.0063,  0.0010, -0.0430,  0.0095, -0.0307,
        -0.0077, -0.0450, -0.0239, -0.0118, -0.0068, -0.0051, -0.0211, -0.0009,
         0.0136,  0.0087,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.8052e-02, -1.0200e-04, -3.4423e-02,  3.6618e-04,  8.7154e-03,
        -9.3754e-03, -1.7236e-01,  1.7416e-02,  2.0618e-02,  5.1928e-03,
         2.5827e-04,  1.6927e-02, -1.1552e-01, -2.1436e-02, -1.0887e-01,
        -1.5921e-01, -3.0138e-02, -5.2581e-03,  5.7924e-03,  1.1862e-02,
        -1.0401e-02, -8.8038e-03,  1.3879e-03, -4.3819e-02, -3.5495e-02,
        -1.6833e-02, -3.8001e-02, -5.2442e-02, -7.8341e-03, -1.3084e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-3.0566e-02, -3.5371e-02,  8.6960e-02,  3.4320e-01,  3.0968e-02,
        -3.2289e-02,  5.3546e-03,  6.5362e-02,  1.1295e-01,  2.2949e-01,
        -1.3638e-02, -7.7073e-05, -1.3789e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0169, -0.0026,  0.0474, -0.0118,  0.0205, -0.0342, -0.0841, -0.1771,
        -0.2332, -0.0162,  0.0208, -0.1013,  0.0090, -0.0516, -0.0869,  0.0434,
        -0.0099, -0.0331,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0035, -0.3186, -0.3580, -0.0341, -0.1465,  0.0154, -0.0889, -0.0350,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0148, -0.1718, -0.0072, -0.0275, -0.0180, -0.0049, -0.0135, -0.0160,
        -0.0034, -0.0237, -0.0028,  0.0154, -0.0216, -0.0038, -0.0024, -0.0050,
         0.0020,  0.0024,  0.0019, -0.0005,  0.0099, -0.0144, -0.0493, -0.0188,
        -0.0111, -0.0150,  0.0025, -0.0092, -0.0636, -0.0251,  0.0053, -0.0617,
        -0.0088, -0.0942, -0.0049, -0.0223, -0.0329, -0.0089,  0.0183,  0.0097,
        -0.0189, -0.0156, -0.0149,  0.0071,  0.0238, -0.0433,  0.0324],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0494, -0.1151, -0.0086, -0.4264, -0.1549, -0.0324,  0.0517, -0.0275,
         0.0409,  0.0766,  0.0165,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0492,  0.4113,  0.0272,  0.0288, -0.0073, -0.0006,  0.0386,  0.1494,
        -0.0331,  0.0601,  0.0232,  0.0152,  0.0562, -0.0184,  0.0344,  0.0048,
        -0.0036, -0.0385,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1298,  0.0956,  0.4061,  0.0297,  0.0027, -0.0234,  0.1562,  0.0067,
         0.0581,  0.0461,  0.0201, -0.0166, -0.0047, -0.0009, -0.0034,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0954, -0.3218, -0.0547, -0.0071,  0.0189, -0.0370, -0.1420, -0.0226,
         0.0581, -0.0828, -0.0837, -0.0061, -0.0152,  0.0449,  0.0097,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0410,  0.0043, -0.0129,  0.0054, -0.0121,  0.0002,  0.0129,  0.0068,
        -0.0926, -0.0526, -0.0149, -0.1234, -0.1588,  0.0015,  0.0032, -0.0084,
        -0.0526, -0.0279,  0.0247,  0.0242, -0.0864, -0.1581,  0.0024,  0.0269,
        -0.0411, -0.0045,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0537,  0.0711,  0.0249, -0.0004, -0.0297, -0.0671,  0.0048,  0.2644,
         0.2842, -0.0432,  0.0242, -0.0203,  0.0192,  0.0270,  0.0659,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0614,  0.4210,  0.0058,  0.0230,  0.0434,  0.0315,  0.0596, -0.0143,
        -0.0037,  0.1121,  0.0029, -0.0049,  0.0310,  0.0097,  0.0033, -0.0059,
         0.0317,  0.0052,  0.0477, -0.0062, -0.0038, -0.0143, -0.0269, -0.0054,
        -0.0127,  0.0125,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.7775e-02, -9.8513e-03,  1.9100e-02,  2.4877e-02, -1.5706e-02,
        -2.0553e-02, -1.2134e-05,  1.9946e-03, -2.7097e-02, -4.5948e-02,
        -1.3506e-01, -2.7185e-02, -8.0209e-03, -1.7498e-02, -1.1302e-01,
        -1.0252e-01, -1.0445e-02, -2.6260e-02, -6.0473e-03, -8.0950e-02,
        -6.0364e-02, -4.0031e-02, -3.5802e-02, -3.3746e-02, -3.3941e-02,
         6.6192e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-1.8255e-02, -3.9174e-01,  7.1712e-03,  2.0717e-02, -3.0054e-02,
        -7.2294e-03, -1.7790e-02, -7.5893e-02, -7.2190e-03, -2.2563e-02,
        -4.4875e-03, -2.3351e-02, -5.2256e-02,  1.2222e-02, -1.4846e-02,
         1.5774e-04,  5.1785e-03,  1.5009e-02, -1.1035e-02, -2.1990e-03,
        -1.3239e-03, -3.2831e-02,  4.6913e-03, -2.9449e-02, -5.8166e-02,
         1.8451e-02, -2.7678e-02, -5.8882e-03, -1.0177e-03,  6.5192e-03,
         4.6158e-03, -3.6056e-02, -2.4219e-03,  4.6886e-03, -1.0600e-02,
         1.6231e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0559,  0.1077,  0.0551,  0.0189,  0.0541, -0.0002,  0.0603,  0.0723,
        -0.0151,  0.0006,  0.0570,  0.0075,  0.0084, -0.0169,  0.0139,  0.0289,
        -0.0030,  0.0011, -0.0049, -0.0153,  0.0471,  0.0392, -0.0040,  0.0052,
         0.0288,  0.0492, -0.0076,  0.0455,  0.0366,  0.0060,  0.0080,  0.0323,
         0.0461, -0.0138,  0.0198, -0.0138,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0159, -0.0602,  0.0048, -0.0044,  0.0612, -0.2206, -0.0296, -0.0121,
         0.0011, -0.0134, -0.0390, -0.1003, -0.2444,  0.0579,  0.0013, -0.0441,
        -0.0023,  0.0009, -0.0046,  0.0818,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0757, -0.1478, -0.0218, -0.0232,  0.0173, -0.0059, -0.0305, -0.0925,
        -0.0219, -0.0345,  0.0017, -0.0049, -0.0005, -0.0111, -0.0689, -0.0523,
        -0.0224, -0.0338, -0.0115, -0.0103, -0.0049,  0.0005,  0.0028,  0.0135,
        -0.0111,  0.0066,  0.0008,  0.0019, -0.0045, -0.0383, -0.0642, -0.0377,
         0.0096, -0.0589, -0.0063,  0.0053, -0.0243,  0.0203,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0430, -0.2030, -0.1173, -0.1006, -0.0175,  0.0020,  0.0133,  0.0086,
        -0.0011,  0.0082, -0.0269, -0.0470, -0.0066,  0.0104,  0.0027, -0.0243,
         0.0056, -0.0370,  0.0084,  0.0062,  0.0056,  0.0045,  0.0163,  0.0047,
        -0.0057, -0.0088, -0.0169, -0.0286, -0.0690,  0.0035, -0.0206, -0.0177,
         0.0076,  0.0034, -0.0342,  0.0096, -0.0128, -0.0162,  0.0003, -0.0160,
        -0.0082], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0063, -0.5257, -0.0163, -0.0094, -0.0308, -0.0870, -0.0202, -0.0097,
        -0.0229,  0.0097,  0.0008, -0.0532, -0.0826,  0.0136, -0.0007, -0.0411,
         0.0148, -0.0329, -0.0042, -0.0031,  0.0106, -0.0044,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0315,  0.0088,  0.0067, -0.0021,  0.0056, -0.0099, -0.0031,  0.0093,
        -0.0197,  0.0076, -0.0080, -0.0537, -0.0004, -0.0198,  0.0174, -0.0592,
         0.0134, -0.0616, -0.0213, -0.0746, -0.1117,  0.0202,  0.0283, -0.0658,
         0.0090,  0.0167, -0.0048, -0.0299, -0.1020, -0.0198, -0.0457, -0.0659,
        -0.0005,  0.0344,  0.0117,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0020, -0.0896, -0.1690, -0.0202, -0.1187,  0.0006,  0.0118, -0.0062,
         0.0064, -0.0029, -0.0198,  0.0187, -0.0158,  0.0226, -0.0095, -0.0058,
         0.0257, -0.0367, -0.0805,  0.0020, -0.0570, -0.0031, -0.0423, -0.0494,
        -0.0126, -0.0126, -0.0764, -0.0542,  0.0280,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0220, -0.3335, -0.0610, -0.0714, -0.0104, -0.0555,  0.0168, -0.0274,
        -0.0213, -0.0083, -0.0548, -0.0431,  0.0044, -0.0051, -0.0080, -0.0679,
        -0.0380, -0.0286, -0.0547,  0.0038, -0.0005, -0.0185, -0.0300,  0.0121,
         0.0028,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0089,  0.0278,  0.0070,  0.0475, -0.1093,  0.0203, -0.0248, -0.1139,
        -0.1416, -0.0117, -0.0010,  0.0147,  0.0085,  0.0296, -0.1146, -0.0884,
         0.0529, -0.0737, -0.0160, -0.0071, -0.0117, -0.0324,  0.0365,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0105, -0.0019,  0.0003, -0.0096, -0.0086, -0.0122, -0.0047,  0.0022,
         0.0048, -0.0057,  0.0081, -0.0092, -0.0981, -0.0965, -0.0114, -0.0034,
         0.0301, -0.0462,  0.0065,  0.0013, -0.0127, -0.0045, -0.0014, -0.0200,
         0.0034, -0.0064, -0.1527, -0.0136, -0.1246, -0.0291, -0.0656, -0.0080,
        -0.0117, -0.0279, -0.0441, -0.0180, -0.0199,  0.0080, -0.0570,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0171,  0.0165,  0.0794, -0.1724, -0.1767, -0.0506, -0.0613, -0.1651,
        -0.0885,  0.0551,  0.0284, -0.0891,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 1.0343e-01, -1.6023e-01, -2.5081e-01, -3.3045e-03, -1.5151e-02,
        -1.7765e-02, -4.1896e-02,  1.4208e-02,  4.4070e-04, -1.1849e-02,
         8.1443e-03,  1.7913e-02,  3.7459e-03, -2.4596e-03, -1.7223e-02,
         6.3900e-04, -2.1485e-02, -4.7785e-02, -2.6860e-03, -9.6425e-03,
        -1.6509e-02, -2.0227e-02,  1.2760e-02, -4.0357e-03, -6.2922e-03,
         2.1956e-03,  1.0861e-02, -9.1348e-03, -7.0767e-03,  2.4832e-03,
         2.2818e-02,  1.7915e-02, -1.2940e-02,  9.6267e-03, -8.8195e-04,
        -5.0456e-03, -3.6564e-02,  1.1818e-04,  1.7051e-03,  9.4732e-04,
         9.2935e-03,  3.3895e-03,  2.5320e-03,  1.2859e-03,  3.2552e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.0932e-02,  5.0644e-01,  4.8122e-02,  6.6700e-02, -1.0283e-01,
         1.6774e-02,  5.0869e-03,  4.6685e-02,  1.1990e-01,  1.3267e-02,
         4.1023e-04, -3.2857e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0561,  0.6538, -0.0786,  0.0976,  0.0215, -0.0214, -0.0181, -0.0531,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.2235e-03,  1.7548e-01,  1.6988e-02, -3.1549e-03,  3.4110e-02,
        -2.3587e-05,  4.3441e-03,  1.0449e-02,  4.9781e-02,  7.4997e-02,
         2.6318e-02,  1.0224e-02,  1.0465e-02,  2.6722e-03,  1.8867e-03,
        -5.0017e-04, -1.1750e-02, -1.0599e-02, -3.7703e-03, -7.0925e-03,
        -9.6401e-04, -3.5216e-03, -1.3534e-03, -5.1874e-03, -3.4437e-03,
         1.4938e-02, -5.7617e-03, -4.9858e-03,  8.3180e-03, -2.0087e-03,
         5.0946e-04, -2.4087e-03, -3.4499e-03, -3.6263e-04, -2.1198e-03,
         1.2930e-03,  9.1065e-04, -4.8107e-03, -3.7270e-03, -9.9615e-04,
        -1.0482e-02,  3.7746e-03,  6.8973e-03,  4.4549e-03, -3.1320e-02,
        -4.2972e-02,  5.4193e-02,  1.2045e-02,  3.5736e-02,  8.7039e-02,
         3.9950e-03, -3.4735e-03,  1.9211e-02, -1.2446e-02,  1.3843e-02,
        -7.6002e-03, -8.6096e-03, -6.5080e-03,  4.1253e-03, -3.4134e-02,
         6.4244e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0116, -0.0138,  0.0270, -0.0318, -0.1065, -0.0104, -0.0282,  0.0152,
         0.0097,  0.0059, -0.0130, -0.0663,  0.0053, -0.0543, -0.0388,  0.0064,
         0.0028,  0.0064, -0.0304,  0.0098, -0.1192, -0.0958, -0.0089, -0.0354,
        -0.0467, -0.0496,  0.0207, -0.0086, -0.0235, -0.0343, -0.0109,  0.0216,
        -0.0312,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0445,  0.0271,  0.1697,  0.0525,  0.0150, -0.0052,  0.0647, -0.0333,
        -0.0178, -0.0058,  0.0045,  0.0048,  0.0338,  0.0097,  0.0162, -0.0611,
        -0.0212,  0.0700,  0.2093,  0.0032, -0.0269,  0.0023, -0.0404,  0.0612,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0214, -0.0823, -0.1444, -0.0616, -0.0100, -0.0860,  0.0034, -0.0662,
        -0.0062, -0.0769, -0.0791, -0.0817, -0.1183,  0.0011,  0.0067, -0.0128,
         0.0158, -0.0185,  0.0074,  0.0135, -0.0367, -0.0199,  0.0048,  0.0254,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0490, -0.0358, -0.0905, -0.0803,  0.0085,  0.0204, -0.0155, -0.0658,
        -0.0062, -0.0060,  0.0037, -0.0127, -0.0050, -0.0005, -0.0477, -0.0939,
        -0.0746,  0.0002,  0.0077, -0.0246, -0.0972,  0.0166, -0.0293, -0.0648,
         0.0053,  0.0003, -0.0371,  0.0031,  0.0179, -0.0131, -0.0280, -0.0115,
         0.0154,  0.0118,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.5913e-02,  2.7922e-02, -1.8848e-02,  3.6167e-03, -5.8603e-02,
         3.6188e-03, -4.7258e-03, -2.3778e-04, -1.0981e-04, -1.1919e-03,
         2.6468e-02,  4.6019e-02, -2.4956e-02, -1.5035e-01,  2.9947e-02,
        -3.6659e-03, -1.4182e-01, -2.3658e-02, -6.6374e-02, -3.8013e-02,
        -7.6753e-02, -1.0364e-01,  2.9759e-02, -6.8737e-02, -2.5047e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1140, -0.0065, -0.0380, -0.0008, -0.0182, -0.0009, -0.1249, -0.0947,
        -0.0985, -0.3318,  0.0419, -0.0340, -0.0418, -0.0020, -0.0117,  0.0402,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0244,  0.2461, -0.0029,  0.1367,  0.0528,  0.0648,  0.1550,  0.0164,
        -0.0629,  0.0076, -0.0067, -0.0272,  0.0231, -0.0028, -0.0142, -0.0136,
         0.0257,  0.0651,  0.0264,  0.0122, -0.0080,  0.0040,  0.0015,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.3072e-01,  1.8348e-02,  2.3790e-03,  1.3175e-01,  3.9194e-01,
         8.3631e-02, -7.2327e-02, -3.7551e-03,  1.7370e-02, -1.5747e-02,
        -7.6172e-02, -3.3249e-02,  2.2300e-02, -3.1214e-04,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.0501,  0.1527,  0.0841,  0.1186,  0.0226,  0.0199,  0.0341, -0.0296,
         0.0258,  0.0117,  0.0977,  0.0827,  0.0411,  0.0024,  0.0047, -0.0353,
        -0.0056,  0.0703,  0.0109,  0.0433,  0.0138,  0.0197, -0.0098, -0.0134,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0859,  0.0054, -0.2518,  0.0199, -0.2149,  0.0101, -0.3795,  0.0232,
         0.0075,  0.0008, -0.0009,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0435, -0.0191,  0.0416, -0.0036,  0.0329, -0.1116, -0.1895,  0.0187,
        -0.0950, -0.0156, -0.0407,  0.0151, -0.0502, -0.0322, -0.0834, -0.0431,
        -0.0656, -0.0582,  0.0239,  0.0107,  0.0056,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0426, -0.0013,  0.0227, -0.0005,  0.0099,  0.0155,  0.0493, -0.2478,
        -0.2467, -0.2307,  0.0012,  0.0577, -0.0087, -0.0654,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0782,  0.0100,  0.0326, -0.0252,  0.0120,  0.0227,  0.0064, -0.0349,
         0.0221, -0.0523,  0.1131,  0.0581,  0.4094,  0.0237,  0.0134,  0.0309,
         0.0079, -0.0028, -0.0097, -0.0009,  0.0336,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0705, -0.3286, -0.0027, -0.1424,  0.0781, -0.0903, -0.0101, -0.1181,
         0.0135, -0.0567, -0.0213, -0.0452,  0.0224,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0040, -0.1596, -0.0250, -0.0061, -0.0129, -0.0028, -0.0431, -0.1259,
        -0.0090, -0.0102,  0.0074,  0.0232, -0.0030,  0.0047, -0.0009, -0.0396,
        -0.0081, -0.0128,  0.0022, -0.0030, -0.0076, -0.0077,  0.0031,  0.0050,
        -0.1623, -0.0112, -0.0683, -0.0638, -0.0461,  0.0009, -0.0085,  0.0063,
         0.0050,  0.0012,  0.0056,  0.0070, -0.0220, -0.0006,  0.0062, -0.0090,
        -0.0243,  0.0029,  0.0008,  0.0188, -0.0024], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.7352e-02,  6.8415e-02,  8.5905e-02,  1.8637e-02,  1.1557e-03,
        -4.3824e-03,  1.9031e-03,  4.4267e-02,  1.3315e-05, -6.2680e-03,
         8.7846e-02,  1.9048e-03, -2.4260e-03, -7.9272e-03,  5.8894e-02,
         1.2998e-02,  1.4065e-01,  1.8108e-02,  7.7538e-04,  4.7357e-03,
         8.4206e-03,  1.0953e-01,  1.1267e-02,  7.0877e-02, -8.7370e-03,
        -1.1615e-02,  3.3794e-02,  1.8529e-02,  6.3408e-02,  1.9036e-03,
        -5.3170e-03,  1.2389e-03,  2.8909e-02,  3.1891e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0235, -0.0453,  0.0206,  0.0675, -0.0039,  0.0021, -0.0059, -0.0109,
        -0.0552,  0.0121,  0.0138, -0.0052, -0.0152, -0.0171,  0.0034, -0.0348,
         0.0027, -0.0162,  0.0252,  0.3224,  0.1790, -0.0064, -0.0259, -0.0153,
         0.0225,  0.0481,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0436, -0.4259,  0.0026, -0.0472, -0.0043, -0.0229,  0.0136, -0.0380,
        -0.0301, -0.0027, -0.0040,  0.0066, -0.0034, -0.0109, -0.0410, -0.0456,
        -0.1240, -0.0115, -0.0255, -0.0212, -0.0492, -0.0042,  0.0219,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0214, -0.1075, -0.0005, -0.0431,  0.0027, -0.0021, -0.0198, -0.0456,
        -0.0661, -0.0083, -0.0008, -0.0145, -0.0081, -0.0024, -0.1023,  0.0106,
        -0.0102, -0.0195, -0.0442, -0.0131, -0.0123, -0.0026, -0.0010,  0.0219,
         0.0132, -0.0145, -0.0605, -0.1401, -0.0270, -0.0187,  0.0115, -0.0085,
         0.0147, -0.0005, -0.0204, -0.0115, -0.0207,  0.0221, -0.0229,  0.0125,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0218,  0.0249,  0.0036, -0.0236, -0.0912, -0.0139, -0.0031, -0.0608,
         0.0122, -0.0154, -0.0059,  0.0021, -0.0238,  0.0079, -0.0275, -0.0913,
        -0.0704, -0.0162, -0.0199, -0.0368, -0.0101, -0.0624, -0.0654, -0.0169,
        -0.0040,  0.0107, -0.0130, -0.0515, -0.0082, -0.0157, -0.0483, -0.0045,
        -0.0543, -0.0625,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0309, -0.0425, -0.0416, -0.0226, -0.0789, -0.0052,  0.0204, -0.0098,
         0.0017,  0.0034, -0.0081, -0.0267,  0.0231, -0.0165, -0.1008,  0.0036,
         0.0095, -0.0060, -0.0443, -0.0014,  0.0028,  0.0121, -0.0033,  0.0157,
        -0.0325, -0.0814, -0.0195, -0.0223, -0.0005,  0.0091,  0.0153,  0.0010,
         0.0056, -0.0011, -0.0749, -0.0052, -0.0211, -0.0297, -0.0056, -0.0299,
        -0.0245, -0.0130, -0.0181, -0.0043, -0.0027,  0.0002, -0.0015, -0.0001,
        -0.0109, -0.0083, -0.0272,  0.0033], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0070, -0.0143,  0.0019,  0.0045,  0.1401, -0.0039, -0.0083, -0.0186,
        -0.0218,  0.0045, -0.0031, -0.0177,  0.0167,  0.1541,  0.0312,  0.0380,
         0.1907,  0.0087,  0.1031,  0.0221,  0.0256, -0.0089,  0.0146,  0.0955,
         0.0060, -0.0005, -0.0163,  0.0113,  0.0061, -0.0003, -0.0047,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0290, -0.3122, -0.0106, -0.0328, -0.0036,  0.0040, -0.0067, -0.0079,
        -0.0122,  0.0367, -0.0049, -0.0632, -0.0101, -0.0213, -0.0547, -0.0467,
        -0.0014, -0.0718, -0.0817, -0.0867, -0.0232,  0.0036, -0.0114,  0.0233,
         0.0402,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1912,  0.0192,  0.0158, -0.0145,  0.0149, -0.2005, -0.3371, -0.0389,
         0.0568, -0.0382,  0.0213, -0.0516,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.7913e-02, -4.1424e-01, -2.5494e-02, -3.9613e-02, -7.4452e-02,
        -1.1477e-02,  2.3654e-03, -3.1562e-03,  8.3431e-03,  1.2533e-02,
        -5.1566e-04, -5.3731e-03, -4.6190e-02, -6.1336e-02, -6.0020e-02,
        -4.1714e-02,  9.1837e-03, -6.7479e-03,  4.9353e-03,  1.4151e-02,
        -1.1740e-02, -5.0255e-03, -2.8935e-02,  1.0253e-02,  1.6625e-02,
        -4.8178e-02,  7.8494e-03, -4.0792e-04, -1.1235e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0214,  0.4719,  0.0239, -0.0253,  0.0032,  0.0983,  0.2545, -0.0035,
        -0.0174,  0.0197, -0.0273,  0.0335,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.9457e-02, -3.6602e-01, -7.7041e-02, -9.9927e-02,  2.9343e-05,
        -1.3146e-01, -1.5084e-01, -1.7703e-02, -6.3483e-02, -8.8438e-03,
         4.4012e-03,  7.9845e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0820,  0.0701, -0.1927,  0.0156, -0.0651, -0.0137,  0.0099, -0.0128,
        -0.1690, -0.0159,  0.0150, -0.1284, -0.0217, -0.0178, -0.0254, -0.1004,
        -0.0010,  0.0025, -0.0028,  0.0384,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0072,  0.1898,  0.0178,  0.0633,  0.1251,  0.0234,  0.0226, -0.0103,
         0.0332,  0.0146,  0.0045,  0.0204,  0.0058,  0.0083,  0.0062,  0.0003,
         0.0211,  0.0013,  0.0088,  0.0174,  0.0431, -0.0242, -0.0120,  0.0151,
         0.0193, -0.0083,  0.0169,  0.0477,  0.0430, -0.0051, -0.0069,  0.0058,
         0.0383,  0.0853, -0.0084,  0.0163,  0.0029,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1463,  0.6932,  0.0402, -0.0010,  0.0380,  0.0112,  0.0400,  0.0266,
         0.0034,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0107,  0.0198,  0.0551, -0.1388, -0.0026,  0.0056, -0.0936, -0.0892,
         0.0112,  0.0085,  0.0188,  0.0189, -0.0759, -0.1658, -0.0209, -0.0120,
         0.0113, -0.0206, -0.0293, -0.0885,  0.0010,  0.0176, -0.0098, -0.0048,
         0.0228,  0.0429,  0.0042,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0744,  0.2901, -0.0131,  0.0834,  0.0094,  0.0146, -0.0101, -0.0073,
        -0.0004, -0.0189, -0.0188,  0.0559,  0.0175, -0.0089,  0.0023, -0.0079,
         0.0246, -0.0128, -0.0024, -0.0028, -0.0117,  0.0434,  0.0545, -0.0023,
        -0.0019,  0.0216,  0.0087,  0.0211,  0.1232,  0.0074, -0.0067,  0.0007,
        -0.0056, -0.0155,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0143,  0.0059,  0.0027,  0.0012, -0.0599, -0.0156,  0.0101,  0.0114,
         0.0119,  0.0270,  0.0072, -0.0811,  0.0026,  0.0079,  0.0062, -0.0451,
        -0.0723,  0.0032, -0.0393, -0.0066,  0.0124,  0.0025, -0.0334, -0.0090,
         0.0038,  0.0076, -0.0022, -0.0032,  0.0058, -0.0078, -0.0025, -0.0973,
        -0.1236, -0.0031, -0.0164, -0.0479, -0.0497,  0.0065, -0.0092,  0.0016,
        -0.0181, -0.0601,  0.0068,  0.0311,  0.0063,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0554, -0.0042, -0.0262, -0.0183, -0.0215, -0.1897, -0.2341, -0.0238,
        -0.0278, -0.1754, -0.0451, -0.0508, -0.1083,  0.0125, -0.0042, -0.0029,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.3132e-02,  8.9883e-03,  1.9724e-02, -1.1379e-04, -1.6280e-02,
        -2.8361e-02, -9.9234e-04, -3.6711e-02, -1.6138e-01, -1.5359e-01,
        -4.7760e-02, -1.0808e-01, -1.3889e-02, -1.2019e-01, -7.1259e-02,
         7.5316e-03, -1.6806e-02, -3.2008e-04,  8.2587e-03, -1.3773e-02,
         7.1133e-03,  2.1438e-02,  2.6404e-03, -2.9265e-02, -5.5381e-03,
         1.3182e-02,  3.6856e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0244,  0.0149, -0.0655, -0.0053, -0.0312, -0.0039, -0.0321, -0.0029,
        -0.0208, -0.0035,  0.0036, -0.0151, -0.0358,  0.0041, -0.0158, -0.0052,
        -0.0225,  0.0036,  0.0156, -0.0268,  0.0068, -0.0204, -0.0279, -0.0309,
        -0.0101,  0.0031, -0.0549, -0.0792,  0.0181, -0.0535,  0.0114, -0.0510,
        -0.0356, -0.0027, -0.0137, -0.0276, -0.0113, -0.0799, -0.0106,  0.0008,
        -0.0223, -0.0367, -0.0110, -0.0031,  0.0198, -0.0053], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.1196e-01, -4.2258e-02,  2.5354e-02, -4.8281e-02,  1.5721e-02,
         7.4351e-03,  4.2273e-03,  9.1640e-02,  2.3669e-02,  7.2804e-03,
         9.3374e-03,  1.3238e-02,  1.2989e-03, -2.6409e-02,  1.3170e-02,
         1.2596e-03, -3.1749e-04, -3.7242e-05,  9.9425e-02,  2.1437e-01,
        -1.7190e-02,  5.3600e-02,  7.2750e-04,  2.7806e-02,  1.5907e-02,
        -4.3275e-03,  1.0096e-01,  1.0982e-02,  9.3039e-03, -2.5016e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1147,  0.0066, -0.0362,  0.0146, -0.0396, -0.0489,  0.0968,  0.3385,
         0.0998,  0.0574,  0.1471,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0891, -0.3288, -0.0232, -0.0299, -0.0424, -0.0166, -0.0310, -0.1021,
        -0.0039, -0.0455, -0.0082, -0.0075,  0.0426, -0.0040, -0.0132, -0.0741,
        -0.0136, -0.0545, -0.0083, -0.0615,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0152, -0.2854, -0.1838, -0.1675, -0.0277, -0.0723, -0.1335,  0.0554,
         0.0137,  0.0391,  0.0063,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0447, -0.2630, -0.1271, -0.1240, -0.0364, -0.0096, -0.0365, -0.0576,
        -0.0911,  0.0041, -0.0099, -0.0179, -0.0299, -0.0169, -0.0122, -0.0330,
        -0.0154, -0.0043, -0.0116, -0.0041,  0.0508,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1013, -0.2204, -0.1456,  0.0348, -0.0405, -0.1994,  0.0053,  0.0197,
        -0.0653, -0.0213, -0.0201, -0.0366,  0.0109, -0.0041,  0.0034, -0.0149,
        -0.0206,  0.0357,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0026, -0.1095, -0.1821, -0.0269,  0.0032,  0.0064,  0.0137,  0.0060,
         0.0023, -0.0056, -0.0034,  0.0081,  0.0033, -0.0478, -0.0550, -0.0016,
        -0.0318, -0.0554, -0.0253, -0.0066, -0.0155, -0.0450, -0.1009, -0.0129,
        -0.0236,  0.0112, -0.0474, -0.0146, -0.0146, -0.0567, -0.0023, -0.0021,
         0.0048,  0.0029, -0.0077,  0.0056,  0.0021,  0.0275,  0.0059,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.2402e-02, -3.5746e-01,  5.6841e-03, -8.3945e-02, -1.0672e-02,
        -8.3343e-04,  3.8475e-03,  4.4593e-03, -1.6407e-02, -2.5383e-02,
        -3.7827e-02, -8.1560e-02,  3.1482e-03,  4.9062e-04,  4.0078e-03,
         1.9938e-03, -3.2371e-02,  1.0384e-02, -1.1524e-02,  2.8416e-03,
        -3.7826e-03,  2.2446e-03,  4.0866e-03,  7.0549e-03,  6.5550e-03,
         3.6528e-03, -1.9007e-02, -4.5446e-03, -5.8354e-02, -1.2284e-01,
        -2.3171e-04, -2.0069e-02, -1.1712e-02,  8.6252e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.1126,  0.0038, -0.2380, -0.3322, -0.0467, -0.0929, -0.0195,  0.0335,
        -0.1208,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0090, -0.1478,  0.0028, -0.0649, -0.0152, -0.1050,  0.0141, -0.0385,
        -0.0473,  0.0207, -0.0430, -0.0318, -0.0165, -0.0062, -0.0039, -0.0927,
        -0.0124, -0.0932, -0.0040,  0.0034,  0.0113,  0.0021,  0.0069,  0.0102,
        -0.0344, -0.0030, -0.0089,  0.0062, -0.0071,  0.0003,  0.0077, -0.0322,
         0.0132, -0.0246,  0.0056, -0.0011,  0.0103, -0.0003,  0.0300,  0.0122,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0223, -0.0932, -0.1194, -0.0272, -0.0747, -0.0195,  0.0062, -0.0083,
        -0.0550, -0.0409, -0.0034,  0.0063,  0.0004, -0.0083,  0.0058, -0.0014,
         0.0078, -0.0123, -0.0292,  0.0084, -0.0092, -0.0031,  0.0079, -0.0209,
        -0.0029, -0.0204, -0.0416, -0.0065, -0.0527, -0.0872,  0.0014, -0.0054,
        -0.0010,  0.0047,  0.0048,  0.0042, -0.0288, -0.0603,  0.0002,  0.0082,
         0.0050,  0.0029,  0.0071,  0.0061,  0.0043,  0.0090, -0.0080,  0.0027,
        -0.0014, -0.0217,  0.0043, -0.0057,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0320, -0.2577, -0.0359, -0.0553, -0.0162, -0.0293, -0.0842, -0.0371,
        -0.0725, -0.0016, -0.0131, -0.0702, -0.0575, -0.0113, -0.0424, -0.0544,
        -0.0442, -0.0064, -0.0404,  0.0133, -0.0211,  0.0039,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0433, -0.5451, -0.0051, -0.0165,  0.0037, -0.0055,  0.0237, -0.0650,
        -0.0247, -0.0132,  0.0092, -0.0320,  0.0166, -0.0205, -0.0078, -0.0622,
        -0.0260,  0.0251,  0.0101,  0.0282,  0.0165,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4690e-02, -1.9231e-01, -3.2517e-02, -4.3164e-02, -2.4544e-02,
        -1.6454e-02, -3.7425e-02,  4.2869e-03, -2.6337e-02, -8.8115e-03,
        -6.1159e-04, -5.3208e-03,  6.4671e-04,  1.6258e-03,  3.2651e-05,
        -1.6424e-03,  3.3917e-03, -1.2862e-03, -1.7602e-02,  6.5631e-03,
         2.9379e-03,  1.1853e-02, -2.1397e-03, -3.8683e-03, -3.0505e-03,
        -4.5535e-03, -2.3000e-02, -2.6657e-02, -2.0745e-03, -4.3801e-02,
        -6.6979e-02, -4.9285e-02, -2.5913e-03, -2.4338e-02, -1.4117e-02,
         2.8388e-03,  1.1960e-02, -1.9459e-03,  1.5467e-02,  4.4500e-03,
        -5.7374e-02, -4.6582e-03, -1.7228e-03,  1.1046e-02, -2.5391e-02,
         2.1301e-03, -4.2321e-02, -2.6749e-02, -1.2316e-03,  1.0449e-02,
        -2.2294e-02, -3.8251e-03, -4.1064e-03,  2.3851e-02, -5.6777e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0166, -0.0162,  0.0241, -0.0193, -0.0130, -0.0274, -0.0796, -0.0007,
         0.0044,  0.0041,  0.0115,  0.0041,  0.0096, -0.0209, -0.0665, -0.0098,
         0.0023, -0.0207, -0.0121, -0.0768, -0.0234, -0.0145, -0.0404,  0.0081,
         0.0022,  0.0134, -0.0053, -0.0837, -0.0245,  0.0044, -0.0053,  0.0053,
        -0.0110, -0.0896,  0.0003, -0.0235,  0.0058, -0.0493, -0.0457, -0.0379,
        -0.0102, -0.0090, -0.0234,  0.0205,  0.0033,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0200, -0.2288, -0.0799, -0.0932, -0.0050, -0.0177, -0.0315, -0.0135,
        -0.0955, -0.0007,  0.0095, -0.0496, -0.0184, -0.0780, -0.0458,  0.0018,
        -0.0162, -0.0172, -0.0260, -0.0281, -0.0454,  0.0109, -0.0135,  0.0539,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0214,  0.0108,  0.0040,  0.0147, -0.0093, -0.0093, -0.0602, -0.0546,
         0.0129,  0.0053,  0.0050,  0.0016, -0.0086, -0.0832, -0.0024,  0.0405,
        -0.0646, -0.0189, -0.0337, -0.0753, -0.0149, -0.0108, -0.0622, -0.0911,
         0.0044, -0.0544, -0.0571,  0.0095, -0.0270, -0.0694, -0.0069, -0.0197,
        -0.0365,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0271, -0.3606, -0.0326, -0.1347, -0.0236, -0.0129, -0.0083,  0.0007,
        -0.1375, -0.0149,  0.0141, -0.0084,  0.0351, -0.0766, -0.0021, -0.0462,
        -0.0484,  0.0012, -0.0149,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0200, -0.0291,  0.0225, -0.0437, -0.1011, -0.0116, -0.0507, -0.0747,
        -0.1705, -0.0074, -0.0097,  0.0117, -0.0215, -0.0212, -0.0021, -0.0025,
         0.0177, -0.1057, -0.0683,  0.0089, -0.0758, -0.0959, -0.0109,  0.0149,
        -0.0016,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0452, -0.2298, -0.1044, -0.0205,  0.0223, -0.1095, -0.1513,  0.0828,
        -0.0236, -0.0160,  0.0106,  0.0016, -0.0536,  0.0047, -0.0075, -0.0317,
        -0.0848,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-3.8537e-02,  1.8754e-02, -9.6923e-03, -1.7696e-04, -8.6651e-03,
        -5.7675e-03,  5.6351e-03, -6.3920e-03,  1.1822e-02, -9.5578e-02,
        -2.1044e-02, -1.0436e-02,  2.6701e-02, -4.7018e-03,  7.8839e-03,
         3.9866e-03, -1.1375e-01, -1.2271e-01, -1.9984e-02, -6.4240e-02,
        -3.7421e-02, -1.0963e-01, -1.1354e-01,  4.7424e-04, -2.2922e-02,
         3.8187e-05,  7.4170e-03,  1.6916e-02, -1.7102e-02, -1.9098e-03,
        -2.5346e-02,  5.0827e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0022, -0.0827, -0.1801, -0.0046, -0.0108,  0.0105, -0.0505, -0.1356,
        -0.0019,  0.0110,  0.0121, -0.0026, -0.0322,  0.0063, -0.0165,  0.0029,
        -0.0277, -0.0088, -0.0590, -0.0087, -0.0709, -0.0053, -0.0563, -0.1134,
        -0.0215, -0.0416, -0.0084, -0.0158,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0680,  0.0114, -0.0019, -0.0058,  0.0091,  0.1184,  0.1823,  0.0203,
         0.0310,  0.0518,  0.0022,  0.0755, -0.0095, -0.0125,  0.0086,  0.0860,
         0.0061,  0.0736,  0.0144, -0.0111, -0.0115,  0.0073,  0.0431,  0.0165,
         0.0630,  0.0099, -0.0119,  0.0146, -0.0229,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.2332e-02, -9.3085e-02, -3.6280e-03, -4.0113e-02, -1.2600e-01,
        -5.0622e-03, -7.1705e-05, -5.9286e-03,  5.6003e-05,  1.1281e-02,
        -5.0253e-02,  1.6893e-04, -4.1224e-02, -1.0380e-01,  4.4114e-02,
        -2.5805e-02,  4.3626e-03, -8.9109e-02, -1.4270e-02, -2.3601e-02,
         5.7677e-03, -3.5504e-03, -2.0178e-02, -1.1741e-02, -2.0460e-02,
        -3.3276e-02, -3.9214e-02, -6.7502e-03,  1.6191e-03, -1.3923e-03,
         1.2756e-02, -1.0196e-01,  1.7066e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7923e-02, -3.0079e-01, -2.2040e-02, -7.8635e-02, -2.8151e-03,
        -6.4968e-02,  3.5515e-02, -3.0019e-03, -2.2650e-03, -6.1107e-02,
        -4.8543e-02, -2.5571e-02,  1.3787e-02,  2.6234e-03, -1.1786e-03,
         1.2600e-02, -6.0887e-02,  7.0616e-03, -1.0917e-01, -2.5857e-03,
        -8.0435e-02, -2.2695e-04, -1.4755e-02, -3.1516e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0322, -0.1458, -0.1976,  0.0148, -0.0648,  0.0047, -0.0308, -0.0536,
         0.0235, -0.0455, -0.0055, -0.0339, -0.0407, -0.0121, -0.0112,  0.0134,
         0.0175, -0.0573,  0.0015, -0.1292, -0.0574, -0.0013,  0.0057,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0278, -0.1519,  0.0159, -0.0060, -0.0292, -0.0661,  0.0171,  0.0099,
         0.0159, -0.0084, -0.0044,  0.0029, -0.0117,  0.0069, -0.0005, -0.0055,
        -0.0615, -0.0340, -0.0388, -0.0135, -0.0057, -0.0183, -0.0017, -0.0168,
        -0.0505, -0.0201, -0.0065,  0.0057, -0.0106, -0.0669, -0.0643, -0.0224,
        -0.0283, -0.0172, -0.0397, -0.0543, -0.0023, -0.0226,  0.0182,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0006, -0.0239, -0.0011, -0.0304, -0.0929, -0.0040,  0.0087,  0.0058,
        -0.0290,  0.0129,  0.0028,  0.0038,  0.0063,  0.0013,  0.0185, -0.0314,
         0.0069, -0.0027,  0.0045, -0.0200, -0.0687, -0.0102,  0.0067, -0.0520,
        -0.0974, -0.0932, -0.0189, -0.0315, -0.0051, -0.0669, -0.0214, -0.0194,
        -0.0287, -0.0154, -0.0006, -0.0043, -0.0053, -0.0880,  0.0158, -0.0035,
        -0.0121,  0.0207, -0.0068,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.8333e-02, -2.0554e-01, -7.7774e-03,  4.9400e-03, -1.5022e-02,
         1.5276e-02, -1.3125e-02,  2.7060e-03, -5.8646e-02, -1.1221e-02,
         6.7529e-03,  5.9311e-03, -2.5009e-02, -1.2592e-04,  2.2909e-03,
        -8.0199e-03, -2.9513e-02, -5.8234e-02, -3.8836e-03, -1.8379e-03,
        -1.1452e-03,  8.3285e-03, -2.3911e-03,  1.8241e-03,  5.6214e-03,
         4.7708e-03, -3.9653e-03,  3.2523e-03,  9.6490e-03,  1.0384e-02,
         4.1499e-03, -3.6830e-02, -6.3060e-02, -1.3233e-02, -2.1052e-03,
         1.6868e-04, -2.7120e-02, -2.0610e-02, -3.9429e-02, -5.5248e-03,
        -3.9115e-02, -4.8255e-02, -1.0262e-02,  1.8814e-03, -2.4526e-03,
         2.7808e-02, -1.0289e-03,  3.3045e-03,  2.1734e-04, -4.7056e-03,
         1.5660e-02,  6.6253e-03, -2.1289e-02, -4.2903e-03, -1.6653e-02,
         1.1816e-02,  6.8902e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0006,  0.4292, -0.0198,  0.0193,  0.0456,  0.0123,  0.1239,  0.1133,
         0.0100,  0.0459,  0.1055,  0.0392, -0.0251, -0.0104,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0131, -0.2447, -0.0507,  0.0076,  0.0024, -0.0243,  0.0060, -0.0022,
        -0.0453,  0.0032, -0.0492, -0.1212,  0.0189, -0.0278, -0.0552, -0.0109,
        -0.0331, -0.0213,  0.0166, -0.0524, -0.0966, -0.0192, -0.0339,  0.0048,
         0.0142,  0.0016,  0.0005,  0.0097, -0.0032,  0.0102,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0688, -0.3804, -0.0408, -0.2424, -0.1325,  0.0331, -0.0894, -0.0125,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0123, -0.2065, -0.0295, -0.0285, -0.0577, -0.1128, -0.0035,  0.0178,
         0.0034, -0.0145,  0.0084, -0.0036, -0.0337, -0.0878, -0.0058, -0.0011,
         0.0101, -0.0061,  0.0002, -0.0021,  0.0218, -0.0072,  0.0037, -0.0009,
        -0.0748, -0.0267, -0.0347, -0.0208, -0.0700, -0.0472, -0.0029,  0.0371,
        -0.0066,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5980e-02, -2.2129e-01, -5.1501e-02, -1.8987e-02, -1.2613e-03,
        -1.2696e-02, -4.4415e-03, -3.9228e-05, -1.9999e-04, -1.4486e-02,
        -5.7274e-03, -1.2430e-02, -2.9738e-02,  7.9000e-03,  1.3882e-02,
         4.1164e-03, -1.2390e-04,  9.6557e-03, -1.4086e-02,  5.2076e-03,
         6.1810e-03,  1.2484e-02, -2.6376e-03,  7.1345e-03,  6.7614e-03,
        -1.0326e-02, -1.0446e-02, -5.9952e-02,  3.4648e-03, -6.7835e-03,
        -1.3863e-03, -9.7263e-03, -3.0070e-02, -6.2643e-02, -1.4206e-03,
        -3.8528e-03, -2.7957e-02, -1.0827e-03,  1.6348e-03,  5.5832e-03,
         1.1181e-04, -5.9994e-04, -3.1380e-03, -3.1131e-02, -3.8066e-03,
        -3.4476e-03,  2.2299e-03,  3.1746e-03, -6.4454e-04, -1.0114e-04,
         4.1924e-03, -2.5071e-02,  5.4184e-04, -2.5961e-02, -6.9547e-03,
         6.3429e-03,  1.6395e-03,  5.4772e-03, -7.9424e-04, -1.5732e-02,
        -4.3224e-03, -1.7321e-02, -5.1882e-02,  1.2179e-04,  2.6901e-03,
         2.2196e-03, -4.0025e-03,  1.0358e-02,  1.2769e-02,  4.8737e-03,
         3.3075e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0663, -0.0360, -0.0269,  0.0019, -0.0456,  0.0020, -0.1056, -0.0102,
         0.0027,  0.0099, -0.0007,  0.0030,  0.0095,  0.0123, -0.0066,  0.0011,
        -0.0019,  0.0126, -0.0172, -0.1231, -0.0272,  0.0188,  0.0029, -0.0451,
         0.0080, -0.0453, -0.0104,  0.0094,  0.0060,  0.0059, -0.0031, -0.0404,
        -0.0756, -0.0886, -0.0100, -0.0091, -0.0601, -0.0190,  0.0024, -0.0127,
         0.0021, -0.0027,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.3176e-02, -2.3666e-01,  2.9277e-03, -2.5116e-02,  1.0427e-03,
        -8.5098e-03,  1.1904e-02,  2.0161e-02, -5.7770e-03, -3.9142e-02,
        -1.0031e-03, -3.6521e-02, -6.3660e-02,  1.9591e-03, -1.5208e-02,
        -6.4854e-03, -1.5631e-02, -6.5139e-02, -9.0004e-02, -9.1065e-03,
        -7.1103e-03, -2.7408e-02, -5.9007e-03,  6.5396e-03, -2.6328e-03,
        -5.1993e-02, -1.2363e-02, -5.4108e-03, -1.0019e-01, -1.0518e-02,
         5.2111e-03, -8.7453e-04, -7.9414e-03, -1.6007e-02, -1.6120e-02,
        -8.9474e-05, -2.4559e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0819, -0.4593, -0.0517, -0.1156, -0.0196,  0.0329,  0.0076, -0.0265,
        -0.0395, -0.0031, -0.0556,  0.0225, -0.0389, -0.0215,  0.0242,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0084, -0.0820,  0.0111, -0.0032, -0.1433, -0.1794, -0.0134, -0.0990,
        -0.2129, -0.0188,  0.0233,  0.0480, -0.0330,  0.0022,  0.0243, -0.0075,
        -0.0231, -0.0144,  0.0503,  0.0024,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0260, -0.0059,  0.0028, -0.0058, -0.1245, -0.0168,  0.0139, -0.0089,
         0.0167,  0.0072, -0.0543, -0.0559, -0.0288, -0.0093, -0.0429, -0.0361,
         0.0041, -0.0223,  0.0140, -0.0309, -0.0093, -0.0004, -0.0073,  0.0018,
         0.0139, -0.0666, -0.0919, -0.0089,  0.0066, -0.0268, -0.0668, -0.0028,
         0.0009,  0.0004, -0.0349, -0.0608,  0.0077, -0.0313, -0.0039, -0.0061,
        -0.0067,  0.0136, -0.0035,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0228, -0.0065, -0.0262,  0.0086, -0.0439, -0.0067, -0.0223, -0.0974,
        -0.0359,  0.0015,  0.0035,  0.0177,  0.0081, -0.0968, -0.1268,  0.0004,
         0.0097,  0.0007, -0.0017, -0.0030, -0.0554, -0.0106,  0.0051, -0.0017,
        -0.0027, -0.0080, -0.0216, -0.0090, -0.0633, -0.0112, -0.0020, -0.0568,
        -0.0116, -0.0515, -0.0804, -0.0231,  0.0172,  0.0286,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0365, -0.2898, -0.0820, -0.1083, -0.0190, -0.0670, -0.0140,  0.0020,
        -0.0435, -0.0951, -0.0068, -0.0244, -0.0617, -0.0158, -0.0062, -0.0007,
        -0.0170, -0.0527,  0.0025,  0.0173,  0.0321,  0.0056,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.9804e-02, -1.5711e-01, -1.9220e-02,  3.0416e-03,  2.7314e-02,
        -5.8272e-03, -1.2550e-02,  6.0830e-03, -1.9769e-02, -7.4926e-02,
        -1.0081e-02,  9.7839e-03, -5.3981e-03,  5.7560e-03, -3.4010e-02,
        -7.1184e-02, -1.0852e-01, -1.6834e-02, -1.3043e-02, -5.1982e-02,
         7.5942e-03, -1.6874e-04, -2.2530e-02, -4.1638e-03, -1.1237e-02,
        -3.6264e-05,  1.3055e-03,  9.5458e-03, -8.6177e-03, -4.2948e-02,
        -2.1802e-03,  4.3891e-03,  1.2933e-02, -1.6956e-02, -2.4013e-02,
        -6.2441e-02,  1.3021e-02, -3.0810e-02,  3.3216e-02, -1.9654e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0246, -0.0232, -0.0007,  0.0035, -0.0033, -0.0391,  0.0115,  0.0142,
        -0.0541, -0.0060, -0.0002, -0.0039, -0.0283, -0.2194, -0.1204, -0.0054,
         0.0101, -0.0060, -0.0807, -0.0046, -0.0037,  0.0104,  0.0173, -0.0322,
        -0.0084, -0.0573, -0.0149, -0.0794, -0.0781, -0.0108,  0.0242,  0.0043,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0295, -0.2743, -0.0161, -0.0065, -0.0166,  0.0051, -0.0054, -0.0757,
        -0.1332, -0.0157,  0.0270, -0.0346, -0.0121, -0.0021,  0.0083, -0.0044,
        -0.0629, -0.0122,  0.0015,  0.0010, -0.0025,  0.0031, -0.0048, -0.0927,
        -0.0081,  0.0160, -0.0459, -0.0169,  0.0063, -0.0200, -0.0039, -0.0078,
        -0.0098,  0.0176,  0.0004,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0091, -0.0488, -0.2269, -0.1169,  0.0136, -0.0403, -0.1470, -0.1291,
         0.0048, -0.0902,  0.0092, -0.0378, -0.0076, -0.0371, -0.0208, -0.0009,
        -0.0167, -0.0025,  0.0044, -0.0074,  0.0289,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0049, -0.4282, -0.0187, -0.0892, -0.1181, -0.0555, -0.0097, -0.0248,
        -0.0578,  0.0089, -0.0093,  0.0047,  0.0105,  0.0080, -0.0746,  0.0086,
        -0.0561,  0.0115, -0.0010,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0372, -0.1107, -0.0861, -0.0324, -0.0366,  0.0063,  0.0137,  0.0124,
        -0.0032, -0.1349, -0.1530, -0.0233, -0.0508, -0.1198, -0.0113, -0.0826,
         0.0041, -0.0461, -0.0064, -0.0292,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0248, -0.3904,  0.0140, -0.0711, -0.0092, -0.0172, -0.0072, -0.0687,
        -0.0138, -0.0234, -0.0207, -0.0129,  0.0012, -0.0446, -0.0205, -0.0208,
        -0.0731, -0.0108, -0.0162, -0.0528,  0.0145, -0.0501, -0.0220,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0018, -0.2078, -0.0696, -0.0028, -0.0164, -0.0626, -0.0174, -0.0195,
        -0.0726, -0.1226, -0.0247, -0.0704, -0.0149, -0.0542,  0.0149,  0.0095,
         0.0063,  0.0003, -0.0310, -0.0075, -0.0388,  0.0075,  0.0288, -0.0290,
         0.0688,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.2105e-04, -3.3754e-01, -2.7665e-02, -5.7271e-02, -1.4835e-02,
        -9.4043e-03, -7.7569e-03, -2.7466e-02,  4.3762e-02, -2.9828e-02,
        -5.6055e-02,  1.1202e-03,  6.0721e-04,  5.3242e-03,  1.4253e-02,
        -1.6398e-02,  7.8089e-05,  1.8903e-03,  9.2677e-03, -1.3121e-02,
        -5.6135e-02, -7.1443e-02, -1.8199e-02, -4.3389e-02, -1.6778e-02,
        -8.1271e-03,  2.0532e-03, -6.7737e-03, -1.0559e-02, -2.0186e-02,
        -1.7469e-02,  6.1835e-03, -1.8961e-02,  1.4688e-03, -2.8206e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0092, -0.0142, -0.0170, -0.1047, -0.0172, -0.0278, -0.0771, -0.0213,
        -0.1344, -0.0022, -0.0241, -0.0154, -0.0008, -0.0457, -0.1074, -0.1356,
        -0.0053, -0.1279,  0.0529,  0.0174, -0.0298,  0.0127,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0213, -0.0167,  0.0702, -0.0061,  0.0023,  0.0457, -0.0107, -0.0012,
        -0.0343,  0.4748,  0.0379,  0.0243,  0.0040,  0.0134, -0.0567,  0.1300,
         0.0027,  0.0092,  0.0143,  0.0100, -0.0142,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0089, -0.4424, -0.0148, -0.0444, -0.0225,  0.0021, -0.0997, -0.1076,
        -0.0097,  0.0229,  0.0150,  0.0259, -0.0071, -0.0550, -0.0097, -0.0411,
         0.0463,  0.0081,  0.0167,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0164, -0.0138, -0.0790, -0.0015, -0.0053, -0.0758,  0.0075, -0.0047,
         0.0086, -0.0041, -0.0071, -0.0202, -0.0036, -0.0165,  0.0111,  0.0008,
        -0.0500, -0.0750,  0.0008, -0.0289, -0.0127, -0.0297, -0.0272,  0.0017,
        -0.0443,  0.0093, -0.0029, -0.0252, -0.0018, -0.0900, -0.0139, -0.0304,
        -0.0356, -0.1089,  0.0045, -0.0138,  0.0065,  0.0031,  0.0037, -0.0396,
         0.0105,  0.0085,  0.0018, -0.0079,  0.0061,  0.0201, -0.0098],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.0033e-03, -2.2318e-01, -1.4415e-01, -1.5548e-02,  1.7483e-02,
        -9.9528e-03, -7.7409e-03,  5.2643e-03, -5.9089e-02,  9.6441e-03,
         8.9973e-03, -1.9548e-04,  9.2129e-03, -3.2038e-02,  1.2338e-02,
        -6.9155e-03, -1.1206e-01,  3.0506e-04,  2.7847e-03, -1.5429e-02,
        -2.0976e-02, -2.2098e-02, -2.7922e-02, -9.9427e-02,  1.4001e-02,
         8.3514e-03, -9.3161e-03, -5.7091e-03,  1.2208e-02,  1.3412e-02,
         2.0354e-02,  1.3227e-02, -2.3060e-03, -3.1361e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.2984e-04, -4.0974e-01, -2.0705e-02,  2.3850e-02, -1.4070e-02,
        -3.9643e-02, -3.2063e-03,  7.1648e-03,  2.6747e-03, -2.1171e-02,
        -1.6058e-02, -8.2370e-03, -5.3671e-02, -1.0273e-01,  3.2155e-03,
        -3.5053e-04,  6.1942e-03,  4.3629e-03,  1.4649e-02, -7.5177e-02,
        -5.0372e-03, -6.7446e-03, -6.3244e-03,  6.1419e-03, -1.5252e-02,
         8.6750e-03, -9.1427e-03, -2.3012e-02, -6.0686e-03, -5.4548e-02,
        -1.1837e-02, -6.9763e-03,  1.8691e-05,  1.3123e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.1119, -0.3889,  0.0062,  0.0302, -0.0187, -0.0822,  0.0214, -0.0256,
         0.0028, -0.0954, -0.0379, -0.0123, -0.0301, -0.0690, -0.0263, -0.0125,
         0.0090, -0.0195,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0617,  0.0133,  0.0153, -0.0181,  0.0046,  0.0041,  0.0014, -0.0024,
        -0.1416, -0.2528,  0.0093, -0.0118,  0.0047,  0.0073,  0.0163,  0.0157,
         0.0048, -0.0193,  0.0569, -0.2136, -0.0107,  0.0283, -0.0261,  0.0027,
         0.0087, -0.0214,  0.0270,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.8369e-02,  7.0291e-03, -6.3210e-03,  1.3634e-04, -3.0108e-02,
        -2.4908e-02, -2.1889e-03, -7.3773e-02, -1.8689e-01,  1.5381e-03,
         2.5051e-02, -1.3519e-02, -2.0030e-01, -3.2601e-02,  1.0403e-02,
        -1.4586e-03, -2.3749e-02, -1.4425e-02, -1.6586e-03, -2.0722e-01,
        -5.4735e-03,  1.2832e-02, -4.2807e-02,  1.2714e-03, -5.3474e-02,
        -2.4921e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0070, -0.2617,  0.0026, -0.0769, -0.1064, -0.0213, -0.0130, -0.0208,
        -0.0647, -0.0435, -0.0596, -0.0284,  0.0121, -0.0132, -0.0274, -0.0717,
        -0.0416, -0.0413, -0.0016, -0.0018,  0.0835,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0113, -0.2964, -0.1365, -0.1856, -0.0193, -0.0104, -0.0965, -0.0051,
        -0.0264, -0.0491, -0.0115, -0.0677,  0.0217,  0.0100, -0.0524,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1434,  0.0212,  0.0044,  0.0129,  0.0750,  0.0513,  0.0174,  0.0199,
         0.0237,  0.0152, -0.0804, -0.0212,  0.2403,  0.0917,  0.1143, -0.0111,
        -0.0215, -0.0352,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0064, -0.3177, -0.1218, -0.1087, -0.0086, -0.0187,  0.0087, -0.0187,
         0.0107, -0.0348, -0.0925, -0.0573,  0.0503, -0.0328, -0.0089, -0.0210,
         0.0227, -0.0137,  0.0460,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.6630e-02, -1.5954e-01, -1.1096e-01,  1.7129e-02, -1.8251e-02,
        -2.9436e-02,  1.8958e-02,  1.1789e-02, -1.2278e-02, -2.3535e-03,
        -2.6224e-03, -4.6322e-03, -1.6810e-02,  6.2973e-03, -1.8034e-03,
         7.5767e-03, -5.5387e-02, -3.7859e-03,  5.0402e-03,  8.5992e-03,
        -6.4479e-02,  3.9142e-03, -3.4223e-02, -4.0221e-03, -4.6292e-02,
        -2.9672e-03, -2.2575e-02,  1.7079e-02,  3.2328e-03, -3.6162e-02,
        -1.0071e-05,  2.6206e-02, -3.9590e-02, -2.1452e-03, -3.0732e-02,
        -2.4899e-02, -4.5819e-02,  2.6983e-04,  3.3478e-03, -8.3069e-03,
         7.3169e-03, -3.3467e-03,  1.1451e-02, -3.1730e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.9673e-02, -1.6551e-01,  1.2753e-03, -1.0523e-01, -1.3297e-01,
         4.6862e-03, -6.8952e-02, -6.8202e-03, -6.9784e-02,  2.0231e-03,
        -2.2272e-02,  3.2967e-02,  7.0024e-03, -2.5003e-03, -6.2502e-03,
         5.5826e-03,  3.4526e-02, -3.1491e-02, -2.9575e-03, -2.3049e-02,
        -4.4320e-02,  2.8472e-03, -3.9823e-03,  1.0306e-02,  1.5451e-02,
         9.5447e-04, -9.6172e-03, -8.6084e-02, -3.0580e-03,  1.5424e-02,
         1.6499e-04, -2.8223e-02,  2.7368e-03,  4.7105e-04, -1.0833e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.1041e-02, -3.7117e-01, -1.5324e-02, -5.4375e-02, -1.4790e-02,
         1.5289e-02, -8.3495e-02, -4.8268e-02, -1.9093e-02, -3.6967e-03,
        -5.7958e-03, -4.2496e-02, -1.3343e-01, -1.4643e-04, -1.5824e-02,
        -7.8922e-02, -3.3789e-03, -2.1881e-03,  9.6005e-03,  2.1080e-03,
         9.6334e-03,  1.3251e-02, -9.9124e-04,  3.5693e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0725, -0.2505, -0.0074, -0.0436, -0.0093, -0.0072,  0.0128, -0.0297,
        -0.0865, -0.0120, -0.0265,  0.0082, -0.0008, -0.0432, -0.0480, -0.0018,
         0.0239,  0.0026, -0.0054, -0.0064,  0.0025, -0.0358,  0.0121, -0.0428,
         0.0055, -0.0532, -0.0489, -0.0190, -0.0118, -0.0174, -0.0272, -0.0095,
         0.0081, -0.0082,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.0939e-02,  1.4734e-02, -1.0336e-02, -1.0909e-02,  1.8760e-02,
        -3.5206e-03, -1.7921e-01, -4.6333e-02,  1.1631e-02,  2.4282e-04,
         2.2143e-02,  2.1526e-02, -1.2238e-01, -1.0576e-03, -1.1325e-01,
        -1.3684e-01, -2.6434e-02,  2.2632e-02,  8.7695e-03,  7.6200e-03,
         1.1836e-02, -1.6258e-02, -3.0306e-03, -7.7070e-02, -2.0173e-02,
         1.3563e-04, -2.6501e-02, -3.2360e-02,  1.3696e-02, -9.6687e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.1139,  0.0290, -0.0011, -0.3000, -0.0659, -0.0357, -0.0257, -0.0194,
        -0.1713, -0.1438, -0.0122,  0.0123,  0.0697,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0205,  0.0269, -0.0316,  0.0396,  0.0011, -0.0456, -0.0666, -0.2094,
        -0.2539,  0.0164,  0.0203, -0.0647,  0.0119, -0.0456, -0.0583,  0.0218,
         0.0428, -0.0229,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0882, -0.1995, -0.3501, -0.0284, -0.2420,  0.0173, -0.0633, -0.0112,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0275, -0.2240, -0.0165, -0.0243, -0.0442, -0.0254, -0.0106,  0.0009,
         0.0065, -0.0132, -0.0055,  0.0207, -0.0203, -0.0095, -0.0026, -0.0101,
         0.0040, -0.0013,  0.0054, -0.0018,  0.0107,  0.0059, -0.0475, -0.0452,
        -0.0059, -0.0247, -0.0018,  0.0013, -0.0612, -0.0127,  0.0149, -0.0369,
        -0.0105, -0.0703,  0.0005, -0.0198, -0.0356, -0.0077,  0.0071,  0.0023,
        -0.0142, -0.0076, -0.0148,  0.0090,  0.0112,  0.0453,  0.0010],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1538, -0.2643,  0.0831, -0.2266, -0.0271, -0.0162, -0.0672, -0.0329,
        -0.0106, -0.0841,  0.0342,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0059, -0.2948, -0.0527, -0.0363, -0.0457, -0.0126, -0.0436, -0.1241,
         0.0112, -0.0573,  0.0111, -0.0067, -0.0756, -0.0221,  0.1682,  0.0078,
        -0.0021, -0.0221,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0695, -0.1682, -0.3329, -0.0194,  0.0232, -0.0023, -0.1856,  0.0113,
        -0.0893, -0.0598,  0.0042, -0.0017, -0.0241,  0.0060,  0.0025,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1000, -0.3414, -0.0426,  0.0034,  0.0303, -0.0762, -0.0922, -0.0294,
        -0.0266, -0.0958, -0.1083, -0.0066, -0.0079,  0.0065,  0.0328,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0040, -0.0064, -0.0049, -0.0337,  0.0130, -0.0060,  0.0016, -0.0141,
         0.0808,  0.0085, -0.0167,  0.0925,  0.1323,  0.0271, -0.0076,  0.0443,
         0.0439, -0.0210,  0.0064, -0.0101,  0.1621,  0.1819, -0.0363,  0.0257,
         0.0002, -0.0189,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.3941e-01,  1.4664e-02,  5.1930e-02,  3.1779e-02, -1.8638e-02,
         1.4833e-02,  1.8232e-02, -1.8221e-01, -3.8316e-01, -2.6970e-02,
         3.4105e-02, -1.2804e-02,  2.8646e-02,  4.2582e-02,  4.3424e-05,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0323, -0.2505,  0.0096, -0.1006, -0.0939, -0.0098, -0.0759, -0.0016,
        -0.0009, -0.1171, -0.0078, -0.0138, -0.0420, -0.0100, -0.0101,  0.0051,
        -0.0291, -0.0111, -0.0872, -0.0226, -0.0210,  0.0048,  0.0170,  0.0022,
         0.0132,  0.0109,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0643, -0.0069,  0.0184,  0.0216, -0.0093, -0.0288,  0.0168,  0.0458,
        -0.0185, -0.0575, -0.0948, -0.0266, -0.0196,  0.0206, -0.1343, -0.1354,
        -0.0100, -0.0462, -0.0071, -0.0719, -0.0200, -0.0281, -0.0178, -0.0368,
         0.0224, -0.0205,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-4.8167e-02, -3.1705e-01,  2.2394e-02,  5.0592e-03, -1.5037e-02,
        -1.3254e-02, -3.5179e-02, -7.1641e-02, -7.4292e-03, -4.1024e-02,
        -1.6019e-02, -1.9661e-02, -7.1842e-02,  6.4482e-03, -2.8151e-02,
         1.2398e-03,  8.6819e-03, -3.7263e-03, -2.4598e-02,  1.6529e-03,
        -8.1683e-03, -2.1588e-02,  1.9157e-04, -2.5173e-02, -6.5105e-02,
         8.3562e-03, -4.6162e-02, -9.4549e-03,  4.8119e-03,  5.6674e-03,
        -2.9824e-03, -3.0800e-02,  1.5249e-03, -2.0223e-03, -3.7529e-03,
         5.9868e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0297, -0.1515, -0.0531, -0.0092, -0.0512,  0.0013, -0.0646, -0.0788,
         0.0054,  0.0210, -0.0306, -0.0002, -0.0098,  0.0090, -0.0029, -0.0141,
         0.0006, -0.0031,  0.0115, -0.0097,  0.0008, -0.0378, -0.0154, -0.0092,
        -0.0465, -0.0403, -0.0071, -0.0669, -0.0638,  0.0074, -0.0187, -0.0424,
        -0.0561,  0.0122,  0.0123,  0.0057,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0830, -0.0406,  0.0203, -0.0165, -0.0095, -0.0901, -0.0038,  0.0306,
        -0.0220,  0.0025, -0.0531, -0.1678, -0.2492, -0.0457, -0.0262, -0.0153,
         0.0148, -0.0406,  0.0099, -0.0584,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7938e-02, -2.0686e-01, -3.4233e-02, -4.1618e-02,  1.2288e-02,
        -3.5891e-03,  6.9854e-03, -4.3639e-02, -5.9668e-03, -3.2140e-02,
         2.5082e-04,  2.6709e-03, -1.7060e-03,  4.4261e-03, -8.1531e-02,
        -6.9805e-02, -5.1296e-03, -6.5788e-02, -1.5726e-02, -7.7913e-03,
        -1.3133e-03, -5.0174e-03,  9.8189e-03,  3.3445e-02,  1.8807e-02,
         6.3440e-03, -2.4117e-03, -5.3098e-03,  1.0181e-02, -3.2048e-02,
        -7.4050e-02, -1.9181e-02, -1.4695e-02, -7.9070e-02,  9.3676e-03,
        -3.7057e-03, -4.3627e-05,  1.5109e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.9754e-02, -1.8017e-01, -1.2310e-01, -4.4266e-02, -1.5421e-03,
        -8.9334e-03, -6.5047e-03,  1.4751e-02,  2.2293e-02,  7.6233e-03,
        -4.7147e-02, -6.0267e-02,  1.0069e-03,  9.6816e-03, -6.0938e-03,
        -4.0292e-02,  5.2259e-03, -9.3643e-02, -7.2946e-03,  7.9449e-04,
        -3.1014e-04,  8.7458e-03,  1.8826e-02, -5.8370e-03,  5.7620e-03,
        -4.7047e-03, -2.4636e-02, -3.3590e-02, -8.5451e-02,  9.4870e-03,
        -9.1453e-03, -1.1233e-02, -1.0699e-03,  9.0926e-03, -2.6073e-02,
        -1.2543e-05,  6.4945e-04,  9.5543e-03, -2.0226e-03,  2.6764e-02,
         6.6554e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0320, -0.3563, -0.0241,  0.0064,  0.0060, -0.1035, -0.0008, -0.0046,
        -0.0355, -0.0026,  0.0059, -0.0680, -0.0841,  0.0217,  0.0322, -0.0504,
        -0.0039, -0.0850, -0.0084,  0.0341,  0.0187, -0.0158,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0505,  0.0047,  0.0028,  0.0070,  0.0181, -0.0197,  0.0134,  0.0038,
        -0.0013,  0.0044, -0.0168, -0.0568, -0.0239, -0.0414, -0.0042, -0.0718,
         0.0279, -0.0627, -0.0067, -0.0668, -0.0867,  0.0135,  0.0385, -0.0555,
         0.0015,  0.0111, -0.0139, -0.0463, -0.0794, -0.0309, -0.0509, -0.0335,
        -0.0184,  0.0034,  0.0116,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.2560e-02, -2.6071e-02, -2.6149e-01, -2.5597e-03, -8.5969e-02,
        -1.5072e-02,  8.7287e-03,  6.7993e-03,  5.2550e-03,  2.1974e-04,
        -3.6281e-02,  4.5188e-02,  1.0512e-02,  3.4894e-03,  8.6320e-03,
        -3.2931e-03,  6.5923e-02,  1.8450e-03, -8.4460e-02,  3.8269e-03,
        -6.1616e-02,  9.4897e-03, -3.0529e-02, -6.1384e-02,  6.2736e-03,
         1.5693e-02, -8.4362e-02,  4.2960e-03,  2.8185e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0004, -0.3659, -0.0739, -0.0221, -0.0221, -0.0585,  0.0031, -0.0094,
        -0.0244,  0.0148, -0.0305, -0.0396,  0.0186, -0.0078,  0.0075, -0.0517,
        -0.0671, -0.0152, -0.0421,  0.0041, -0.0463, -0.0327,  0.0021, -0.0020,
         0.0383,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0403, -0.0138,  0.0041,  0.0258, -0.1930,  0.0066,  0.0121, -0.0927,
        -0.1279, -0.0254,  0.0446,  0.0438,  0.0044,  0.0197, -0.0774, -0.1186,
         0.0324, -0.0272,  0.0017,  0.0124, -0.0377,  0.0142, -0.0243,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0016,  0.0144,  0.0118, -0.0123,  0.0157,  0.0132,  0.0097, -0.0046,
        -0.0112,  0.0014, -0.0227, -0.0085, -0.1068, -0.1534,  0.0062, -0.0314,
        -0.0178, -0.0463, -0.0014,  0.0013,  0.0171, -0.0152, -0.0100,  0.0107,
         0.0057, -0.0012, -0.1236, -0.0090, -0.0702,  0.0142, -0.0612, -0.0116,
        -0.0151, -0.0125, -0.0649,  0.0215, -0.0285,  0.0016, -0.0147,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0400,  0.0229,  0.0098, -0.1494, -0.2806, -0.0028, -0.0687, -0.1228,
        -0.1808, -0.0471, -0.0324, -0.0430,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 2.3389e-02, -7.2590e-02, -2.5662e-01, -1.8366e-02, -2.0946e-02,
        -9.2921e-03, -4.2737e-02,  1.5614e-02, -1.2921e-02, -3.2106e-02,
        -3.6247e-03,  6.7958e-03, -3.8411e-02, -2.6285e-02, -4.4801e-02,
         8.7627e-03, -5.6916e-03, -4.4894e-02, -7.7842e-03, -1.4661e-02,
        -4.2482e-03, -2.8811e-02,  7.0891e-03,  1.5504e-03,  3.3303e-03,
        -5.9682e-03,  2.4805e-02, -8.5121e-04,  4.8744e-03, -6.0756e-03,
         2.8622e-03,  1.3472e-02, -5.4178e-02,  6.2521e-03,  3.6306e-03,
        -1.0431e-04, -7.4468e-02, -7.1475e-03, -7.7954e-04,  4.7182e-03,
         4.3584e-03,  3.7426e-03,  1.2546e-02,  1.0979e-02, -6.8651e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0030, -0.5396, -0.0429, -0.1267,  0.0218, -0.0062,  0.0097, -0.0279,
        -0.1354, -0.0069, -0.0363,  0.0436,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0012, -0.5496, -0.0228, -0.2150, -0.0269,  0.0182, -0.0945,  0.0718,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.1453e-02, -1.9169e-01, -5.3695e-03,  8.9138e-03, -3.4542e-02,
         1.0912e-03,  1.2716e-02, -1.1166e-02, -2.5945e-02, -6.6677e-02,
        -6.5863e-03, -1.3155e-02, -6.0765e-03, -2.5636e-04, -7.6431e-04,
        -1.0730e-03,  7.1419e-03, -9.9152e-03,  1.3900e-03, -3.4087e-03,
         4.9527e-03,  1.7045e-03, -1.2667e-02, -2.5760e-03, -6.1263e-03,
        -1.0154e-02, -1.2154e-03, -8.6319e-03, -2.1225e-02, -1.1490e-04,
        -5.1316e-03, -1.1548e-02,  3.8524e-03, -2.3334e-03, -1.2688e-02,
         1.8299e-05,  8.6167e-03, -2.9204e-03, -5.3721e-03,  1.3961e-03,
        -7.3157e-03, -3.0846e-03, -5.9787e-03, -3.6963e-03, -1.1841e-02,
        -4.6960e-03, -5.5000e-02, -6.6357e-03, -6.7610e-02, -9.0701e-02,
        -2.6412e-02, -6.7973e-03, -3.9755e-02,  4.5812e-03, -1.2764e-03,
        -6.9554e-03, -1.1722e-02, -7.1648e-03,  1.9514e-03,  1.9013e-02,
         5.5243e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0240, -0.0041,  0.0025, -0.0600, -0.1049,  0.0003, -0.0663, -0.0037,
        -0.0050,  0.0007, -0.0066, -0.0774, -0.0048, -0.0900, -0.1040,  0.0116,
        -0.0135,  0.0040, -0.0523, -0.0347, -0.0603, -0.0785, -0.0314, -0.0028,
        -0.0271, -0.0277, -0.0039, -0.0128, -0.0243, -0.0136, -0.0194,  0.0019,
        -0.0260,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0309, -0.0242, -0.1989, -0.1142,  0.0069,  0.0255, -0.0664, -0.0092,
         0.0091, -0.0194, -0.0069,  0.0029, -0.0621,  0.0008, -0.0019,  0.0402,
        -0.0114, -0.0522, -0.1756,  0.0185, -0.0109, -0.0435,  0.0430, -0.0254,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0073, -0.1321, -0.1249, -0.0728,  0.0074, -0.1028, -0.0047, -0.0799,
        -0.0066, -0.0264, -0.0871, -0.0431, -0.1230, -0.0071,  0.0081, -0.0050,
         0.0072, -0.0025,  0.0425,  0.0328, -0.0445, -0.0092, -0.0100,  0.0130,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0255, -0.0059, -0.0893, -0.1563, -0.0090, -0.0021, -0.0026, -0.0381,
        -0.0051,  0.0257,  0.0048, -0.0055,  0.0078,  0.0030, -0.0451, -0.0749,
        -0.0610,  0.0009, -0.0052, -0.0217, -0.0730,  0.0060, -0.0569, -0.0483,
         0.0110,  0.0004, -0.0672, -0.0146,  0.0117, -0.0311, -0.0029, -0.0195,
         0.0181,  0.0496,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.2531e-02, -9.6035e-03, -7.8883e-05,  5.6383e-03, -7.3253e-02,
        -2.3318e-02, -5.2426e-03,  1.6535e-03, -8.9770e-03, -1.5479e-02,
        -2.8191e-02,  2.6848e-04,  3.4301e-03, -1.8612e-01,  2.0449e-02,
        -6.9616e-02, -1.7099e-01, -1.3091e-02, -6.7783e-02, -2.5467e-03,
        -6.3058e-02, -1.4814e-01,  2.8640e-02, -2.6001e-02, -1.5900e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0024, -0.0083, -0.0570, -0.0219, -0.0475, -0.0101, -0.2006, -0.1078,
        -0.1014, -0.2716, -0.0299, -0.0346, -0.0260, -0.0015, -0.0354, -0.0439,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0143, -0.2546, -0.0245, -0.1126, -0.0378, -0.0789, -0.1658, -0.0366,
         0.0373, -0.0118,  0.0031,  0.0125, -0.0204,  0.0085,  0.0100,  0.0043,
        -0.0215, -0.0757, -0.0344, -0.0097,  0.0091, -0.0069, -0.0097,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0415, -0.0620,  0.0640, -0.2630, -0.3448, -0.0086,  0.0535,  0.0249,
        -0.0055,  0.0025, -0.0681,  0.0147, -0.0015,  0.0452,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.0270, -0.1055, -0.0847, -0.1346, -0.0157,  0.0017, -0.0610,  0.0283,
        -0.0375, -0.0321, -0.0856, -0.1233, -0.0120, -0.0028,  0.0050,  0.0146,
        -0.0183, -0.0555, -0.0049, -0.0596, -0.0088,  0.0060, -0.0600,  0.0155,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0227, -0.0880, -0.2404, -0.0269, -0.1120, -0.0099, -0.2754, -0.0271,
        -0.0485, -0.0732, -0.0761,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0225,  0.0208, -0.0018,  0.0315,  0.0337, -0.0954, -0.2067, -0.0250,
        -0.1506,  0.0024, -0.0766, -0.0014, -0.0568,  0.0280, -0.0574, -0.0005,
        -0.0365, -0.1155,  0.0017, -0.0074, -0.0280,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0015, -0.0031,  0.0037,  0.0691, -0.0042,  0.0263, -0.0013, -0.3253,
        -0.0602, -0.2178, -0.0438, -0.0559,  0.0731, -0.1148,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.8285e-02,  4.6494e-02,  4.2646e-02,  2.6144e-02,  2.5435e-02,
         3.3296e-02,  6.4166e-03, -2.4468e-02, -1.1318e-02,  2.2673e-02,
         7.6170e-02, -1.0042e-01,  3.6015e-01, -1.1587e-02,  1.0262e-02,
        -2.6156e-04,  4.5799e-02, -7.5138e-03, -2.8821e-03, -5.2861e-02,
         4.9147e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0191, -0.4143,  0.0123, -0.1606,  0.0333, -0.0796,  0.0035, -0.0914,
        -0.0047, -0.0412, -0.0471, -0.0914, -0.0016,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0373, -0.1815,  0.0057, -0.0014,  0.0154, -0.0065, -0.0348, -0.0622,
        -0.0154,  0.0147,  0.0029, -0.0134, -0.0125, -0.0028,  0.0052, -0.0172,
         0.0049, -0.0043, -0.0045, -0.0108,  0.0061,  0.0238,  0.0100, -0.0256,
        -0.1026,  0.0363, -0.0393, -0.0745, -0.0614, -0.0041, -0.0102,  0.0004,
         0.0065, -0.0066,  0.0010, -0.0010, -0.0437,  0.0024, -0.0029, -0.0135,
        -0.0159,  0.0002,  0.0075, -0.0379, -0.0134], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0283, -0.0443, -0.0643, -0.0007, -0.0070,  0.0067,  0.0155, -0.0267,
        -0.0091, -0.0006, -0.0675, -0.0022, -0.0137,  0.0094, -0.0597, -0.0433,
        -0.1409,  0.0029,  0.0071, -0.0090,  0.0077, -0.0742,  0.0185, -0.0632,
         0.0027, -0.0037, -0.0249, -0.0250, -0.1416, -0.0137, -0.0104, -0.0171,
        -0.0373,  0.0009,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0494,  0.0326,  0.0101, -0.0269, -0.0030, -0.0112,  0.0134,  0.0209,
         0.0621,  0.0072, -0.0112, -0.0166, -0.0133,  0.0026, -0.0007,  0.0091,
        -0.0060,  0.0233,  0.0009, -0.2286, -0.3721,  0.0007, -0.0026, -0.0006,
        -0.0557,  0.0194,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.2387e-03, -3.7092e-01, -8.0616e-03, -1.2726e-01,  1.6002e-03,
        -1.5672e-02, -1.7747e-02, -7.1821e-02, -7.6978e-03, -1.9515e-02,
         2.8366e-02, -2.6800e-04, -5.7851e-03,  1.2314e-03, -4.0915e-02,
        -7.9924e-02, -9.2031e-02, -5.3791e-03, -4.3182e-02, -1.8547e-02,
        -2.7336e-02,  7.4614e-03,  2.0348e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0098, -0.1762, -0.0222, -0.0238,  0.0052,  0.0029, -0.0089, -0.0773,
        -0.0798, -0.0134,  0.0121,  0.0027,  0.0173, -0.0095, -0.1370,  0.0194,
        -0.0034, -0.0310, -0.0561,  0.0061,  0.0068, -0.0046,  0.0060,  0.0052,
         0.0110,  0.0009, -0.0420, -0.0946, -0.0022, -0.0140, -0.0142,  0.0101,
        -0.0007,  0.0077,  0.0033,  0.0009, -0.0103,  0.0197, -0.0051,  0.0267,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0015, -0.0181, -0.0117, -0.0469, -0.1374, -0.0184,  0.0538, -0.1044,
        -0.0072, -0.0388, -0.0051, -0.0028, -0.0429, -0.0180, -0.0321, -0.0682,
        -0.0620, -0.0052, -0.0016, -0.0423, -0.0097, -0.0458, -0.0308,  0.0033,
        -0.0141, -0.0110, -0.0115, -0.0478,  0.0002,  0.0048, -0.0629, -0.0040,
        -0.0313,  0.0043,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 7.8264e-02,  9.1327e-03, -4.4088e-02, -2.2158e-02, -8.0967e-02,
        -5.5656e-04,  1.9338e-02,  6.9937e-03,  2.3239e-04,  6.8700e-03,
         8.5167e-03, -2.3510e-02,  2.5894e-03, -4.6796e-03, -1.5912e-01,
         2.2998e-02, -7.0122e-06,  5.1266e-03, -5.8068e-02, -1.1844e-02,
         2.1404e-03, -1.8675e-03,  4.7832e-03, -7.6740e-04, -5.3170e-02,
        -1.0449e-01,  6.7195e-03,  2.3827e-02,  8.6905e-03,  9.9709e-03,
         1.1629e-02,  5.5563e-04,  1.2598e-02, -3.2058e-03, -2.5718e-02,
        -9.6100e-03, -1.1022e-02, -4.2960e-02, -4.6974e-03, -8.0928e-03,
        -1.7280e-02, -5.0785e-03, -1.8963e-02, -7.7678e-04,  1.1157e-03,
        -1.2163e-03,  1.0949e-03, -1.1016e-03, -1.4750e-02, -1.4913e-02,
         6.2724e-03, -5.8629e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0465,  0.0119,  0.0113, -0.0008, -0.1509, -0.0154, -0.0334, -0.0033,
        -0.0039, -0.0013,  0.0098,  0.0130, -0.0069, -0.1177, -0.0102, -0.1090,
        -0.1274, -0.0119, -0.0556,  0.0128, -0.0216, -0.0140, -0.0382, -0.0907,
        -0.0066, -0.0085,  0.0300, -0.0126, -0.0087,  0.0153, -0.0010,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0374, -0.1068,  0.0037, -0.0078,  0.0123,  0.0208, -0.0072,  0.0142,
        -0.0141,  0.0426,  0.0042, -0.0505,  0.0180,  0.0128, -0.0399, -0.0538,
         0.0127, -0.0573, -0.1599, -0.1461, -0.0558, -0.0466, -0.0255, -0.0288,
         0.0214,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1540, -0.0281,  0.0129, -0.0587, -0.0160, -0.2235, -0.3134, -0.0186,
        -0.0008,  0.0486,  0.1049, -0.0205,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.3199e-03, -2.8946e-01, -2.8143e-02, -9.3629e-02, -1.0104e-01,
        -7.2676e-04,  2.0732e-02, -1.1685e-03,  4.6455e-03,  1.2829e-03,
         4.6089e-03, -1.6395e-02, -6.5990e-02, -9.1985e-02, -2.5481e-02,
        -7.1135e-02, -2.4974e-02, -2.9525e-03, -4.6307e-03,  2.1249e-04,
        -2.5396e-02, -6.8554e-03, -2.6798e-02,  6.6033e-03,  1.2105e-02,
        -4.9748e-02, -6.9636e-04,  1.1694e-02, -5.5984e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0716, -0.3808,  0.0310,  0.0396,  0.0211, -0.1309, -0.1393,  0.0398,
        -0.0010,  0.0643, -0.0657, -0.0149,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0528, -0.3894, -0.0617, -0.1043,  0.0374, -0.1305, -0.1024,  0.0132,
        -0.0722, -0.0089, -0.0148,  0.0125,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1191,  0.0653, -0.1500,  0.0028, -0.1150,  0.0198,  0.0113, -0.0183,
        -0.1193, -0.0098,  0.0040, -0.1126,  0.0038, -0.0011, -0.0558, -0.1182,
        -0.0175,  0.0072,  0.0229, -0.0263,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0104, -0.2131, -0.0038, -0.0626, -0.1299, -0.0135, -0.0212, -0.0029,
        -0.0324, -0.0061, -0.0128, -0.0437, -0.0282, -0.0146, -0.0128, -0.0047,
        -0.0375,  0.0161,  0.0007,  0.0037, -0.0173,  0.0152,  0.0003, -0.0007,
        -0.0099,  0.0054, -0.0248, -0.0541, -0.0308, -0.0006, -0.0028, -0.0082,
        -0.0468, -0.0579,  0.0080, -0.0248,  0.0217,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0633, -0.5768, -0.0462,  0.1612, -0.0406, -0.0040, -0.0366,  0.0387,
         0.0326,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0718,  0.0185, -0.0132, -0.1761, -0.0071, -0.0571, -0.0964, -0.0604,
         0.0098, -0.0032,  0.0060, -0.0526, -0.0580, -0.1134,  0.0339, -0.0086,
        -0.0137,  0.0221, -0.0318, -0.0429,  0.0028,  0.0057, -0.0205,  0.0023,
         0.0150,  0.0441, -0.0129,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.2862e-02, -2.7428e-01,  1.1709e-03, -7.8706e-02, -1.8454e-02,
        -3.5328e-02,  1.0877e-05,  1.0317e-03, -1.6464e-03,  6.7458e-03,
         4.2640e-02, -6.8660e-02, -5.8035e-03, -1.9967e-04, -6.3479e-03,
        -1.8844e-03, -9.0129e-02, -6.7814e-03, -1.5133e-02, -3.1696e-03,
         8.2425e-03, -4.9275e-02, -4.7537e-02,  4.1695e-03, -3.5167e-04,
        -1.0542e-02, -3.8542e-03, -4.6794e-02, -8.2301e-02, -1.3667e-03,
         1.3801e-02,  1.8143e-03, -4.5094e-02,  3.8742e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0793, -0.0030, -0.0044,  0.0086, -0.0551, -0.0261, -0.0006,  0.0096,
         0.0087,  0.0205, -0.0035, -0.0773, -0.0024,  0.0012,  0.0034, -0.0476,
        -0.0689,  0.0111, -0.0247, -0.0100,  0.0152, -0.0028, -0.0371,  0.0004,
         0.0025, -0.0070, -0.0012, -0.0100,  0.0020, -0.0114,  0.0121, -0.0620,
        -0.1019, -0.0059, -0.0181, -0.0370, -0.0591,  0.0005,  0.0119, -0.0105,
        -0.0207, -0.0739,  0.0108, -0.0140, -0.0059,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0003,  0.0256, -0.0099, -0.0098, -0.0334, -0.2213, -0.2694, -0.0183,
        -0.0098, -0.1332,  0.0031, -0.0498, -0.1491, -0.0469, -0.0041, -0.0158,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0935,  0.0052,  0.0199, -0.0147,  0.0084, -0.0047, -0.0202,  0.0048,
         0.1671,  0.1895,  0.0330,  0.1530, -0.0348,  0.0418,  0.0554,  0.0142,
         0.0163,  0.0026, -0.0163, -0.0033,  0.0064,  0.0057,  0.0105,  0.0096,
         0.0274, -0.0147, -0.0271,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.3029e-02,  1.2965e-02, -6.0591e-02,  2.8976e-03, -5.9103e-02,
        -1.5027e-02, -2.6710e-02, -8.7930e-03, -2.3914e-02, -7.3936e-03,
         1.3721e-05, -1.0640e-02, -3.4074e-02,  3.4029e-04, -2.1683e-02,
        -6.8337e-03, -3.0155e-02,  1.3791e-02, -1.8529e-03, -2.8305e-02,
         2.4788e-03, -1.7242e-02, -1.4860e-02, -6.0032e-02, -4.5019e-03,
        -8.5455e-03, -3.0750e-02, -7.5785e-02, -6.9164e-03, -3.7255e-02,
        -1.1359e-02, -4.3473e-02, -4.3486e-02,  2.0642e-03, -1.5653e-02,
        -7.1087e-02, -2.9737e-02, -4.3140e-02,  4.9496e-03,  1.0250e-02,
        -1.0778e-02, -2.6961e-02, -4.3491e-03,  8.2698e-04,  6.8354e-03,
        -8.5730e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.1376e-02, -2.1091e-02, -2.6953e-02, -1.7915e-02, -2.5967e-02,
         2.2526e-03, -2.7380e-02,  9.7007e-02,  2.0563e-03, -6.2313e-03,
        -3.5119e-02,  1.8111e-02, -1.6094e-03, -9.6935e-03, -3.4538e-02,
         5.5542e-04,  3.6305e-03,  3.3850e-02,  9.5991e-02,  2.9722e-01,
        -7.5903e-04, -2.2595e-02, -3.8958e-05,  1.8500e-02, -2.9568e-03,
         1.9934e-02,  6.1946e-02, -1.6054e-02, -2.0509e-03, -6.6177e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2043, -0.0670, -0.0837, -0.0123,  0.0275,  0.0141,  0.0574,  0.3872,
         0.0712,  0.0433,  0.0321,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0655, -0.2622, -0.0077, -0.0331, -0.0718,  0.0240, -0.0596, -0.1024,
         0.0103, -0.0290,  0.0102,  0.0039, -0.0551, -0.0155, -0.0194, -0.1001,
        -0.0302, -0.0134,  0.0262,  0.0603,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1146, -0.3301, -0.2137, -0.0654,  0.0529, -0.0563, -0.0922,  0.0078,
        -0.0226, -0.0330,  0.0115,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0141,  0.2579,  0.1506,  0.1845,  0.0160, -0.0188, -0.0063,  0.0606,
         0.0838,  0.0191, -0.0022,  0.0123, -0.0022,  0.0235, -0.0118,  0.0456,
        -0.0074, -0.0178,  0.0144, -0.0035, -0.0476,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0230, -0.3582, -0.1540, -0.0064,  0.0145, -0.1961,  0.0092, -0.0412,
        -0.0382,  0.0058, -0.0235, -0.0512, -0.0173, -0.0100, -0.0128, -0.0050,
         0.0296,  0.0040,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.0489e-02, -1.0282e-01, -1.9044e-01, -1.6776e-02,  2.7817e-03,
         1.0737e-02, -2.1394e-03,  1.6182e-03, -8.3834e-05, -3.0448e-03,
         2.9637e-03,  8.2367e-03,  5.9655e-03, -3.3738e-02, -3.9459e-02,
         2.4063e-02, -2.5477e-02, -3.7553e-02, -2.5110e-02, -3.0407e-02,
        -7.5144e-03, -3.1049e-02, -7.9019e-02, -2.1375e-03, -3.1554e-02,
        -2.8062e-02, -3.7064e-02, -1.2476e-02, -4.3897e-02, -5.9475e-02,
         2.3350e-03, -1.1006e-02,  8.0215e-03,  3.9406e-03, -1.2847e-02,
        -2.9117e-04, -1.8610e-03,  4.1911e-02,  1.6378e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0047, -0.3147, -0.0281, -0.0532, -0.0057, -0.0039,  0.0097,  0.0189,
        -0.0004, -0.0175, -0.0131, -0.0991,  0.0037,  0.0015, -0.0047, -0.0011,
        -0.0632, -0.0049,  0.0096, -0.0069, -0.0078, -0.0060,  0.0041,  0.0030,
        -0.0079, -0.0018, -0.0346, -0.0390, -0.0647, -0.0901, -0.0136, -0.0181,
        -0.0131, -0.0317,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.0340, -0.1490, -0.2437, -0.2827,  0.0097, -0.0939,  0.0556, -0.0735,
         0.0579,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.4458e-02, -2.0783e-01, -2.8197e-02, -6.8737e-02, -2.9815e-02,
        -7.8405e-02, -1.5553e-03, -2.7114e-02, -3.8149e-02, -1.5193e-02,
        -1.6925e-02, -1.4757e-02, -7.9671e-03, -1.7924e-02, -1.4245e-02,
        -4.9811e-02, -2.8291e-02, -9.4722e-02, -2.3347e-02, -1.6164e-02,
        -2.7211e-03, -6.9941e-05, -2.3873e-03,  3.7364e-03, -3.4689e-02,
        -4.5803e-03,  3.3087e-03, -3.2188e-03, -8.5888e-03, -1.5349e-04,
        -2.0440e-02, -4.4090e-02,  4.0138e-03, -4.0245e-02,  5.0565e-04,
        -7.5255e-03, -2.6887e-03,  1.5820e-02, -1.1450e-03, -6.4622e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.0960e-03, -3.4398e-02, -1.7602e-01, -9.9930e-03, -6.0340e-02,
        -1.0732e-02, -1.9611e-03,  1.1144e-03, -3.7869e-02, -8.1006e-02,
         1.1706e-02,  5.0817e-03,  2.5613e-03,  2.9106e-03, -8.7261e-03,
         3.6560e-03,  2.8309e-03,  3.9109e-03, -3.1237e-02,  1.0254e-02,
        -4.5596e-02,  3.1087e-03, -3.2623e-03, -1.4612e-02, -2.0430e-03,
        -2.2597e-02, -3.6200e-02,  2.2556e-03, -4.9890e-02, -6.9134e-02,
        -7.6800e-04, -1.7428e-02, -3.1005e-03,  7.9960e-03,  2.7431e-03,
         1.7173e-02, -6.7241e-02, -6.2057e-02, -1.7936e-03,  8.7070e-03,
         7.0614e-03,  4.8313e-03, -1.1938e-05,  2.1262e-03, -1.8430e-03,
         2.9811e-03, -3.1778e-03,  2.6848e-03,  3.8993e-03,  9.4097e-03,
         5.5285e-05, -2.6808e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0148, -0.2973, -0.0021, -0.0496, -0.0084, -0.0158, -0.0981, -0.0055,
        -0.0293, -0.0355, -0.0090, -0.0264, -0.0843,  0.0051, -0.0463, -0.0533,
        -0.0514, -0.0039, -0.0737,  0.0107,  0.0239,  0.0556,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0533,  0.4227, -0.0238,  0.0227, -0.0032,  0.0084, -0.0244,  0.0823,
         0.0313,  0.0043,  0.0344,  0.0817,  0.0180, -0.0248, -0.0023,  0.0347,
         0.0308, -0.0009, -0.0302, -0.0126,  0.0533,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0158, -0.2212, -0.0489, -0.0250, -0.0005, -0.0026, -0.0157,  0.0069,
        -0.0339, -0.0078,  0.0032, -0.0017, -0.0020,  0.0041,  0.0093, -0.0007,
         0.0071,  0.0028, -0.0296,  0.0040,  0.0089, -0.0058, -0.0024,  0.0046,
         0.0007, -0.0074, -0.0095, -0.0295, -0.0004, -0.0286, -0.0673, -0.0512,
         0.0011, -0.0261, -0.0415, -0.0171,  0.0124,  0.0092,  0.0083, -0.0034,
        -0.0516, -0.0104, -0.0112, -0.0009, -0.0369,  0.0146, -0.0179, -0.0158,
        -0.0033,  0.0119, -0.0209, -0.0108, -0.0045, -0.0053,  0.0060],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0747, -0.0031,  0.0289, -0.0225, -0.0028, -0.0305, -0.0794, -0.0095,
        -0.0019,  0.0086,  0.0001,  0.0151,  0.0030, -0.0246, -0.0591, -0.0200,
        -0.0075, -0.0338, -0.0016, -0.0597,  0.0006, -0.0373, -0.0566,  0.0082,
         0.0274,  0.0031, -0.0091, -0.1137,  0.0055, -0.0045,  0.0155, -0.0049,
        -0.0211, -0.0700, -0.0056, -0.0130, -0.0030, -0.0292, -0.0253, -0.0278,
        -0.0045,  0.0005,  0.0098,  0.0147, -0.0026,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0345, -0.2482, -0.0858, -0.0823, -0.0205, -0.0228, -0.0190, -0.0069,
        -0.0622, -0.0085,  0.0063, -0.0369,  0.0014, -0.0728, -0.0638, -0.0095,
        -0.0248, -0.0301, -0.0224, -0.0337, -0.0442,  0.0194, -0.0040,  0.0397,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0125,  0.0222,  0.0179,  0.0201,  0.0146,  0.0034, -0.0185, -0.0507,
         0.0042,  0.0048, -0.0023,  0.0035, -0.0083, -0.0603, -0.0031,  0.0463,
        -0.0498, -0.0414, -0.0521, -0.0632, -0.0172,  0.0034, -0.0750, -0.1033,
        -0.0498, -0.0331, -0.0630, -0.0161, -0.0289, -0.0699, -0.0068, -0.0211,
        -0.0132,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0170,  0.4255,  0.0164,  0.1029, -0.0048, -0.0043,  0.0179,  0.0114,
         0.1313, -0.0038, -0.0042,  0.0013, -0.0553,  0.0688,  0.0203,  0.0710,
         0.0099, -0.0038,  0.0301,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0568, -0.0314,  0.0167, -0.0071, -0.1311, -0.0141, -0.0528, -0.0916,
        -0.1504,  0.0034,  0.0062,  0.0496,  0.0089,  0.0259, -0.0043, -0.0010,
        -0.0242, -0.0845, -0.0451, -0.0255, -0.0672, -0.0434, -0.0275,  0.0161,
        -0.0151,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0019, -0.2951, -0.0958, -0.0149, -0.0496, -0.1351, -0.2016,  0.0259,
        -0.0317, -0.0483,  0.0078, -0.0160, -0.0354, -0.0048,  0.0047, -0.0176,
         0.0138,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 4.5473e-02,  1.4308e-02, -1.9589e-02, -7.6036e-05,  1.8039e-04,
        -8.7230e-03, -8.2288e-04, -1.5106e-02, -1.3764e-03, -5.8657e-02,
        -2.6026e-02,  1.4689e-03,  2.3385e-02,  2.2508e-03,  6.3022e-03,
        -2.5805e-02, -7.6090e-02, -1.5637e-01,  2.7644e-02, -7.7451e-02,
         1.2224e-02, -9.8500e-02, -1.3409e-01, -1.9550e-02, -7.8446e-03,
        -9.5394e-03, -5.9877e-03,  6.3553e-03,  1.3225e-02, -1.3410e-02,
        -3.4169e-02,  5.7997e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0081, -0.1218, -0.1501, -0.0220,  0.0150, -0.0052, -0.0560, -0.0913,
        -0.0094,  0.0057,  0.0004,  0.0105,  0.0719, -0.0159,  0.0038,  0.0080,
        -0.0241, -0.0222, -0.0495, -0.0113, -0.0558, -0.0151, -0.0491, -0.1236,
        -0.0018, -0.0073, -0.0100, -0.0352,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0147,  0.0345,  0.0150, -0.0125,  0.0117,  0.1084,  0.1724,  0.0463,
        -0.0103, -0.0133,  0.0006,  0.0632,  0.0132,  0.0314, -0.0087,  0.1299,
        -0.0123,  0.0612, -0.0083,  0.0025, -0.0006, -0.0067,  0.0137,  0.0316,
         0.0511, -0.0251, -0.0377, -0.0401, -0.0229,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.2968e-03, -1.3533e-01, -1.3203e-02, -4.8924e-02, -1.1349e-01,
         2.6760e-02, -2.3025e-02,  8.0995e-03, -5.9943e-03,  7.4963e-03,
         2.9073e-02,  7.2800e-05, -1.9074e-02, -1.0962e-01,  5.0134e-02,
        -3.4987e-02,  8.7500e-03, -1.1096e-01, -1.2923e-02, -1.5743e-02,
        -7.5152e-03,  2.6225e-03, -3.0010e-02, -5.7550e-04, -1.8504e-02,
        -5.0870e-02, -4.3779e-02,  1.5287e-03, -4.2185e-03, -1.0657e-02,
        -9.0016e-04, -2.6150e-02,  2.2719e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0696, -0.2902, -0.0163, -0.0890, -0.0051, -0.0678,  0.0118, -0.0235,
        -0.0087, -0.0379, -0.0593, -0.0034, -0.0041,  0.0047, -0.0050,  0.0117,
        -0.0448,  0.0127, -0.0624,  0.0003, -0.1102, -0.0063,  0.0332,  0.0219,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0313, -0.2314, -0.1359,  0.0026, -0.0441,  0.0052, -0.0381, -0.0744,
         0.0111, -0.0501,  0.0077, -0.0373, -0.0919, -0.0213, -0.0162,  0.0037,
         0.0210, -0.0360, -0.0094, -0.0290, -0.0299,  0.0357, -0.0367,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0717, -0.1678, -0.0131, -0.0207, -0.0212, -0.0395,  0.0054,  0.0023,
         0.0183,  0.0007, -0.0006, -0.0123, -0.0232,  0.0029, -0.0013, -0.0226,
        -0.0637, -0.0466, -0.0391,  0.0096, -0.0004, -0.0429, -0.0028, -0.0182,
        -0.0457, -0.0118, -0.0073, -0.0099, -0.0205, -0.0590, -0.0309, -0.0141,
        -0.0297, -0.0180, -0.0321, -0.0605,  0.0007,  0.0101,  0.0028,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0018, -0.0910, -0.0239, -0.0070, -0.1149, -0.0220,  0.0089,  0.0030,
         0.0059,  0.0109, -0.0014,  0.0107,  0.0042,  0.0094,  0.0154, -0.0630,
        -0.0065, -0.0033, -0.0009, -0.0032, -0.0612, -0.0083, -0.0052, -0.0144,
        -0.0799, -0.1178, -0.0286, -0.0338, -0.0123, -0.0384,  0.0013,  0.0053,
         0.0134, -0.0145, -0.0195,  0.0014, -0.0258, -0.0782,  0.0085,  0.0046,
        -0.0029,  0.0107,  0.0067,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0083, -0.1778, -0.0029, -0.0016,  0.0170,  0.0006, -0.0131, -0.0179,
        -0.0667, -0.0088, -0.0075, -0.0020, -0.0327, -0.0063, -0.0071, -0.0034,
        -0.0493, -0.0679, -0.0113, -0.0072, -0.0020,  0.0066, -0.0237, -0.0141,
         0.0147, -0.0113,  0.0034,  0.0022, -0.0026,  0.0017, -0.0092, -0.0267,
        -0.0568,  0.0175, -0.0076, -0.0030, -0.0214, -0.0210, -0.0481, -0.0052,
        -0.0344, -0.0502, -0.0031,  0.0029, -0.0034, -0.0162, -0.0013,  0.0024,
         0.0037, -0.0013,  0.0102, -0.0198, -0.0005,  0.0021, -0.0044, -0.0198,
         0.0161], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0370, -0.4552, -0.0013, -0.0405, -0.0479, -0.0330, -0.1012, -0.1155,
         0.0284, -0.0304, -0.0420,  0.0075,  0.0045,  0.0555,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0278, -0.1858, -0.0441, -0.0299, -0.0104, -0.0445,  0.0013, -0.0076,
        -0.0395,  0.0005, -0.0548, -0.1181, -0.0171, -0.0168, -0.0511, -0.0153,
        -0.0121, -0.0342,  0.0179, -0.0524, -0.0763, -0.0125, -0.0414,  0.0057,
         0.0149,  0.0159, -0.0051,  0.0003, -0.0402, -0.0066,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1451, -0.1610,  0.0603, -0.3773, -0.1187,  0.0471, -0.0785, -0.0120,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([-0.0202, -0.1778, -0.0088, -0.0122, -0.0426, -0.0991, -0.0016,  0.0321,
         0.0135, -0.0065,  0.0156,  0.0014, -0.0514, -0.0885, -0.0051, -0.0013,
         0.0003,  0.0027,  0.0020,  0.0029,  0.0072, -0.0005,  0.0207,  0.0177,
        -0.0634,  0.0060, -0.0238,  0.0009, -0.0838, -0.1280,  0.0027,  0.0294,
        -0.0303,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.5789e-02, -2.5283e-01, -9.4445e-02, -3.3441e-03, -1.1156e-02,
        -1.8016e-02, -1.1212e-03, -1.6138e-02, -1.3291e-03, -2.8214e-02,
        -1.0253e-02, -2.0737e-02, -1.6679e-02, -7.4937e-03, -3.4102e-04,
        -7.2822e-04,  2.9793e-03, -1.4996e-03, -1.7258e-02,  5.2304e-03,
        -9.5237e-04,  5.5478e-03,  3.7717e-03,  8.5714e-04,  8.7620e-03,
        -2.7026e-03, -1.7825e-04, -4.7493e-02,  4.4445e-03, -5.8500e-04,
         4.6272e-03, -6.2037e-03, -2.4749e-02, -3.1002e-02,  1.8001e-02,
        -4.1703e-03, -2.3446e-02,  5.7071e-04, -1.4360e-03,  3.1476e-03,
        -4.9155e-03,  5.4526e-03, -4.2479e-03, -3.3831e-02, -6.9632e-03,
         1.1680e-03, -2.0672e-03, -2.3327e-04, -9.4151e-03, -7.7274e-04,
         3.7494e-03, -2.1761e-02, -1.2942e-02, -2.5882e-02, -7.6897e-03,
         8.3999e-03,  1.8892e-04, -5.4675e-03,  6.1524e-04, -1.9069e-02,
        -3.2452e-04, -1.6529e-02, -3.2843e-02,  1.2983e-03, -8.8695e-04,
        -3.7783e-03, -7.4832e-03,  3.2258e-04,  7.9300e-03, -5.8920e-03,
        -9.6559e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0168, -0.0098, -0.0398,  0.0022, -0.0820, -0.0110, -0.1089, -0.0142,
         0.0004, -0.0064, -0.0079, -0.0058,  0.0017,  0.0081,  0.0185,  0.0145,
         0.0018,  0.0048, -0.0035, -0.0784, -0.0123,  0.0094,  0.0152, -0.0446,
         0.0055, -0.0529, -0.0232,  0.0044,  0.0007, -0.0073,  0.0103,  0.0077,
        -0.0745, -0.0859, -0.0250, -0.0092, -0.0990, -0.0471,  0.0031, -0.0021,
        -0.0058, -0.0186,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.1550e-03,  2.5186e-01,  1.5166e-02,  1.0405e-02,  2.9628e-03,
         3.4213e-03, -2.0637e-02,  1.0076e-02, -1.4694e-03,  5.9357e-02,
         1.2944e-02,  3.3811e-02,  9.2818e-02,  1.5939e-03,  2.6850e-03,
        -2.9632e-03,  1.7355e-02,  6.1732e-02,  1.1839e-01, -1.4509e-02,
         9.2737e-03,  2.3266e-02, -1.9805e-02,  3.7743e-03, -2.3476e-04,
         4.5924e-02,  3.8925e-03,  6.9441e-03,  7.4741e-02,  4.9428e-03,
         6.1696e-03, -3.4647e-04, -1.6891e-02, -5.6739e-03,  2.8964e-03,
         2.2427e-02, -1.5492e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0367, -0.4973, -0.0506, -0.1069,  0.0209,  0.0177, -0.0341, -0.0169,
        -0.0113,  0.0123, -0.0368,  0.0367, -0.0885, -0.0293, -0.0037,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0213, -0.0169,  0.0160,  0.0358, -0.1082, -0.2232,  0.0074, -0.1547,
        -0.2089, -0.0283, -0.0166,  0.0423, -0.0509,  0.0104,  0.0040,  0.0083,
        -0.0099, -0.0227, -0.0129,  0.0013,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0112, -0.0046, -0.0010, -0.0071, -0.1353, -0.0092, -0.0055, -0.0198,
         0.0068,  0.0007, -0.0439, -0.0821, -0.0088, -0.0054, -0.0509, -0.0233,
         0.0006, -0.0224,  0.0148, -0.0332, -0.0047, -0.0015,  0.0065, -0.0036,
         0.0062, -0.0546, -0.0677,  0.0021,  0.0044, -0.0537, -0.0819, -0.0160,
        -0.0009, -0.0029, -0.0480, -0.0601,  0.0064, -0.0446,  0.0033, -0.0140,
         0.0125,  0.0175, -0.0005,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.1741e-03, -8.3208e-03, -5.9071e-03,  3.6548e-03, -2.6599e-02,
        -1.0247e-02, -3.2913e-02, -7.6067e-02, -3.3283e-03, -5.2960e-05,
         1.7730e-02,  2.3090e-03,  3.9501e-03, -7.8072e-02, -9.2847e-02,
        -5.2230e-03,  4.5511e-03,  4.0422e-03,  4.5352e-03, -1.4678e-02,
        -4.8239e-02, -4.7821e-03,  3.6438e-02,  5.7317e-03, -3.1573e-03,
        -1.0887e-02, -4.4730e-03,  8.5081e-04, -1.0210e-01, -1.0396e-02,
         2.0962e-02, -9.5838e-02, -1.4033e-02, -9.4816e-02, -1.0478e-01,
        -2.0603e-02,  1.5100e-02, -8.6095e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0701, -0.2895, -0.1314, -0.1022, -0.0088, -0.0580, -0.0062,  0.0093,
        -0.0266, -0.0704, -0.0024,  0.0103, -0.0267, -0.0114,  0.0073, -0.0220,
        -0.0125, -0.0538,  0.0011,  0.0149, -0.0340,  0.0311,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0712, -0.0783,  0.0090,  0.0139,  0.0164, -0.0102,  0.0055,  0.0192,
        -0.0219, -0.0908, -0.0101,  0.0166, -0.0036,  0.0039, -0.0438, -0.0457,
        -0.1127, -0.0116, -0.0102, -0.0839, -0.0031,  0.0002, -0.0739, -0.0145,
        -0.0013, -0.0021, -0.0123,  0.0105,  0.0038, -0.0222, -0.0079, -0.0059,
         0.0008, -0.0051, -0.0272, -0.0759,  0.0183,  0.0092, -0.0256,  0.0018,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0088,  0.0080, -0.0097,  0.0004, -0.0195, -0.0389, -0.0257, -0.0219,
        -0.0658, -0.0455, -0.0116, -0.0136, -0.0098, -0.1173, -0.0791,  0.0165,
         0.0057,  0.0183, -0.0505, -0.0118, -0.0020, -0.0034,  0.0006, -0.0121,
        -0.0017, -0.1461,  0.0135, -0.0499, -0.0812, -0.0219, -0.0197, -0.0694,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.4479e-02, -3.4513e-01, -1.5209e-02,  3.5924e-03, -5.2619e-03,
         1.4107e-03, -1.3362e-02, -5.7879e-02, -9.4914e-02,  6.0716e-03,
        -2.4749e-03, -2.8103e-02,  8.9874e-04,  1.1959e-02,  1.0287e-02,
        -1.4635e-02, -4.4896e-02, -6.0831e-03, -1.8092e-04,  5.7379e-03,
        -7.1596e-03, -2.0951e-02, -1.0448e-02, -1.1246e-01, -1.7333e-02,
         2.4809e-03, -5.5953e-02, -1.2771e-02, -9.4113e-03, -2.2493e-02,
        -1.8045e-04,  7.5630e-03, -1.3711e-02,  6.4576e-03,  8.0615e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0938, -0.0277, -0.0982, -0.0730, -0.0272, -0.0424, -0.1869, -0.1034,
        -0.0141, -0.0918,  0.0049, -0.0366,  0.0225, -0.0554, -0.0106,  0.0053,
        -0.0190, -0.0076,  0.0182, -0.0473,  0.0142,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0187, -0.3540,  0.0107, -0.0725, -0.1609, -0.0758, -0.0391, -0.0287,
        -0.0101,  0.0311,  0.0089,  0.0166, -0.0043,  0.0255, -0.0229,  0.0182,
        -0.0373,  0.0452, -0.0193,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0238, -0.0521, -0.1110,  0.0261, -0.0684, -0.0065,  0.0112,  0.0010,
         0.0104, -0.0477, -0.1443, -0.0281, -0.0881, -0.2076, -0.0071, -0.0646,
        -0.0026, -0.0655,  0.0102,  0.0237,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0492, -0.3603, -0.0216, -0.0826,  0.0130, -0.0017, -0.0254, -0.0593,
         0.0012, -0.0145, -0.0096, -0.0153,  0.0042, -0.0573, -0.0071, -0.0492,
        -0.0691, -0.0098, -0.0340, -0.0607, -0.0013, -0.0013, -0.0523,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0321, -0.1636, -0.0999, -0.0703, -0.0046, -0.0375,  0.0035, -0.0481,
        -0.0535, -0.1018, -0.0277, -0.0594, -0.0191, -0.0745,  0.0108, -0.0053,
         0.0147, -0.0030, -0.0222, -0.0340, -0.0394, -0.0170,  0.0119, -0.0054,
         0.0410,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0549, -0.2517, -0.0133, -0.0219,  0.0088,  0.0022, -0.0197,  0.0011,
         0.0106, -0.0059, -0.0296,  0.0009,  0.0069,  0.0144,  0.0122, -0.0284,
         0.0017, -0.0209,  0.0092, -0.0026, -0.1395, -0.1054, -0.0610, -0.0460,
        -0.0121, -0.0190, -0.0091, -0.0133, -0.0004, -0.0147, -0.0203,  0.0009,
        -0.0106,  0.0037,  0.0270,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0450, -0.0024, -0.0171, -0.1135, -0.0149,  0.0213, -0.0672,  0.0073,
        -0.1510, -0.0179,  0.0710, -0.0045, -0.0239, -0.0749, -0.0675, -0.1212,
        -0.0329, -0.0815, -0.0304, -0.0113, -0.0031,  0.0200,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0722,  0.0052, -0.0106, -0.0049, -0.0245,  0.0296,  0.0064, -0.0258,
         0.0126,  0.1878,  0.0033, -0.0038, -0.0056,  0.0487, -0.0276,  0.2909,
         0.0194, -0.0078,  0.0749,  0.0595,  0.0790,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0004, -0.3997, -0.0148, -0.0156, -0.0125,  0.0007, -0.1601, -0.1428,
         0.0312, -0.0049, -0.0422,  0.0064, -0.0063, -0.0832, -0.0144, -0.0438,
         0.0074, -0.0014,  0.0121,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.7727e-02,  7.8879e-03, -2.0116e-02,  1.1789e-02,  4.4402e-03,
        -8.5769e-02, -7.0275e-04, -1.7450e-03, -2.6478e-03, -1.0706e-03,
         4.3931e-03, -9.7149e-03, -3.0086e-03, -3.4602e-02,  1.0293e-02,
         1.4151e-02, -2.3626e-02, -4.3915e-02,  6.7907e-03, -2.7124e-02,
         6.5272e-03, -1.9315e-02, -4.0621e-02, -7.1017e-03, -3.9185e-02,
         3.3716e-03, -2.1018e-02, -4.6149e-02, -8.1930e-05, -9.0671e-02,
         2.9803e-03, -4.1817e-03, -8.7206e-02, -9.2116e-02,  1.6283e-03,
        -9.2938e-03,  9.5877e-03,  9.6308e-03, -1.7875e-02, -4.2410e-02,
         5.7498e-03,  8.8104e-03,  4.8864e-03, -4.1582e-03,  1.1804e-03,
         1.2109e-02, -1.0641e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0334, -0.2214, -0.1468, -0.0142,  0.0059,  0.0021, -0.0101,  0.0044,
        -0.0443,  0.0143,  0.0029, -0.0090,  0.0050,  0.0010, -0.0058, -0.0021,
        -0.0707,  0.0038,  0.0006, -0.0198,  0.0067, -0.0239, -0.0712, -0.1541,
         0.0234,  0.0191, -0.0350, -0.0053, -0.0025,  0.0016,  0.0125,  0.0090,
        -0.0078, -0.0103,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0178, -0.3756, -0.0021,  0.0233, -0.0007, -0.0478, -0.0116, -0.0053,
        -0.0070, -0.0310, -0.0106,  0.0072, -0.0902, -0.0978,  0.0021, -0.0006,
        -0.0265, -0.0018, -0.0172, -0.0163,  0.0048,  0.0058, -0.0079, -0.0021,
        -0.0032,  0.0118,  0.0045, -0.0143, -0.0130, -0.0412, -0.0123,  0.0182,
        -0.0364,  0.0318,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0034,  0.4297, -0.0308,  0.0302,  0.0414,  0.0569, -0.0202,  0.0138,
         0.0127,  0.1054,  0.0228, -0.0125,  0.0552,  0.1074,  0.0129,  0.0225,
         0.0138,  0.0085,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0357,  0.0029,  0.0174, -0.0388,  0.0116,  0.0036, -0.0136, -0.0138,
         0.1041,  0.2601, -0.0352,  0.0074,  0.0111,  0.0014, -0.0091,  0.0405,
         0.0213, -0.0042, -0.0364,  0.1527,  0.0190, -0.0014,  0.0110, -0.0071,
         0.0045, -0.0288, -0.1075,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.3189e-02, -2.3192e-02, -1.7466e-03,  1.8950e-02, -9.4592e-03,
         3.7604e-02, -1.4093e-03, -1.0080e-02, -1.5016e-01, -3.9480e-02,
        -2.0347e-02, -8.8150e-03, -2.0830e-01, -5.4955e-02,  4.9465e-03,
        -1.3728e-03, -9.5765e-03,  5.5205e-03,  1.2702e-02, -1.8692e-01,
         1.3224e-02, -2.0269e-04, -5.5576e-02,  1.3788e-02, -1.8728e-02,
        -4.9758e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0529, -0.2021, -0.0054, -0.0870, -0.1450, -0.0343, -0.0181, -0.0035,
        -0.0627, -0.0366, -0.0742, -0.0364,  0.0021, -0.0004, -0.0134, -0.0621,
        -0.0300, -0.0449, -0.0049, -0.0017,  0.0823,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0312, -0.2939, -0.2169, -0.1776, -0.0085,  0.0287, -0.0506, -0.0031,
        -0.0345, -0.0033, -0.0077, -0.0400,  0.0446,  0.0519,  0.0073,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0879,  0.0138, -0.0210,  0.0209,  0.0259,  0.0278,  0.0042, -0.0177,
         0.0365, -0.0646, -0.0144, -0.0659, -0.4225, -0.0754,  0.0379,  0.0101,
        -0.0263, -0.0273,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0747, -0.2935, -0.1097, -0.1218, -0.0064, -0.0066,  0.0152, -0.0691,
        -0.0253, -0.0553, -0.0788, -0.0337,  0.0191, -0.0368, -0.0232,  0.0228,
        -0.0026,  0.0032,  0.0023,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0150, -0.1738, -0.1109, -0.0238, -0.0109, -0.0361, -0.0048,  0.0094,
        -0.0185, -0.0026,  0.0010,  0.0021, -0.0106,  0.0029,  0.0028,  0.0247,
        -0.0464,  0.0050,  0.0207,  0.0241, -0.0262,  0.0054, -0.0355,  0.0065,
        -0.0537, -0.0115, -0.0447,  0.0060,  0.0024, -0.0390, -0.0125, -0.0155,
        -0.0330, -0.0004, -0.0385, -0.0179, -0.0349, -0.0017, -0.0302, -0.0018,
        -0.0093,  0.0010, -0.0168, -0.0093], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0376, -0.2470, -0.0149, -0.0797, -0.0796, -0.0222, -0.0432, -0.0136,
        -0.0547, -0.0060, -0.0401, -0.0023,  0.0015,  0.0060,  0.0131, -0.0044,
         0.0301, -0.0914, -0.0112, -0.0304, -0.0459,  0.0011,  0.0051,  0.0064,
         0.0031,  0.0069, -0.0026, -0.0526,  0.0101,  0.0012, -0.0004, -0.0143,
        -0.0051, -0.0089, -0.0074,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.9222e-02, -4.5015e-01,  6.7940e-04, -5.7571e-02, -1.4437e-02,
         1.2521e-02, -7.4452e-02, -6.6856e-02, -1.2288e-02,  5.6848e-03,
        -1.5476e-03, -4.9860e-02, -7.6184e-02,  2.0022e-03,  2.6360e-03,
        -5.7134e-02, -7.3670e-03, -6.3304e-03, -7.1197e-03, -9.9285e-05,
         1.8024e-02, -1.2926e-02, -1.3258e-03,  3.3586e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0319, -0.2751, -0.0333, -0.0475, -0.0083, -0.0133,  0.0221, -0.0145,
        -0.1015, -0.0156, -0.0223, -0.0123,  0.0050, -0.0187, -0.0595, -0.0024,
         0.0240, -0.0115, -0.0633, -0.0076,  0.0011, -0.0455,  0.0114, -0.0294,
         0.0019, -0.0254, -0.0140, -0.0108, -0.0035, -0.0156, -0.0279,  0.0041,
        -0.0152,  0.0045,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-9.1795e-03,  3.3208e-02, -5.7242e-03, -1.5994e-02,  1.5306e-02,
         3.8140e-03, -1.3108e-01, -7.6326e-03,  2.9629e-02,  8.6837e-03,
         3.8196e-02, -6.3565e-03, -1.1397e-01, -2.2794e-02, -1.2322e-01,
        -1.8358e-01, -1.5468e-02, -6.4112e-03, -2.1044e-03, -1.5293e-02,
         2.0769e-03,  2.8901e-03, -2.3136e-02, -5.0750e-02, -2.5436e-02,
        -1.5311e-04, -7.2061e-02,  1.6876e-03, -1.1999e-02, -2.2166e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0631,  0.0346, -0.0297, -0.2362,  0.0609,  0.0121, -0.0115, -0.0256,
        -0.1153, -0.3242, -0.0443, -0.0057, -0.0367,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0025,  0.0354,  0.0678,  0.0197, -0.0021, -0.0733, -0.0761, -0.1330,
        -0.1714,  0.0089, -0.0075, -0.1345,  0.0083, -0.0706, -0.1508, -0.0087,
         0.0158,  0.0137,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0542, -0.2804, -0.4288, -0.0076, -0.1342,  0.0543, -0.0161,  0.0243,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1173e-03, -2.1637e-01,  1.4416e-03, -4.9285e-02, -3.4128e-02,
        -8.0110e-03, -2.1250e-02, -2.4582e-03,  3.7611e-03, -1.4397e-02,
         9.0288e-03,  5.0505e-03, -7.4667e-03,  7.8659e-03,  6.2565e-03,
        -8.3705e-03, -2.5609e-03,  1.2630e-03,  4.3917e-03, -2.8075e-03,
         2.9200e-03,  1.0454e-02, -3.6105e-02, -3.5703e-02,  3.2560e-03,
        -1.7682e-02,  2.9764e-03, -1.0069e-02, -7.7505e-02, -1.6215e-02,
         8.4745e-03, -5.8325e-02,  5.4812e-04, -8.0794e-02, -1.1253e-02,
        -3.4325e-02, -4.7472e-02, -1.4240e-02,  6.3721e-03,  1.4656e-04,
        -2.6042e-02, -1.6338e-02, -2.1077e-02, -7.3073e-04,  1.4421e-02,
         1.5220e-02,  2.4051e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1103, -0.0314,  0.1237, -0.4103, -0.0293, -0.0375, -0.0872, -0.0587,
         0.0517, -0.0471, -0.0129,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0377,  0.2926, -0.0009,  0.0455,  0.0183,  0.0031,  0.0696,  0.1653,
         0.0300,  0.1104,  0.0217, -0.0278,  0.0808, -0.0043, -0.0188,  0.0125,
        -0.0364,  0.0243,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0628, -0.2130, -0.2389, -0.0363,  0.0090,  0.0017, -0.1846, -0.0144,
        -0.0817, -0.0331, -0.0018, -0.0314, -0.0278,  0.0147,  0.0488,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0116, -0.3329,  0.0304, -0.0313,  0.0181, -0.0809, -0.1909, -0.0798,
         0.0195, -0.0819, -0.0721, -0.0173, -0.0173, -0.0116,  0.0044,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0421,  0.0104,  0.0269,  0.0295,  0.0105, -0.0009, -0.0159, -0.0037,
        -0.0863, -0.0514, -0.0228, -0.0883, -0.0966, -0.0249, -0.0044,  0.0054,
        -0.0412,  0.0209, -0.0088,  0.0057, -0.1042, -0.1902,  0.0154,  0.0718,
        -0.0144, -0.0072,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0121,  0.0273,  0.0289,  0.0967,  0.0337, -0.0385, -0.0121, -0.2018,
        -0.3952, -0.0030,  0.0316, -0.0313,  0.0482,  0.0374,  0.0022,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0606, -0.2389, -0.0021, -0.0611, -0.0876, -0.0187, -0.0864, -0.0023,
        -0.0080, -0.1714,  0.0081,  0.0213, -0.0356, -0.0109,  0.0079,  0.0039,
        -0.0102, -0.0042, -0.0470,  0.0043, -0.0247,  0.0003,  0.0292,  0.0263,
         0.0191, -0.0098,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0593, -0.0010,  0.0027,  0.0076, -0.0069, -0.0334, -0.0028,  0.0201,
        -0.0603, -0.0756, -0.1123,  0.0050, -0.0135,  0.0274, -0.1464, -0.0715,
         0.0571, -0.0489, -0.0015, -0.0854, -0.0411, -0.0451,  0.0086, -0.0499,
        -0.0123,  0.0044,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.0421, -0.3367,  0.0077,  0.0058, -0.0098, -0.0010, -0.0308, -0.0467,
        -0.0018, -0.0309, -0.0122, -0.0242, -0.0623,  0.0087, -0.0193, -0.0264,
        -0.0042,  0.0024, -0.0187,  0.0044, -0.0092, -0.0104, -0.0097, -0.0153,
        -0.0533,  0.0182, -0.0245, -0.0113, -0.0150,  0.0026, -0.0146, -0.0320,
        -0.0036, -0.0094, -0.0096,  0.0652,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0178, -0.1265, -0.0518, -0.0083, -0.0604,  0.0012, -0.0512, -0.0826,
         0.0040, -0.0040, -0.0355, -0.0092, -0.0089,  0.0088, -0.0011, -0.0221,
         0.0009, -0.0063,  0.0280, -0.0098, -0.0123, -0.0570, -0.0010, -0.0087,
        -0.0335, -0.0792, -0.0058, -0.0594, -0.0517, -0.0048, -0.0051, -0.0381,
        -0.0595,  0.0106,  0.0038,  0.0311,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0433,  0.0596, -0.0225, -0.0207,  0.0346, -0.2404, -0.0511, -0.0075,
         0.0439,  0.0006, -0.0012, -0.1714, -0.1511,  0.0163,  0.0104, -0.0173,
         0.0261, -0.0307, -0.0157,  0.0357,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0491, -0.0585, -0.0239, -0.0725, -0.0181, -0.0156, -0.0116, -0.0291,
        -0.0183, -0.0167,  0.0036, -0.0014, -0.0063, -0.0336, -0.0821, -0.0538,
        -0.0390, -0.0609, -0.0096,  0.0044, -0.0241, -0.0099, -0.0092,  0.0008,
        -0.0130,  0.0011, -0.0037, -0.0016, -0.0069, -0.0354, -0.0818, -0.0193,
        -0.0090, -0.0702, -0.0376,  0.0086, -0.0567, -0.0030,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.0473e-02, -2.0405e-01, -1.8494e-01, -7.3327e-02, -1.9582e-02,
         1.4249e-03, -6.9985e-03, -6.4210e-03, -1.9774e-02,  1.3876e-02,
        -3.9815e-02, -6.4678e-02,  1.5530e-03, -1.1332e-02, -3.3286e-03,
        -4.4787e-02, -7.3660e-05, -3.9211e-02, -9.9121e-03,  8.1550e-04,
         1.5126e-02,  3.9010e-03,  1.6371e-02,  1.0685e-02, -1.0433e-02,
         1.9794e-04, -1.4508e-02, -4.2669e-02, -4.8037e-02,  4.9990e-03,
        -1.3244e-02, -9.2525e-03,  4.6921e-03,  1.8558e-03, -1.0230e-02,
        -5.2494e-05, -7.9519e-03, -1.1131e-03, -2.9485e-03,  2.0847e-03,
        -3.2765e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.6052e-05, -4.1360e-01, -3.0581e-02,  6.2125e-02, -3.8144e-03,
        -5.5718e-02,  3.5939e-02, -4.1794e-04, -2.8276e-02, -3.5585e-03,
         6.9033e-03, -5.2083e-02, -8.2416e-02, -1.3987e-02,  2.2085e-02,
        -4.1355e-02, -2.8861e-03, -2.2859e-02, -3.4270e-03,  1.0882e-02,
         5.9146e-02, -4.7882e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0045,  0.0052,  0.0012, -0.0110,  0.0072, -0.0171, -0.0008,  0.0070,
        -0.0123,  0.0017, -0.0029, -0.0264, -0.0012, -0.0500, -0.0070, -0.0892,
         0.0105, -0.0682, -0.0191, -0.0993, -0.0920, -0.0052,  0.0362, -0.0351,
         0.0030,  0.0151,  0.0009, -0.0196, -0.1212, -0.0082, -0.0530, -0.0756,
        -0.0131,  0.0620, -0.0178,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1365, -0.0468, -0.1151, -0.0161, -0.0849,  0.0095, -0.0091,  0.0121,
        -0.0131, -0.0067, -0.0439, -0.0033, -0.0025, -0.0158, -0.0045,  0.0034,
         0.0345, -0.0080, -0.0729, -0.0220, -0.0613, -0.0081,  0.0112, -0.0325,
        -0.0016, -0.0321, -0.0516,  0.0643,  0.0768,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0615, -0.3131, -0.0249, -0.0455,  0.0005, -0.0511,  0.0084, -0.0286,
        -0.0344, -0.0053, -0.0747, -0.0570,  0.0068, -0.0042,  0.0061, -0.0892,
        -0.0322, -0.0243, -0.0316,  0.0262, -0.0135, -0.0357, -0.0051,  0.0174,
        -0.0031,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0388,  0.0091, -0.0165,  0.0312, -0.2047,  0.0144, -0.0190, -0.1186,
        -0.1600, -0.0400,  0.0016, -0.0048,  0.0023,  0.0085, -0.0903, -0.1036,
         0.0349, -0.0325, -0.0074, -0.0007, -0.0103,  0.0195,  0.0315,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0586,  0.0035,  0.0025,  0.0059,  0.0184,  0.0076, -0.0092, -0.0013,
         0.0059, -0.0030, -0.0017, -0.0264, -0.0516, -0.1572, -0.0262, -0.0373,
        -0.0009, -0.0501, -0.0127, -0.0019,  0.0054, -0.0148,  0.0078,  0.0242,
         0.0051,  0.0149, -0.1270, -0.0256, -0.0654,  0.0114, -0.0397, -0.0079,
        -0.0070,  0.0032, -0.0239,  0.0616, -0.0146,  0.0577,  0.0009,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0833,  0.0121, -0.0145, -0.1740, -0.2545, -0.0122, -0.0307, -0.1540,
        -0.1621,  0.0476,  0.0268, -0.0282,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-1.0619e-02, -1.1145e-01, -3.3899e-01,  7.3313e-04, -6.4269e-03,
        -4.7949e-02, -3.9287e-02,  2.1235e-02, -1.1308e-02, -1.4511e-02,
         5.7273e-03, -8.9688e-03, -8.9591e-03, -3.1590e-04, -7.0929e-02,
        -1.8455e-03, -1.5551e-02, -2.6813e-02,  8.3468e-03,  4.0139e-03,
        -5.6169e-03, -8.0078e-03,  1.8948e-02, -4.3035e-03,  1.9779e-03,
         2.1863e-03, -1.7254e-02, -1.2006e-02, -4.5590e-03,  3.8970e-03,
         2.2345e-03,  1.1540e-02, -2.5001e-02, -6.4102e-03,  3.0788e-03,
         5.4819e-03, -7.0727e-02, -6.3495e-03,  3.4406e-04,  2.0602e-03,
         3.4420e-03,  1.4967e-02,  3.3892e-03,  8.5906e-03, -3.6462e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1151, -0.5475,  0.0310, -0.0624, -0.0088,  0.0277, -0.0950, -0.0058,
        -0.0724,  0.0032, -0.0210,  0.0100,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1425, -0.6387,  0.0536, -0.0213, -0.0489, -0.0285,  0.0652,  0.0011,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.8045e-02, -2.4209e-01, -2.6536e-02, -5.8129e-03, -5.1815e-02,
        -4.7394e-03,  1.0546e-02, -9.1875e-03, -4.6132e-02, -5.6125e-02,
        -2.0801e-02, -1.4630e-02, -4.2782e-03,  3.0790e-04, -1.0373e-03,
        -7.1797e-04,  1.0234e-02, -5.8041e-03,  1.6334e-03, -7.6350e-03,
        -1.1903e-03,  1.8038e-03, -1.7002e-03, -3.8987e-03, -9.3216e-03,
         1.8095e-03, -1.4569e-03, -2.4835e-03, -2.1139e-02, -4.8969e-03,
         6.6577e-05,  1.1064e-03,  1.3992e-03,  4.4606e-03, -8.5027e-03,
        -1.7155e-04,  8.0507e-03,  1.9268e-03,  2.0892e-03,  1.2192e-03,
        -3.0274e-03,  1.0881e-03, -2.6743e-03, -2.6679e-03, -3.1984e-03,
         2.0880e-02, -4.1240e-02, -1.1038e-03, -4.5841e-02, -1.2274e-01,
        -1.9063e-02, -7.1856e-03, -3.1628e-02, -1.0295e-02, -1.6047e-02,
        -2.4679e-03, -1.1449e-02,  1.5298e-03,  1.7571e-03,  3.0449e-03,
        -2.4271e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0064, -0.0041, -0.0056, -0.0346, -0.1114, -0.0165, -0.0436,  0.0103,
         0.0133, -0.0034, -0.0088, -0.0756,  0.0155, -0.0575, -0.1053, -0.0098,
        -0.0077, -0.0042, -0.0461, -0.0346, -0.0629, -0.0834, -0.0267,  0.0108,
        -0.0426, -0.0278, -0.0146, -0.0158, -0.0316, -0.0215,  0.0072, -0.0038,
        -0.0367,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0500, -0.0461, -0.2018, -0.0298,  0.0273, -0.0082, -0.0823,  0.0007,
         0.0197,  0.0054, -0.0260, -0.0023, -0.0319, -0.0112, -0.0164,  0.0065,
         0.0174, -0.0666, -0.2087,  0.0220,  0.0268, -0.0133, -0.0018, -0.0778,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0541, -0.1056, -0.1579, -0.0447, -0.0005, -0.0711, -0.0168, -0.0345,
        -0.0187, -0.0831, -0.0657, -0.0498, -0.1463, -0.0050,  0.0292, -0.0107,
        -0.0012,  0.0137,  0.0151,  0.0059, -0.0421, -0.0044, -0.0127,  0.0110,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0922, -0.0048, -0.0885, -0.1197, -0.0036,  0.0038, -0.0022, -0.0350,
         0.0078,  0.0112, -0.0032,  0.0012,  0.0058,  0.0001, -0.0412, -0.0751,
        -0.0680,  0.0023,  0.0075, -0.0357, -0.0660,  0.0007, -0.0309, -0.0704,
        -0.0074,  0.0143, -0.0725,  0.0100, -0.0049, -0.0244, -0.0221, -0.0153,
         0.0330,  0.0191,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0189, -0.0020,  0.0087,  0.0064, -0.1158, -0.0235, -0.0128, -0.0037,
        -0.0123,  0.0034, -0.0091, -0.0176, -0.0059, -0.1367,  0.0006,  0.0299,
        -0.2550, -0.0187, -0.0718, -0.0149, -0.0656, -0.0919,  0.0214, -0.0005,
         0.0529,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0392,  0.0064, -0.0014, -0.0024, -0.0104, -0.0042, -0.1710, -0.0987,
        -0.1420, -0.3683,  0.0101, -0.0200, -0.0252, -0.0101,  0.0395, -0.0511,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0070, -0.2353,  0.0140, -0.0933, -0.0503, -0.0934, -0.1681, -0.0134,
         0.0451,  0.0045, -0.0054,  0.0143, -0.0501,  0.0044,  0.0197, -0.0126,
        -0.0345, -0.0797, -0.0307, -0.0051,  0.0010, -0.0092, -0.0090,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0078,  0.0558, -0.0338,  0.1492,  0.5421, -0.0062, -0.0647,  0.0159,
         0.0177, -0.0151, -0.0225, -0.0281,  0.0162,  0.0248,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0662, -0.1210, -0.0527, -0.1370, -0.0125,  0.0031, -0.0418, -0.0169,
        -0.0760, -0.0164, -0.0807, -0.0724,  0.0117,  0.0003, -0.0035,  0.0309,
         0.0027, -0.1083, -0.0020, -0.0623, -0.0029,  0.0124, -0.0041,  0.0623,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0793,  0.0064, -0.1932,  0.0094, -0.3121, -0.0076, -0.3021,  0.0088,
        -0.0119,  0.0096, -0.0597,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0072,  0.0328,  0.0069,  0.0074,  0.0362, -0.1073, -0.1331, -0.0086,
        -0.1795, -0.0043, -0.0738, -0.0057, -0.0423, -0.0165, -0.0670, -0.0193,
        -0.0285, -0.1323, -0.0128,  0.0733,  0.0050,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0022, -0.0062, -0.0091, -0.0338, -0.0118,  0.0076, -0.0071, -0.3418,
        -0.1457, -0.2513, -0.0210,  0.0457, -0.0199, -0.0969,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1513,  0.0204,  0.0099,  0.0080,  0.0099, -0.0219, -0.0143, -0.0829,
        -0.0137, -0.0115,  0.0200,  0.0928,  0.3975, -0.0015,  0.0139,  0.0351,
        -0.0707,  0.0032, -0.0044,  0.0036, -0.0134,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0313, -0.3583,  0.0574, -0.1458,  0.0197, -0.1550,  0.0023, -0.1480,
         0.0015, -0.0417, -0.0303,  0.0022,  0.0066,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.9906e-03, -1.0750e-01,  3.8828e-03, -4.7068e-03, -1.1906e-04,
        -1.0321e-02, -4.0199e-02, -7.6719e-02, -1.8478e-02, -1.5680e-03,
         5.1206e-03,  9.8108e-03,  6.1315e-03, -1.3545e-03,  2.8647e-04,
        -5.2532e-03,  2.2607e-03, -2.5870e-06,  2.5735e-03,  1.3042e-02,
        -7.2020e-03,  1.5196e-02,  1.0409e-03, -1.1790e-02, -1.6916e-01,
         2.9627e-02, -9.5233e-02, -7.7171e-02, -1.2798e-01, -1.7046e-02,
        -1.2978e-02,  6.0903e-03, -7.8109e-03, -6.0734e-04, -8.7998e-03,
         1.0962e-03, -3.2497e-02,  3.2445e-03,  4.5346e-03, -1.1038e-02,
        -2.0946e-02,  3.0080e-04, -6.1222e-05,  5.3703e-03,  1.4861e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0512, -0.0577, -0.0698, -0.0211, -0.0041,  0.0224, -0.0193, -0.0245,
        -0.0189,  0.0078, -0.0803, -0.0062, -0.0089,  0.0035, -0.0579, -0.0191,
        -0.1209, -0.0035,  0.0001, -0.0015, -0.0082, -0.0699, -0.0159, -0.1004,
        -0.0077,  0.0020, -0.0149, -0.0307, -0.0830, -0.0201,  0.0145,  0.0040,
        -0.0218,  0.0083,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0116, -0.0240,  0.0049,  0.0728,  0.0217,  0.0213, -0.0028,  0.0184,
        -0.0147,  0.0106,  0.0111,  0.0044,  0.0022,  0.0050, -0.0173, -0.0138,
         0.0043, -0.0337,  0.0404,  0.2642,  0.2813, -0.0168, -0.0207,  0.0316,
        -0.0161,  0.0344,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0028, -0.3613,  0.0194, -0.0648,  0.0185,  0.0068,  0.0180, -0.0835,
        -0.0168,  0.0045,  0.0106,  0.0051,  0.0056,  0.0061, -0.0251, -0.0813,
        -0.1443,  0.0066, -0.0375,  0.0111, -0.0235,  0.0287,  0.0182,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.2930e-02, -1.6501e-01, -7.2120e-03, -2.9233e-02,  8.0810e-03,
        -1.0345e-02, -1.2690e-02, -2.2004e-02, -1.0019e-01, -3.0926e-02,
        -7.6461e-04, -2.0389e-02,  8.6776e-03, -1.1911e-02, -1.2997e-01,
         5.6250e-03, -7.2898e-05, -3.3764e-02, -6.1323e-02, -1.6903e-02,
        -4.6493e-03, -3.9465e-03,  2.1126e-03, -6.2970e-03,  5.0955e-03,
         6.7927e-04, -2.8347e-02, -9.0823e-02, -1.2458e-03, -1.4195e-02,
         4.6944e-03, -1.3861e-03,  1.2266e-02,  6.2248e-03, -7.6011e-03,
         7.2652e-03, -2.4617e-02,  1.2399e-02, -1.5233e-02,  1.2903e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0138,  0.0371,  0.0003, -0.0278, -0.0940, -0.0075,  0.0096, -0.0855,
         0.0046, -0.0369, -0.0034,  0.0081, -0.0440,  0.0029, -0.0452, -0.0815,
        -0.0613,  0.0065,  0.0241, -0.0412, -0.0058, -0.0536, -0.0823, -0.0023,
        -0.0058,  0.0089, -0.0029, -0.0223, -0.0047, -0.0168, -0.0896,  0.0063,
         0.0214,  0.0420,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-1.9941e-02, -3.4826e-02, -3.8044e-02,  5.1552e-03, -8.1835e-02,
        -3.7518e-03,  6.8288e-03, -1.0324e-02, -3.4385e-03,  7.5786e-03,
        -5.5514e-03, -1.2847e-02,  2.5667e-02, -2.2245e-02, -1.3425e-01,
         8.1890e-03, -1.1380e-02, -1.2220e-03, -5.1444e-02,  2.5933e-03,
         1.7109e-02,  7.1063e-03, -6.8811e-03, -9.2345e-04, -4.5304e-02,
        -9.9264e-02,  5.9500e-03, -2.1720e-02, -5.6769e-03,  8.6324e-03,
         2.3324e-02,  6.1927e-03,  1.0191e-02,  7.7283e-03, -4.1403e-02,
        -4.2384e-03, -1.6682e-02, -2.4892e-02, -1.0212e-02, -2.3237e-02,
        -1.2374e-02, -1.9891e-02, -2.3704e-02, -1.0264e-02,  4.2521e-03,
         2.2583e-03, -4.1799e-03,  1.9239e-06, -9.1433e-03, -5.0304e-03,
        -3.0513e-02, -4.6138e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.0052e-03,  4.8306e-03, -9.7906e-03, -2.8432e-02,  1.4348e-01,
        -1.6968e-02,  1.1357e-04, -3.3412e-02, -1.2523e-02,  2.8036e-02,
        -2.8016e-03,  6.4302e-03,  1.6265e-02,  1.7738e-01,  1.3715e-02,
         8.6009e-02,  1.2659e-01,  4.9065e-02,  3.7999e-02, -2.5422e-02,
         1.4748e-02,  1.1346e-02,  1.0918e-02,  9.0118e-02,  1.0242e-02,
         9.5356e-03, -2.2301e-03,  9.0359e-03,  2.6420e-03, -1.8462e-03,
        -1.2068e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.3292e-02, -1.8302e-01,  8.5102e-03, -3.5958e-02,  1.2552e-04,
         1.8191e-02, -2.0667e-02, -7.3333e-04,  1.1251e-02,  4.9902e-02,
        -1.4401e-03, -1.2310e-01,  7.7770e-03, -1.0377e-02, -8.4881e-02,
        -8.0642e-02,  1.6689e-02, -9.0112e-02, -7.7449e-02, -1.2450e-01,
        -6.2859e-03,  6.2305e-03, -1.0386e-02,  7.1511e-03, -1.1329e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1115, -0.0344, -0.0092,  0.0073,  0.0762,  0.2326,  0.3594,  0.0539,
         0.0092, -0.0222, -0.0517,  0.0325,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.5011e-03, -3.0729e-01, -1.6237e-02, -6.1206e-02, -1.5840e-01,
        -3.0970e-02,  1.8500e-02, -3.2881e-03, -1.5387e-03, -1.0869e-03,
        -9.4909e-03,  3.5925e-03, -3.1984e-02, -1.0468e-01,  2.0453e-02,
        -4.2749e-02, -2.8295e-03, -4.3053e-03, -4.6090e-03, -2.5934e-03,
        -1.2519e-02,  1.0607e-04, -2.6160e-02,  1.0922e-02,  1.6758e-02,
        -7.0430e-02, -5.5497e-03,  1.5101e-02,  1.4147e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0757, -0.4267, -0.0083, -0.0289, -0.0087, -0.1341, -0.2311,  0.0169,
        -0.0055,  0.0249, -0.0130, -0.0261,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0144, -0.4517,  0.0257, -0.0568,  0.0411, -0.1272, -0.1357,  0.0078,
        -0.0901, -0.0146, -0.0290, -0.0060,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0579,  0.0849, -0.1400, -0.0282, -0.1781,  0.0456,  0.0020, -0.0362,
        -0.1502, -0.0144,  0.0094, -0.1110, -0.0044, -0.0086, -0.0163, -0.0715,
        -0.0110, -0.0012,  0.0100, -0.0187,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0391, -0.2156, -0.0063, -0.0746, -0.0747,  0.0027, -0.0303, -0.0018,
        -0.0202,  0.0039, -0.0190, -0.0359, -0.0402, -0.0123,  0.0096,  0.0052,
        -0.0251, -0.0009, -0.0038,  0.0050, -0.0198,  0.0031,  0.0013,  0.0070,
         0.0027,  0.0228, -0.0022, -0.0456, -0.0373,  0.0202, -0.0070,  0.0072,
        -0.0433, -0.1105, -0.0363, -0.0046,  0.0030,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1077,  0.5613,  0.0351,  0.1603, -0.0043,  0.0037, -0.0154,  0.1033,
        -0.0090,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0666, -0.0035,  0.0243, -0.1177, -0.0023, -0.0354, -0.0598, -0.0950,
         0.0221,  0.0008, -0.0012, -0.0277, -0.0761, -0.1467,  0.0262,  0.0114,
        -0.0084,  0.0136, -0.0693, -0.1175,  0.0015, -0.0069, -0.0178,  0.0057,
         0.0053,  0.0301, -0.0072,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0067, -0.2505,  0.0113, -0.0455, -0.0022, -0.0114, -0.0021,  0.0005,
        -0.0046,  0.0044,  0.0490, -0.1133, -0.0156, -0.0042, -0.0014,  0.0039,
        -0.0270,  0.0050, -0.0167,  0.0086, -0.0171, -0.0544, -0.0736, -0.0124,
         0.0088,  0.0021, -0.0041, -0.0213, -0.0414, -0.0156,  0.0168, -0.0029,
         0.1149,  0.0310,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 1.2930e-03, -3.9734e-03,  5.3779e-03, -3.7341e-03, -4.8966e-02,
        -2.0620e-02, -2.3615e-04, -1.7887e-03,  1.2429e-02,  3.9338e-02,
         1.0834e-02, -9.4738e-02,  4.5935e-03, -1.2280e-02,  5.4820e-03,
        -4.0060e-02, -9.3242e-02, -2.9849e-02, -3.2469e-02, -2.0435e-03,
         1.3115e-02,  1.0545e-02, -4.5776e-02, -8.4163e-03,  9.3626e-03,
        -8.9610e-05, -5.4129e-03, -6.8256e-03,  2.4811e-03, -9.0041e-03,
        -3.3069e-03, -7.7920e-02, -1.0376e-01, -1.3515e-02, -1.9840e-02,
        -2.0101e-02, -6.9050e-02,  3.3553e-03,  5.6745e-03, -6.7535e-03,
        -2.0240e-02, -4.5939e-02,  4.7001e-03,  5.4054e-03,  2.6065e-02,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.6962e-02, -1.8180e-06,  1.7466e-02,  2.6479e-04, -2.8633e-02,
        -1.8630e-01, -2.4328e-01, -1.8208e-02, -2.1207e-02, -1.8031e-01,
        -1.2445e-02, -7.7810e-03, -1.3650e-01,  1.6898e-02,  2.5237e-02,
        -3.8500e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.4061e-02,  6.2185e-04, -8.6831e-03, -2.3019e-02, -8.0497e-04,
        -2.3287e-02, -6.5703e-03, -3.9515e-02, -1.5482e-01, -2.3156e-01,
        -7.4380e-02, -1.1857e-01,  2.1985e-03, -9.6040e-02, -9.8654e-02,
         3.5343e-03, -9.1062e-03,  4.1097e-03,  5.4178e-03,  8.1446e-03,
        -2.0067e-05, -9.2413e-03,  8.4186e-03,  1.3860e-03, -6.9069e-03,
        -7.6861e-03,  4.3241e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0015, -0.0121, -0.0813, -0.0054, -0.0523,  0.0027, -0.0266, -0.0100,
        -0.0281, -0.0132,  0.0079, -0.0062, -0.0251,  0.0035, -0.0278, -0.0074,
        -0.0147,  0.0066, -0.0127, -0.0432,  0.0098, -0.0232, -0.0276, -0.1004,
        -0.0076,  0.0051, -0.0186, -0.0613,  0.0001, -0.0183,  0.0071, -0.0289,
        -0.0491, -0.0090, -0.0168, -0.0492, -0.0159, -0.0427, -0.0033,  0.0073,
        -0.0232, -0.0136, -0.0028, -0.0024,  0.0589,  0.0098], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0570,  0.0147, -0.0303,  0.0286, -0.0086, -0.0041, -0.0099, -0.1041,
        -0.0080, -0.0223, -0.0320, -0.0169,  0.0100,  0.0042, -0.0331,  0.0004,
         0.0035,  0.0193, -0.0801, -0.2364, -0.0315, -0.0206, -0.0022, -0.0206,
         0.0047,  0.0074, -0.0693,  0.0149, -0.0581,  0.0473,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1539,  0.0072,  0.0014, -0.0464, -0.0148, -0.0749,  0.0655,  0.4347,
         0.0358,  0.0506,  0.1148,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0532,  0.3727, -0.0030, -0.0268,  0.0988,  0.0169,  0.0568,  0.0740,
        -0.0079,  0.0170,  0.0085,  0.0078,  0.0116,  0.0436,  0.0213,  0.0599,
         0.0133,  0.0381, -0.0123,  0.0563,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0745, -0.3826, -0.1806, -0.0662,  0.0169, -0.0528, -0.1152,  0.0075,
         0.0536,  0.0073,  0.0428,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1179, -0.2530, -0.0718, -0.1152, -0.0323, -0.0250, -0.0164, -0.0867,
        -0.0835, -0.0171, -0.0005, -0.0064,  0.0043, -0.0278, -0.0126, -0.0453,
         0.0131, -0.0233, -0.0150, -0.0314,  0.0013,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.9183e-03, -3.3009e-01, -1.9364e-01, -1.1540e-02,  2.2429e-02,
        -2.1041e-01, -3.3358e-02,  6.7579e-02, -1.9193e-02, -3.1044e-03,
        -1.1431e-02, -5.9306e-02,  1.4599e-04,  7.0232e-03,  6.5908e-03,
        -5.7352e-03,  7.0682e-03,  6.4339e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.6069e-02, -1.0935e-01, -1.8105e-01,  2.2509e-02, -2.1790e-02,
         3.5457e-03,  7.6920e-03,  2.7164e-03, -7.8720e-04, -3.1755e-03,
         1.0662e-02,  1.1850e-03,  4.5914e-03, -4.2435e-02, -7.5610e-02,
        -6.8218e-03, -5.4862e-02, -7.7985e-02, -2.3946e-02,  6.2584e-03,
         3.8903e-03, -4.1898e-02, -9.1403e-02, -2.1139e-03, -2.3587e-02,
        -6.0331e-04, -2.1119e-02, -1.5671e-02, -2.4503e-02, -4.7667e-02,
        -1.0687e-03, -1.7841e-04,  1.9435e-03,  8.4942e-03, -1.4087e-02,
         3.5821e-03,  5.5885e-03,  9.8489e-04,  1.8571e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1272e-02, -4.7733e-01, -5.1583e-05, -6.0945e-02, -9.8672e-03,
         8.5484e-03,  1.2574e-02,  1.3090e-02, -7.4314e-04, -1.6979e-02,
        -6.1899e-03, -8.2795e-02, -1.2389e-02,  2.8247e-03, -4.8800e-03,
         2.0245e-03, -5.8669e-02, -8.3823e-03,  1.7346e-02, -1.2508e-02,
        -9.6802e-03,  9.2168e-03,  3.3861e-03,  1.0793e-02,  9.7974e-03,
         2.5879e-03, -1.4028e-02, -2.4980e-03, -3.0979e-02, -4.3818e-02,
        -1.0256e-02, -1.4653e-03,  5.8022e-03, -2.6287e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.1288, -0.0142, -0.2807, -0.2892,  0.0030, -0.1276,  0.0611, -0.0682,
         0.0273,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.2534e-02, -1.6776e-01,  3.2490e-03, -7.6106e-02, -3.8411e-02,
        -1.2097e-01, -5.2764e-03, -1.5734e-02, -3.1339e-02, -5.6902e-03,
        -3.9003e-02, -1.6035e-02, -4.3957e-03, -1.3364e-02, -5.3013e-03,
        -6.4641e-02, -1.9818e-02, -9.0295e-02, -6.4456e-03, -9.4400e-04,
        -8.9375e-03,  1.0947e-03,  9.6414e-03, -2.4768e-03, -3.1652e-02,
         8.0484e-03, -5.2604e-03, -4.9704e-03,  7.7126e-03,  5.3736e-05,
        -5.5930e-03, -3.3258e-02,  3.3574e-03, -4.0511e-02,  1.1634e-02,
         3.1548e-03,  1.3822e-03, -3.2219e-03,  1.6626e-02,  4.1048e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0029, -0.0960, -0.2142, -0.0304, -0.0538, -0.0288, -0.0026, -0.0102,
        -0.0546, -0.0484,  0.0008,  0.0034,  0.0016, -0.0012, -0.0108,  0.0098,
         0.0016,  0.0032, -0.0306,  0.0121, -0.0372, -0.0092,  0.0062,  0.0051,
         0.0017, -0.0126, -0.0253, -0.0031, -0.0428, -0.0780, -0.0053, -0.0063,
        -0.0037,  0.0009,  0.0044,  0.0053, -0.0171, -0.0397,  0.0055,  0.0003,
         0.0143,  0.0085,  0.0009,  0.0012, -0.0008,  0.0003, -0.0075, -0.0054,
         0.0024, -0.0022,  0.0253, -0.0048,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0390, -0.3310, -0.0435, -0.0477, -0.0077, -0.0090, -0.0644,  0.0029,
        -0.0392, -0.0064,  0.0157, -0.0276, -0.0614, -0.0252, -0.0310, -0.0730,
        -0.0652, -0.0094, -0.0610, -0.0258,  0.0077,  0.0062,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0621, -0.5337, -0.0283, -0.0212, -0.0147,  0.0218, -0.0061, -0.0426,
        -0.0015,  0.0049, -0.0144, -0.0319,  0.0114, -0.0074,  0.0170, -0.0948,
        -0.0254,  0.0028,  0.0135,  0.0125, -0.0321,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0080, -0.1790, -0.0135, -0.0086, -0.0044,  0.0021,  0.0012,  0.0127,
        -0.0258,  0.0078,  0.0021, -0.0049,  0.0049, -0.0015,  0.0033,  0.0008,
         0.0030,  0.0008, -0.0243,  0.0033,  0.0140, -0.0071, -0.0038,  0.0119,
         0.0036,  0.0065, -0.0122, -0.0389,  0.0004, -0.0529, -0.0679, -0.0463,
        -0.0011, -0.0496, -0.0382, -0.0081,  0.0009,  0.0056,  0.0080,  0.0039,
        -0.0883, -0.0079,  0.0051,  0.0048, -0.0346,  0.0079, -0.0412, -0.0220,
         0.0063,  0.0113, -0.0095, -0.0048, -0.0220,  0.0252, -0.0163],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0064,  0.0085,  0.0335, -0.0022,  0.0064, -0.0282, -0.0690,  0.0015,
         0.0092,  0.0065, -0.0170,  0.0002, -0.0035, -0.0245, -0.0412, -0.0083,
         0.0068, -0.0240, -0.0027, -0.0722, -0.0208, -0.0166, -0.0721,  0.0016,
        -0.0052,  0.0085,  0.0092, -0.1028,  0.0067,  0.0031, -0.0052,  0.0021,
        -0.0288, -0.0977, -0.0038, -0.0295,  0.0056, -0.0364, -0.0607, -0.0396,
        -0.0114, -0.0032, -0.0072,  0.0141,  0.0362,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0255, -0.2555, -0.0884, -0.1021,  0.0137, -0.0164,  0.0027, -0.0292,
        -0.0339, -0.0077,  0.0004, -0.0430, -0.0024, -0.0682, -0.1166,  0.0025,
        -0.0232, -0.0571, -0.0190, -0.0187, -0.0322,  0.0253, -0.0136, -0.0026,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0026,  0.0267, -0.0010, -0.0097, -0.0250,  0.0006, -0.0604, -0.0703,
         0.0104,  0.0010, -0.0263,  0.0008, -0.0043, -0.0518,  0.0054,  0.0228,
        -0.0452, -0.0308, -0.0376, -0.0854, -0.0167,  0.0059, -0.0419, -0.0845,
        -0.0038, -0.0469, -0.0761, -0.0291, -0.0388, -0.0856, -0.0009, -0.0093,
        -0.0425,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0195, -0.4247, -0.0097, -0.1374, -0.0214, -0.0170, -0.0025, -0.0221,
        -0.1111, -0.0074, -0.0007, -0.0080, -0.0052, -0.0846,  0.0097, -0.0195,
        -0.0417, -0.0046, -0.0530,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0754, -0.0274,  0.0512, -0.0030, -0.0554, -0.0004, -0.0828, -0.1224,
        -0.1867,  0.0021,  0.0017,  0.0237,  0.0170,  0.0104, -0.0072,  0.0030,
        -0.0126, -0.0339, -0.0361, -0.0111, -0.0711, -0.1161,  0.0096,  0.0211,
        -0.0189,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0295, -0.3603, -0.1150,  0.0055, -0.0018, -0.0804, -0.1850, -0.0100,
        -0.0195, -0.0280, -0.0240, -0.0074, -0.0534, -0.0127, -0.0074, -0.0073,
         0.0527,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0232,  0.0042,  0.0025,  0.0057,  0.0078, -0.0095,  0.0034, -0.0065,
         0.0066, -0.1022, -0.0423, -0.0207, -0.0058,  0.0040, -0.0155, -0.0305,
        -0.0906, -0.1425, -0.0272, -0.0937, -0.0076, -0.1214, -0.1459, -0.0012,
         0.0252, -0.0083,  0.0086, -0.0094,  0.0061, -0.0038, -0.0066,  0.0117,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0203, -0.1466, -0.1453, -0.0014,  0.0292, -0.0127, -0.0857, -0.1279,
        -0.0006,  0.0051,  0.0084,  0.0040, -0.0075,  0.0008, -0.0175,  0.0280,
        -0.0172, -0.0122, -0.0503, -0.0359, -0.0538, -0.0063, -0.0517, -0.0873,
        -0.0064, -0.0228, -0.0112, -0.0039,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0780, -0.0186, -0.0120,  0.0062, -0.0328, -0.1361, -0.1727, -0.0299,
        -0.0245,  0.0049, -0.0036, -0.0637,  0.0090, -0.0012, -0.0148, -0.0804,
        -0.0192, -0.0417, -0.0114,  0.0223, -0.0023,  0.0087, -0.0183, -0.0004,
        -0.0861, -0.0074,  0.0245, -0.0121,  0.0574,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0058, -0.1339, -0.0012, -0.0510, -0.0905, -0.0123, -0.0107, -0.0076,
        -0.0011,  0.0239,  0.0125, -0.0111, -0.0454, -0.0744,  0.0312, -0.0430,
         0.0218, -0.0954,  0.0075, -0.0056,  0.0125,  0.0125, -0.0510, -0.0151,
        -0.0517, -0.0782, -0.0537, -0.0026, -0.0053, -0.0008, -0.0080,  0.0076,
         0.0152,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0307, -0.2810,  0.0074, -0.0629,  0.0137, -0.0476,  0.0109, -0.0345,
        -0.0234, -0.1110, -0.0559, -0.0139, -0.0050,  0.0007, -0.0092,  0.0108,
        -0.0400, -0.0015, -0.1161,  0.0200, -0.0824, -0.0108,  0.0011,  0.0096,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0147, -0.2881, -0.0982,  0.0054, -0.0577,  0.0088, -0.0121, -0.0407,
         0.0319, -0.0589, -0.0116, -0.0384, -0.0556,  0.0109,  0.0013,  0.0062,
         0.0014, -0.0573,  0.0205, -0.1038, -0.0200, -0.0264,  0.0300,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0459, -0.1322,  0.0163, -0.0071, -0.0259, -0.0347,  0.0273,  0.0036,
         0.0016,  0.0191,  0.0101, -0.0034, -0.0119,  0.0131, -0.0003, -0.0090,
        -0.0899, -0.0611, -0.0438, -0.0172,  0.0005, -0.0302, -0.0037,  0.0022,
        -0.0447, -0.0198, -0.0052, -0.0135, -0.0135, -0.0492, -0.0392, -0.0137,
        -0.0360, -0.0048, -0.0381, -0.0771, -0.0082, -0.0096, -0.0170,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.7623e-02, -6.5964e-02, -1.3521e-02, -4.3536e-02, -1.1852e-01,
        -7.2025e-03,  1.1794e-02,  1.0261e-02, -6.9773e-03,  2.6730e-02,
         1.1658e-03,  7.5188e-03, -2.5085e-03, -9.6653e-03,  4.5826e-05,
        -6.2240e-02, -1.6514e-02, -5.0961e-03,  1.8865e-03, -6.9129e-03,
        -6.0945e-02,  4.2495e-03, -6.5407e-03, -4.1146e-02, -8.5437e-02,
        -1.4867e-01, -9.2540e-03, -8.5003e-03, -3.3523e-03, -3.7734e-02,
        -4.7952e-03,  5.8968e-03,  9.8266e-04, -6.8786e-04, -3.0919e-03,
        -7.5033e-03, -1.3172e-02, -8.2554e-02,  7.3696e-04, -1.0116e-02,
        -4.3384e-03,  1.6846e-02, -7.7766e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.8959e-02,  2.8197e-01,  1.8190e-02, -7.8235e-03, -7.9624e-03,
         3.8729e-03, -4.5627e-05,  5.3325e-03,  6.4795e-02,  4.5734e-03,
        -1.3316e-04, -9.8162e-03,  2.8831e-02, -2.5754e-04, -3.0837e-03,
         1.1208e-02,  3.6193e-02,  6.4441e-02,  4.2350e-03,  8.0759e-03,
        -3.7827e-03,  1.0387e-02,  2.1900e-03,  5.8675e-04, -2.4893e-03,
         3.8068e-03, -8.1832e-03,  4.9492e-03,  1.8199e-03,  5.1150e-04,
        -8.9910e-03,  2.8327e-02,  7.6360e-02, -6.5696e-04,  2.8553e-03,
        -9.6581e-03, -1.5037e-02,  5.5273e-03,  4.3633e-02,  1.0604e-02,
         3.2381e-02,  5.1763e-02,  9.4059e-03, -9.2641e-03,  3.2012e-04,
         6.4264e-03,  1.9305e-03,  4.0576e-03, -3.2984e-03,  3.8642e-03,
        -2.2569e-03,  1.3507e-03, -8.1531e-03, -3.5442e-03,  4.1428e-03,
         2.1058e-02,  6.6333e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0027,  0.4968,  0.0540,  0.0085,  0.0469, -0.0135,  0.0800,  0.0922,
         0.0182,  0.0564,  0.0729,  0.0109,  0.0062, -0.0410,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0522, -0.2129, -0.0394,  0.0296, -0.0008, -0.0080, -0.0066,  0.0007,
        -0.0237, -0.0135, -0.0473, -0.1123, -0.0152, -0.0062, -0.0439,  0.0099,
        -0.0401, -0.0366,  0.0228, -0.0363, -0.0753, -0.0064, -0.0822,  0.0025,
         0.0109, -0.0139,  0.0059,  0.0239,  0.0032,  0.0178,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0077,  0.3559, -0.0167,  0.2792,  0.1717, -0.0880,  0.0024,  0.0783,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0333, -0.1186, -0.0041, -0.0224, -0.0452, -0.1323, -0.0025,  0.0010,
        -0.0048, -0.0106, -0.0058,  0.0081, -0.0598, -0.1275, -0.0043,  0.0046,
         0.0003, -0.0026, -0.0059, -0.0027,  0.0069,  0.0039,  0.0016,  0.0315,
        -0.0831, -0.0201, -0.0490,  0.0054, -0.0771, -0.0892, -0.0118,  0.0219,
        -0.0022,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0371, -0.2316, -0.0816, -0.0254, -0.0082, -0.0167, -0.0077, -0.0057,
         0.0102, -0.0371, -0.0052, -0.0086, -0.0100,  0.0096,  0.0044,  0.0105,
         0.0033,  0.0016, -0.0146,  0.0025,  0.0020,  0.0022,  0.0026,  0.0022,
         0.0090,  0.0085, -0.0056, -0.0632,  0.0014, -0.0056,  0.0022, -0.0037,
        -0.0293, -0.0341,  0.0042, -0.0057, -0.0104,  0.0063,  0.0009,  0.0044,
        -0.0008,  0.0027, -0.0094, -0.0354, -0.0045,  0.0032,  0.0048, -0.0010,
         0.0046,  0.0017, -0.0015, -0.0235, -0.0016, -0.0154, -0.0053,  0.0134,
        -0.0009,  0.0210,  0.0012, -0.0271, -0.0041, -0.0108, -0.0360, -0.0029,
         0.0019, -0.0027, -0.0106,  0.0026,  0.0054,  0.0070,  0.0013],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0196, -0.0284, -0.0427, -0.0030, -0.0744, -0.0296, -0.1298, -0.0331,
        -0.0056,  0.0093,  0.0005, -0.0059,  0.0150,  0.0158,  0.0182,  0.0159,
         0.0016,  0.0069, -0.0151, -0.0852, -0.0128,  0.0069,  0.0076, -0.0389,
         0.0052, -0.0365, -0.0197,  0.0063,  0.0058, -0.0094,  0.0036,  0.0120,
        -0.0617, -0.0548, -0.0192, -0.0136, -0.0719, -0.0374, -0.0013,  0.0047,
         0.0080, -0.0070,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0097, -0.2228, -0.0256, -0.0076,  0.0062,  0.0002,  0.0106, -0.0037,
         0.0027, -0.0762, -0.0298, -0.0351, -0.0411,  0.0114,  0.0013, -0.0112,
        -0.0105, -0.0607, -0.0835,  0.0073, -0.0161, -0.0273,  0.0074,  0.0030,
        -0.0047, -0.0803, -0.0187, -0.0072, -0.0645, -0.0046,  0.0031, -0.0126,
        -0.0068, -0.0184, -0.0072,  0.0089, -0.0519,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0729, -0.4995, -0.0172, -0.1074, -0.0024,  0.0250,  0.0238,  0.0197,
        -0.0156, -0.0015, -0.0393,  0.0293, -0.0752, -0.0210,  0.0503,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0496,  0.0085,  0.0245,  0.0144, -0.1497, -0.2598, -0.0209, -0.1182,
        -0.1202, -0.0095, -0.0092,  0.0186, -0.0947, -0.0132,  0.0013, -0.0058,
         0.0236, -0.0327,  0.0101,  0.0154,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.3067e-02,  1.0141e-02,  5.6741e-03,  6.5076e-03, -1.3774e-01,
        -3.8930e-03, -7.9175e-03, -1.2511e-02,  1.2243e-04,  1.7351e-03,
        -3.7879e-02, -6.3884e-02, -2.4219e-02, -7.9623e-03, -5.1183e-02,
        -2.3834e-02,  3.0219e-03, -3.2105e-02,  1.4776e-02, -4.2451e-02,
        -3.7149e-03, -7.6834e-03, -5.1093e-03, -5.3821e-03,  4.2025e-03,
        -5.8918e-02, -8.6664e-02,  9.0107e-03, -7.1552e-03, -3.0823e-02,
        -6.2067e-02, -1.7962e-02,  5.2507e-03,  1.2677e-03, -3.4677e-02,
        -5.9878e-02,  5.9920e-03, -2.1742e-02, -1.2582e-02, -4.2684e-03,
        -1.3585e-05,  2.2378e-02,  1.2633e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.6161e-02,  8.0874e-03, -2.1193e-02,  9.6417e-03, -4.9633e-02,
        -1.7165e-02, -3.3163e-02, -1.2745e-01, -3.3257e-02,  3.9816e-03,
         2.4890e-02,  1.7022e-02,  7.5912e-03, -9.7661e-02, -8.4032e-02,
         1.3472e-03,  2.4158e-03, -2.6466e-03,  7.7368e-03, -7.7933e-03,
        -3.2977e-02, -1.1149e-02,  3.8313e-03,  1.9169e-02, -1.0246e-02,
         3.2969e-04,  3.7269e-03, -2.0668e-02, -1.0904e-01, -3.9369e-05,
        -1.6715e-02, -4.5940e-02, -9.6484e-03, -4.9278e-02, -5.9213e-02,
        -1.1730e-02, -1.6241e-02,  7.1848e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.5788e-04, -3.0548e-01, -9.7550e-02, -1.1636e-01,  4.6649e-04,
        -7.4040e-02,  6.4296e-03,  6.0814e-03, -5.2835e-02, -9.0233e-02,
        -1.3067e-02, -2.5389e-02, -7.3342e-02, -7.9638e-03, -1.4649e-03,
        -1.0057e-02, -3.7145e-03, -5.9187e-02, -2.4880e-02,  1.9257e-02,
         1.1642e-02,  2.0936e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.1930e-02, -1.1389e-01, -1.0198e-02,  3.8922e-03, -4.2168e-03,
        -2.0628e-02, -1.8986e-02,  4.0066e-02,  2.6625e-03, -1.1358e-01,
        -8.9427e-03, -2.2810e-03, -1.7759e-02,  1.6768e-03, -5.9906e-02,
        -6.7815e-02, -1.1375e-01, -9.1469e-03, -1.1794e-02, -6.7695e-02,
        -1.1604e-03, -7.4283e-04, -5.9160e-02, -6.1822e-03, -2.3741e-03,
         6.8910e-03, -5.8927e-03,  3.2837e-03,  4.6145e-03, -5.4693e-02,
        -1.0081e-02, -1.6549e-02,  1.2258e-02, -2.7697e-03, -2.2340e-02,
        -4.9018e-02,  1.3124e-02, -1.1644e-02,  7.7021e-05,  6.3273e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0003,  0.0027,  0.0029, -0.0101,  0.0101, -0.0822,  0.0072,  0.0104,
        -0.0602, -0.0040,  0.0017, -0.0055, -0.0265, -0.1292, -0.0630, -0.0051,
         0.0248, -0.0078, -0.0909, -0.0030, -0.0155,  0.0065,  0.0063, -0.0121,
        -0.0324, -0.1435,  0.0029, -0.0690, -0.0766, -0.0120, -0.0338,  0.0418,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.8828e-02, -3.4330e-01, -7.2069e-03,  2.0895e-02, -1.8718e-02,
         1.3350e-02,  3.3876e-03, -5.3858e-02, -1.4882e-01,  1.4753e-03,
         4.8893e-03, -4.7258e-02,  1.8719e-04, -4.4502e-03,  7.4197e-05,
        -5.2022e-03, -4.9570e-02, -9.5826e-03,  6.4057e-03,  1.3301e-02,
        -9.8568e-04, -5.1647e-04, -1.4043e-02, -5.8640e-02,  6.4957e-03,
         8.2233e-04, -4.7788e-02, -3.7297e-03,  1.5203e-03, -1.9000e-02,
         1.0695e-02, -1.3794e-03, -9.5477e-03,  3.1971e-02, -1.2107e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([-0.0654,  0.0033, -0.1165, -0.1329, -0.0282, -0.0177, -0.1615, -0.1045,
        -0.0514, -0.0780,  0.0050, -0.0679, -0.0052, -0.0317, -0.0133, -0.0244,
        -0.0428, -0.0149,  0.0133,  0.0161, -0.0057,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0308, -0.3619,  0.0350, -0.0699, -0.1034, -0.0452,  0.0134, -0.0602,
        -0.0764,  0.0065, -0.0125,  0.0207,  0.0128, -0.0157, -0.0603, -0.0019,
        -0.0338,  0.0204, -0.0194,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0377, -0.0313, -0.1171,  0.0159, -0.0566, -0.0171,  0.0066, -0.0145,
        -0.0046, -0.0722, -0.2021, -0.0126, -0.0552, -0.2234, -0.0200, -0.0395,
         0.0043, -0.0252,  0.0429, -0.0012,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0340, -0.3988,  0.0011, -0.0997,  0.0171, -0.0122,  0.0093, -0.0493,
        -0.0055, -0.0042, -0.0046, -0.0218, -0.0072, -0.0573, -0.0231, -0.0518,
        -0.0543,  0.0036, -0.0462, -0.0587,  0.0089,  0.0112, -0.0203,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0129, -0.1718, -0.0656, -0.0353, -0.0176, -0.0882, -0.0002, -0.0571,
        -0.0837, -0.1105, -0.0212, -0.0704, -0.0131, -0.1070, -0.0115, -0.0073,
        -0.0082,  0.0104, -0.0082,  0.0111, -0.0395, -0.0111, -0.0160, -0.0039,
         0.0183,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.3263e-02, -3.0106e-01,  9.7802e-03, -1.9601e-02, -1.0811e-02,
        -3.6166e-03,  1.5338e-02, -1.0536e-02,  1.2644e-02,  1.0210e-02,
        -7.1511e-02, -1.8639e-02, -1.2146e-03, -2.1513e-04,  4.5005e-04,
         2.2362e-04, -1.7172e-02, -7.4686e-04, -1.6123e-02, -1.4024e-02,
        -5.2064e-02, -9.1064e-02, -5.2878e-02, -2.7985e-02, -3.4677e-02,
        -1.4334e-03, -1.1136e-02, -5.5040e-03, -3.8641e-04, -4.2720e-02,
        -3.5246e-02, -8.7002e-03, -8.6783e-03,  4.2519e-03, -4.6094e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0696,  0.0035,  0.0286, -0.0935,  0.0013,  0.0411, -0.0765, -0.0190,
        -0.1039, -0.0342, -0.0197, -0.0238,  0.0042, -0.0399, -0.0700, -0.1481,
        -0.0058, -0.1261,  0.0004,  0.0218, -0.0345,  0.0343,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1224,  0.0161, -0.0300,  0.0133,  0.0287,  0.0486,  0.0338, -0.0089,
         0.0656,  0.3296, -0.0024, -0.0033, -0.0303,  0.0214,  0.0622,  0.0730,
         0.0301, -0.0048,  0.0164, -0.0022, -0.0568,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0224,  0.3927,  0.0311,  0.0017,  0.0225, -0.0143,  0.0965,  0.0584,
        -0.0106, -0.0105, -0.0246, -0.0035, -0.0194,  0.1268,  0.0169,  0.0577,
        -0.0151, -0.0536,  0.0218,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0295,  0.0046, -0.0349, -0.0033,  0.0018, -0.0481, -0.0056,  0.0029,
         0.0054, -0.0081, -0.0057, -0.0302, -0.0090, -0.0395,  0.0063,  0.0037,
        -0.0465, -0.0652, -0.0180, -0.0462, -0.0095, -0.0099, -0.0373,  0.0038,
        -0.0643, -0.0205, -0.0090, -0.0451,  0.0138, -0.1106, -0.0121, -0.0103,
        -0.0454, -0.0872,  0.0112, -0.0017, -0.0015, -0.0013,  0.0083, -0.0160,
         0.0056,  0.0100,  0.0053, -0.0019, -0.0057,  0.0343,  0.0038],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0615, -0.1976, -0.1057, -0.0026,  0.0148,  0.0126, -0.0051,  0.0106,
        -0.0375,  0.0061, -0.0004, -0.0092,  0.0089, -0.0127,  0.0105,  0.0099,
        -0.0685, -0.0027, -0.0026, -0.0178,  0.0084, -0.0427, -0.0951, -0.1330,
         0.0032,  0.0063, -0.0327,  0.0061,  0.0008,  0.0037,  0.0002,  0.0117,
        -0.0256,  0.0331,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0247,  0.3354,  0.0115,  0.0041,  0.0140,  0.0611, -0.0014,  0.0062,
        -0.0022,  0.0338, -0.0212, -0.0066,  0.0449,  0.0937,  0.0025, -0.0165,
         0.0020,  0.0081,  0.0089,  0.1267,  0.0175,  0.0178,  0.0128,  0.0003,
         0.0130, -0.0015, -0.0086,  0.0130,  0.0082,  0.0321,  0.0038, -0.0144,
         0.0072,  0.0246,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0515, -0.4661, -0.0037, -0.0178, -0.0155, -0.0661,  0.0231,  0.0333,
        -0.0247, -0.0765, -0.0132, -0.0039, -0.0296, -0.1024, -0.0083, -0.0153,
        -0.0317, -0.0175,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0262, -0.0082, -0.0116,  0.0272,  0.0077,  0.0051,  0.0004, -0.0084,
        -0.1642, -0.2723, -0.0312, -0.0024,  0.0066, -0.0512, -0.0085,  0.0232,
        -0.0299, -0.0113,  0.0315, -0.1881, -0.0129, -0.0219,  0.0038,  0.0071,
         0.0109,  0.0275,  0.0008,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0694, -0.0111,  0.0011, -0.0113,  0.0434,  0.0045, -0.0011, -0.0161,
        -0.1108, -0.0116,  0.0077, -0.0324, -0.1526, -0.0427,  0.0187, -0.0003,
        -0.0233, -0.0323, -0.0081, -0.2633, -0.0320,  0.0069, -0.0657, -0.0055,
        -0.0180, -0.0100,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.1771e-02, -2.7096e-01, -2.1270e-04, -1.1616e-01, -8.4329e-02,
        -2.4987e-03, -1.0908e-02,  5.1004e-03, -6.7224e-02, -3.4624e-02,
        -8.7292e-02, -5.9069e-02,  7.4672e-03,  1.1419e-02, -9.4738e-03,
        -5.0373e-02, -1.8761e-02, -2.1863e-02, -1.6512e-03,  2.3265e-02,
         3.5576e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0142, -0.3412, -0.1858, -0.1633, -0.0100,  0.0302, -0.0882,  0.0109,
        -0.0036,  0.0153,  0.0272, -0.0447,  0.0156,  0.0195,  0.0301,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1065,  0.0171,  0.0037,  0.0057,  0.0223,  0.0293,  0.0696, -0.0522,
         0.0216, -0.0159, -0.0056, -0.0524, -0.3956, -0.0383,  0.0963,  0.0053,
         0.0379, -0.0245,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.2973e-02,  3.3655e-01,  1.0265e-01,  1.1786e-01,  1.7386e-02,
         1.1280e-02,  6.3911e-03,  3.3045e-02, -1.9365e-04,  9.5451e-02,
         6.9219e-02,  3.0579e-02, -6.1099e-03,  4.8709e-02,  3.3818e-03,
        -2.7839e-02, -1.6291e-02, -2.2359e-02, -4.1735e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0239, -0.0863, -0.1061, -0.0262, -0.0114, -0.0189, -0.0026, -0.0013,
        -0.0084, -0.0019, -0.0036,  0.0014, -0.0217, -0.0005, -0.0055,  0.0155,
        -0.0415, -0.0042,  0.0042,  0.0005, -0.0651,  0.0048, -0.0510, -0.0026,
        -0.0686, -0.0061, -0.0583,  0.0116, -0.0004, -0.0165, -0.0088,  0.0183,
        -0.0587, -0.0070, -0.0467, -0.0193, -0.0922, -0.0077, -0.0149, -0.0278,
        -0.0074, -0.0004,  0.0112, -0.0089], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0008, -0.2285, -0.0093, -0.1072, -0.0564, -0.0213, -0.0373, -0.0039,
        -0.1123, -0.0111, -0.0333,  0.0097, -0.0081, -0.0023,  0.0139,  0.0047,
         0.0384, -0.1019,  0.0064, -0.0251, -0.0365,  0.0011,  0.0005,  0.0032,
         0.0003,  0.0021,  0.0038, -0.0377,  0.0090, -0.0188, -0.0091, -0.0247,
         0.0027,  0.0030,  0.0155,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0044, -0.4218, -0.0011, -0.0757, -0.0164,  0.0083, -0.1051, -0.0690,
        -0.0023,  0.0037,  0.0047, -0.0442, -0.0978, -0.0016,  0.0153, -0.0711,
         0.0056,  0.0031, -0.0068,  0.0043,  0.0085, -0.0057,  0.0153, -0.0081,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0790, -0.2243,  0.0006, -0.0629, -0.0191, -0.0107,  0.0219,  0.0004,
        -0.0678,  0.0041, -0.0141,  0.0177,  0.0122, -0.0239, -0.0194, -0.0093,
         0.0047, -0.0044, -0.0426, -0.0137, -0.0074, -0.0553,  0.0086, -0.0347,
         0.0016, -0.0311, -0.0417, -0.0243,  0.0025, -0.0148, -0.0157, -0.0041,
         0.0577, -0.0477,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.6880e-02,  2.0158e-02,  1.3045e-02, -2.3597e-02,  1.0590e-02,
         1.1291e-02, -9.0948e-02,  8.3832e-05,  4.6534e-03,  4.3128e-04,
         1.9608e-02,  2.1401e-02, -7.2587e-02, -6.4098e-03, -1.0870e-01,
        -2.4585e-01, -5.5750e-03, -2.5358e-03,  1.9072e-03,  1.4302e-02,
         1.0348e-02, -6.8615e-03,  2.6105e-02, -1.0743e-01, -4.0467e-02,
         3.5794e-03, -5.0401e-02, -1.0847e-02, -2.2675e-02, -7.4246e-04,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0110,  0.0733, -0.0597, -0.3414, -0.0157, -0.0287, -0.0079, -0.0291,
        -0.1460, -0.1852, -0.0345, -0.0473,  0.0202,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0649,  0.0101, -0.0051, -0.0078,  0.0069, -0.0609, -0.0539, -0.1935,
        -0.1943,  0.0108,  0.0090, -0.0883,  0.0249, -0.0296, -0.1403,  0.0207,
         0.0307, -0.0483,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0132, -0.3297, -0.2975, -0.0025, -0.2022, -0.0488, -0.1041, -0.0020,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0100, -0.1885, -0.0265, -0.0417, -0.0175, -0.0140, -0.0127, -0.0111,
         0.0089, -0.0103,  0.0020,  0.0035, -0.0143,  0.0022, -0.0078, -0.0128,
        -0.0013,  0.0008, -0.0032, -0.0026,  0.0097, -0.0063, -0.0417, -0.0340,
        -0.0071, -0.0320, -0.0033, -0.0161, -0.0572, -0.0654,  0.0058, -0.0673,
        -0.0147, -0.0644, -0.0165, -0.0272, -0.0370, -0.0031,  0.0056,  0.0024,
        -0.0166, -0.0026, -0.0147,  0.0015,  0.0048,  0.0042,  0.0469],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1231, -0.2651,  0.1671, -0.2006, -0.0829,  0.0440, -0.0236, -0.0064,
        -0.0534,  0.0071,  0.0268,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0678, -0.3268, -0.0865, -0.1087, -0.0157,  0.0014, -0.0585, -0.1349,
        -0.0105, -0.0692, -0.0156, -0.0084, -0.0351,  0.0083,  0.0094, -0.0060,
         0.0207, -0.0166,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0173, -0.1421, -0.4405,  0.0102,  0.0056, -0.0164, -0.2035, -0.0121,
        -0.0791, -0.0053,  0.0020,  0.0088, -0.0073,  0.0431,  0.0067,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0747, -0.3251, -0.0074, -0.0083, -0.0209, -0.0844, -0.2228, -0.0033,
         0.0254, -0.0704, -0.1050,  0.0140,  0.0107, -0.0126,  0.0150,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0070,  0.0078,  0.0124,  0.0118, -0.0200,  0.0095,  0.0057,  0.0139,
        -0.1182,  0.0106,  0.0252, -0.1034, -0.1630,  0.0184, -0.0088,  0.0011,
        -0.0754, -0.0043, -0.0033, -0.0152, -0.1388, -0.1783,  0.0088,  0.0080,
        -0.0094, -0.0213,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1048, -0.0367,  0.0204, -0.0086,  0.0210,  0.0107, -0.0456, -0.2752,
        -0.2667, -0.0533, -0.0014,  0.0399,  0.0514,  0.0343, -0.0299,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.4420e-02, -3.2403e-01, -2.2936e-02, -5.3813e-02, -4.8120e-02,
        -1.9469e-02, -5.2123e-02,  3.0054e-04, -4.9587e-04, -9.2476e-02,
         2.0229e-02,  1.6389e-02, -7.5655e-02, -6.9818e-03,  4.5584e-03,
         1.5155e-03, -2.0002e-02,  1.2014e-02, -3.1332e-02,  1.7708e-02,
        -1.1025e-02,  1.1556e-02,  2.8831e-02,  1.8599e-02, -1.5705e-02,
         4.9719e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0359, -0.0060,  0.0189,  0.0287, -0.0065, -0.0190,  0.0014, -0.0048,
        -0.0313, -0.1097, -0.1311, -0.0089, -0.0338,  0.0112, -0.1105, -0.0991,
         0.0218, -0.0331,  0.0268, -0.0532,  0.0166, -0.0146, -0.0139, -0.0527,
         0.0964, -0.0142,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-7.3275e-02, -2.9492e-01, -2.3406e-03, -9.2994e-04, -2.2038e-02,
         8.9894e-03, -4.8582e-02, -5.1757e-02, -1.2272e-02, -2.5430e-02,
        -1.1451e-02, -3.1037e-02, -4.4951e-02, -6.5458e-03, -2.4314e-02,
        -8.3338e-03,  2.6472e-04, -4.3384e-03, -1.7657e-02, -9.9701e-03,
         1.0452e-02, -1.8921e-02,  1.1402e-02, -2.7681e-02, -5.9971e-02,
         1.7997e-02, -3.7072e-02, -8.2578e-03,  3.7916e-02, -1.4621e-03,
        -8.1472e-03, -2.8471e-02, -1.3406e-02, -6.2822e-03,  6.4232e-03,
         6.7446e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0346, -0.1048, -0.0244, -0.0191, -0.0382, -0.0099, -0.0700, -0.0602,
         0.0069, -0.0125, -0.0261,  0.0034, -0.0037,  0.0145, -0.0008, -0.0079,
         0.0034, -0.0025,  0.0091,  0.0050,  0.0060, -0.0661, -0.0039,  0.0104,
        -0.0341, -0.0684,  0.0075, -0.0548, -0.0523,  0.0006, -0.0192, -0.0490,
        -0.0678,  0.0215, -0.0287,  0.0528,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0711,  0.0072,  0.0222, -0.0232,  0.0104, -0.1953, -0.0040, -0.0184,
         0.0040,  0.0084, -0.0404, -0.1078, -0.2734, -0.0004, -0.0101, -0.0079,
        -0.0311, -0.0268, -0.0402,  0.0977,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0046, -0.1193, -0.0325, -0.0422, -0.0039, -0.0054, -0.0309, -0.0775,
        -0.0019, -0.0246,  0.0105, -0.0064,  0.0006, -0.0014, -0.0962, -0.0846,
        -0.0268, -0.0562, -0.0076, -0.0033, -0.0151, -0.0037, -0.0019,  0.0051,
        -0.0051,  0.0058,  0.0010, -0.0049,  0.0165, -0.0552, -0.0944, -0.0355,
        -0.0116, -0.0780,  0.0011,  0.0120, -0.0038,  0.0129,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1818e-02, -2.7553e-01, -1.1961e-01, -7.6624e-02, -1.2298e-02,
         2.0899e-03, -4.1670e-04,  1.4259e-03, -1.2710e-02,  9.5130e-03,
        -4.4279e-02, -6.8547e-02,  7.4522e-03, -3.3887e-03,  6.0845e-03,
        -4.5399e-02,  7.7951e-03, -4.8253e-02,  4.8524e-04, -1.4214e-04,
         1.0119e-02,  1.3858e-02,  1.5298e-02,  3.6432e-03, -9.9129e-03,
         7.0834e-03, -7.7143e-03, -2.3946e-02, -4.8669e-02,  8.0690e-03,
        -1.8565e-03, -1.2585e-02,  5.3030e-03, -9.5761e-05, -1.2322e-02,
        -5.7781e-03,  4.8728e-03,  2.1354e-02, -4.1176e-03, -2.7796e-02,
        -1.7452e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0061, -0.3642, -0.0432,  0.0054,  0.0017, -0.0648, -0.0289, -0.0117,
        -0.0245,  0.0075,  0.0023, -0.1305, -0.0825, -0.0230,  0.0020, -0.0701,
        -0.0128, -0.0492, -0.0074,  0.0160,  0.0173, -0.0292,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0255,  0.0096,  0.0068,  0.0020,  0.0087, -0.0388, -0.0039,  0.0052,
        -0.0049,  0.0033,  0.0132, -0.0778, -0.0019, -0.0176, -0.0153, -0.0848,
         0.0241, -0.0700, -0.0201, -0.0788, -0.1583,  0.0146,  0.0098, -0.0732,
         0.0186, -0.0088, -0.0101, -0.0244, -0.0748,  0.0012, -0.0251, -0.0461,
        -0.0123, -0.0042,  0.0064,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.2683e-02, -4.6254e-02, -2.3311e-01, -7.0048e-03, -1.0290e-01,
        -1.3434e-02,  3.3484e-02,  2.0099e-04,  7.9871e-03,  1.6601e-03,
        -6.5624e-02, -3.1999e-04, -4.1105e-02,  1.4935e-02, -4.5284e-03,
         3.3949e-03,  4.2897e-03,  5.8148e-03, -1.0977e-01,  3.6963e-03,
        -4.6128e-02,  4.0018e-04, -2.8628e-02, -7.6878e-02,  7.5547e-03,
        -6.1975e-03, -3.8077e-02, -1.7389e-02,  6.5484e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0811, -0.3272, -0.0698, -0.0744, -0.0156, -0.0758, -0.0104,  0.0076,
        -0.0319,  0.0030, -0.0333, -0.0181,  0.0188,  0.0127, -0.0006, -0.0450,
        -0.0302,  0.0166, -0.0356, -0.0033,  0.0045, -0.0086, -0.0478,  0.0077,
         0.0204,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0288,  0.0049,  0.0114, -0.0026, -0.2459,  0.0410, -0.0109, -0.1333,
        -0.1559, -0.0370,  0.0365,  0.0019,  0.0022, -0.0027, -0.0884, -0.0744,
         0.0204, -0.0352,  0.0059, -0.0118,  0.0078,  0.0398, -0.0013,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0775, -0.0074,  0.0099, -0.0017, -0.0015,  0.0196,  0.0079, -0.0004,
         0.0014,  0.0045,  0.0060,  0.0039, -0.0517, -0.1181,  0.0116, -0.0076,
        -0.0069, -0.0359,  0.0070, -0.0005, -0.0130,  0.0073,  0.0060, -0.0062,
        -0.0103, -0.0115, -0.1521, -0.0429, -0.1101, -0.0162, -0.0765, -0.0117,
        -0.0040, -0.0395, -0.0301, -0.0309, -0.0061, -0.0387, -0.0059,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0612,  0.0315, -0.0314,  0.1095,  0.2622,  0.0087,  0.0350,  0.1503,
         0.1908, -0.0087,  0.0577, -0.0530,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.0027, -0.1315, -0.2525, -0.0072, -0.0109, -0.0217, -0.0732,  0.0170,
        -0.0122, -0.0405, -0.0051,  0.0054, -0.0230, -0.0050, -0.0693, -0.0077,
        -0.0212, -0.0283,  0.0007,  0.0029, -0.0008, -0.0266,  0.0139, -0.0024,
        -0.0003, -0.0039,  0.0080,  0.0021,  0.0022, -0.0006,  0.0118,  0.0146,
        -0.0508,  0.0021,  0.0066,  0.0059, -0.0496, -0.0103,  0.0011, -0.0040,
         0.0077,  0.0005, -0.0035,  0.0150,  0.0176,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1332, -0.4796, -0.0055, -0.0876,  0.0514, -0.0092,  0.0280, -0.0271,
        -0.1120,  0.0342,  0.0213, -0.0108,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0672, -0.6600,  0.0181, -0.0808, -0.0100, -0.0460, -0.0444,  0.0735,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.5260e-02, -1.6870e-01, -2.7085e-02, -2.2683e-03, -3.7348e-02,
        -1.5704e-03,  7.9657e-03, -9.5502e-03, -3.9028e-02, -9.4947e-02,
        -1.7414e-02, -9.8162e-03,  3.5687e-03, -1.6249e-03,  6.0458e-04,
         4.1150e-03, -4.1848e-03,  1.5116e-02, -6.0303e-03, -3.1818e-03,
         3.7472e-03,  5.5130e-03,  9.9601e-05, -1.3171e-02, -5.5725e-04,
        -3.5176e-03,  3.6983e-03, -3.8318e-03, -9.4321e-03,  2.4781e-03,
         8.1466e-05,  9.0207e-04,  6.1371e-04, -1.7315e-03, -2.6263e-03,
        -3.2395e-03,  1.3621e-03,  3.0664e-03, -3.4182e-03, -7.8833e-04,
         1.5258e-03,  4.0589e-03, -7.8995e-03, -5.2475e-03,  5.6077e-03,
         1.1465e-02, -7.8436e-02,  8.0470e-03, -6.1757e-02, -1.1274e-01,
        -2.1915e-02,  7.3049e-04, -1.7457e-02, -1.3829e-03, -6.2542e-03,
         2.3017e-03, -1.7488e-03,  4.2376e-03,  6.2852e-03,  3.3194e-02,
        -4.4457e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0112, -0.0211,  0.0029, -0.0605, -0.1057, -0.0433, -0.0481, -0.0016,
         0.0045,  0.0072, -0.0143, -0.0654,  0.0008, -0.0630, -0.0648, -0.0178,
        -0.0004, -0.0115, -0.0358, -0.0152, -0.0290, -0.0937, -0.0160, -0.0063,
        -0.0430, -0.0551, -0.0406, -0.0191, -0.0293, -0.0422, -0.0166, -0.0096,
         0.0044,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0735,  0.0429,  0.1326,  0.0915,  0.0277, -0.0040,  0.0626, -0.0110,
        -0.0157,  0.0079, -0.0060,  0.0008,  0.0660,  0.0106,  0.0190, -0.0334,
         0.0065,  0.0899,  0.1792, -0.0082, -0.0274,  0.0196, -0.0290,  0.0347,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0348, -0.1178, -0.1284, -0.0265, -0.0122, -0.0969,  0.0285, -0.0464,
        -0.0185, -0.0954, -0.0565, -0.0482, -0.1382, -0.0083, -0.0037,  0.0066,
         0.0149,  0.0061,  0.0164,  0.0181, -0.0167, -0.0032, -0.0077, -0.0502,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 8.0641e-02, -1.4030e-02, -6.2022e-02, -6.8784e-02, -4.6545e-03,
        -6.3026e-03, -1.0520e-03, -6.5005e-02,  1.1175e-02,  1.3615e-02,
         7.3065e-03, -9.8615e-03,  1.3360e-02,  3.8087e-03, -5.4692e-02,
        -8.1544e-02, -6.5029e-02,  8.3541e-03,  1.5474e-05, -1.7413e-02,
        -7.7992e-02, -4.1057e-03, -3.9341e-02, -1.1916e-01, -1.6836e-02,
        -9.3434e-03, -4.1846e-02, -6.3052e-03,  1.1024e-02, -2.7622e-02,
        -1.9623e-02, -4.6759e-03,  3.4628e-03, -2.9998e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0359, -0.0154,  0.0085,  0.0053, -0.0787, -0.0055, -0.0091, -0.0074,
        -0.0025,  0.0053,  0.0264,  0.0311,  0.0089, -0.1657, -0.0160, -0.0227,
        -0.1246,  0.0054, -0.0800, -0.0098, -0.0709, -0.1824,  0.0213,  0.0117,
        -0.0494,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0005,  0.0317, -0.0073, -0.0098, -0.0267,  0.0437, -0.2232, -0.1650,
        -0.1128, -0.2193, -0.0115, -0.0382, -0.0499,  0.0086, -0.0171, -0.0347,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0126, -0.2423, -0.0067, -0.1492, -0.0503, -0.0961, -0.0601, -0.0143,
         0.0501, -0.0064,  0.0060,  0.0108, -0.0569, -0.0057,  0.0044, -0.0058,
        -0.0615, -0.0842, -0.0369, -0.0246,  0.0121,  0.0013,  0.0019,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0261, -0.0097, -0.0037, -0.2317, -0.3987, -0.0690,  0.0327,  0.0175,
        -0.0263,  0.0559, -0.0152, -0.0430,  0.0368, -0.0339,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0107, -0.1223, -0.1253, -0.1616, -0.0191,  0.0172, -0.0535,  0.0291,
        -0.0528, -0.0233, -0.0440, -0.0787, -0.0286,  0.0006, -0.0050,  0.0047,
        -0.0159, -0.0595, -0.0025, -0.0559, -0.0115,  0.0068,  0.0270,  0.0445,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0058,  0.0460, -0.1846, -0.0091, -0.2690, -0.0636, -0.2925,  0.0384,
        -0.0083, -0.0283,  0.0543,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0373, -0.0362, -0.0120, -0.0053,  0.0254, -0.0950, -0.2090, -0.0078,
        -0.0962, -0.0009, -0.0610,  0.0099, -0.0490, -0.0139, -0.0893, -0.0028,
        -0.0426, -0.1171, -0.0118, -0.0512, -0.0264,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0266,  0.0288, -0.0051,  0.0415,  0.0310, -0.0094, -0.0185, -0.3483,
        -0.1631, -0.1997, -0.0248,  0.0183, -0.0075, -0.0773,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0128,  0.0209, -0.0221, -0.0199,  0.0044, -0.0011,  0.0295, -0.0907,
        -0.0441,  0.0314, -0.0215,  0.0580,  0.4200, -0.0346, -0.0266,  0.0364,
        -0.0195,  0.0202,  0.0009,  0.0309, -0.0545,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0204, -0.3231, -0.0069, -0.1639,  0.0415, -0.1423, -0.0331, -0.1507,
        -0.0011, -0.0683, -0.0113, -0.0041,  0.0333,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0532, -0.1745, -0.0061, -0.0066, -0.0129, -0.0104, -0.0427, -0.0400,
        -0.0267,  0.0005,  0.0005,  0.0191,  0.0045,  0.0070,  0.0105, -0.0258,
        -0.0013, -0.0020, -0.0062,  0.0180,  0.0132,  0.0278,  0.0215, -0.0141,
        -0.0350, -0.0188, -0.0918, -0.0566, -0.0661,  0.0031,  0.0104,  0.0344,
         0.0025, -0.0004,  0.0019, -0.0018, -0.0179, -0.0003, -0.0022, -0.0106,
        -0.0151, -0.0038, -0.0113,  0.0367,  0.0342], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0150, -0.0326, -0.0898, -0.0187,  0.0009, -0.0081, -0.0119, -0.0123,
        -0.0104,  0.0044, -0.0909,  0.0037,  0.0074,  0.0013, -0.0308, -0.0159,
        -0.1277,  0.0064,  0.0056, -0.0024, -0.0199, -0.0756,  0.0120, -0.0911,
        -0.0043,  0.0077, -0.0243, -0.0248, -0.1462, -0.0310, -0.0293, -0.0057,
        -0.0283, -0.0033,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0353, -0.0401, -0.0193,  0.0731,  0.0195,  0.0223, -0.0162, -0.0058,
        -0.0010,  0.0010,  0.0213,  0.0077,  0.0283, -0.0023, -0.0030, -0.0161,
        -0.0093, -0.0097,  0.0434,  0.2656,  0.2789,  0.0227,  0.0073,  0.0307,
        -0.0089, -0.0109,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.0950e-02, -3.2870e-01,  6.3104e-03, -1.0401e-01,  9.7321e-03,
        -1.0746e-02, -5.3110e-03, -5.8434e-02,  1.6663e-03,  3.0356e-02,
        -2.3996e-03, -2.2656e-04,  8.8421e-03, -1.2586e-02, -2.6902e-02,
        -6.3668e-02, -1.0558e-01, -1.1245e-02, -6.7523e-02, -2.2540e-02,
        -3.9850e-02,  1.1668e-02, -1.0755e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.0019e-02, -1.9496e-01, -3.1844e-02, -2.8880e-02, -1.2393e-04,
        -6.5236e-03, -1.0617e-02, -2.1930e-02, -7.0154e-02, -8.2414e-03,
         6.8403e-03, -1.3614e-02,  5.5648e-03, -6.6641e-03, -1.8292e-01,
         2.4737e-02,  1.9840e-03, -3.3796e-02, -6.3903e-02,  1.0012e-02,
         4.1280e-03, -6.6914e-03,  5.0276e-03,  1.1031e-02,  4.6503e-03,
         6.0308e-03, -4.5080e-02, -8.7265e-02,  5.8689e-03, -4.6129e-03,
        -7.6545e-03, -8.1707e-04, -1.6467e-03, -1.8202e-03, -1.2864e-02,
        -7.9173e-03, -1.3011e-02,  4.4675e-03,  7.7926e-03,  1.8290e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1378e-02,  2.2670e-02,  1.2502e-02, -4.6084e-02, -8.3803e-02,
        -9.0420e-03, -3.1624e-02, -6.5843e-02,  2.1899e-02, -1.7071e-02,
         6.5459e-03,  3.0300e-03, -4.7255e-02, -1.5021e-02, -5.7178e-02,
        -6.4854e-02, -8.1031e-02,  5.1360e-03,  3.5611e-02, -3.8159e-02,
        -9.1047e-03, -5.3580e-02, -6.3495e-02,  5.1303e-03, -2.2620e-03,
        -1.4901e-03, -1.5437e-05, -3.2979e-02, -8.2870e-03, -1.7596e-02,
        -7.4925e-02, -5.5872e-05, -3.7537e-03,  5.1591e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0178, -0.0522, -0.0447, -0.0237, -0.0767, -0.0039,  0.0176,  0.0065,
         0.0031,  0.0010, -0.0035, -0.0184,  0.0084, -0.0163, -0.1834,  0.0082,
         0.0040, -0.0119, -0.0468,  0.0008,  0.0043,  0.0016,  0.0007,  0.0034,
        -0.0331, -0.0410,  0.0034, -0.0062,  0.0051,  0.0093, -0.0038,  0.0065,
         0.0029, -0.0099, -0.0376,  0.0009, -0.0192, -0.0295, -0.0033, -0.0459,
        -0.0508, -0.0218, -0.0345, -0.0101,  0.0063,  0.0093, -0.0010, -0.0016,
        -0.0085, -0.0051, -0.0122,  0.0223], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0681,  0.0350, -0.0030,  0.0263, -0.1017, -0.0048,  0.0141,  0.0220,
         0.0080,  0.0336, -0.0024,  0.0105, -0.0146, -0.1447, -0.0044, -0.0312,
        -0.0971, -0.0375, -0.0538, -0.0106, -0.0081, -0.0037, -0.0091, -0.1055,
        -0.0098, -0.0008,  0.0209, -0.0057,  0.0019,  0.0282, -0.0829,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0515, -0.1879,  0.0095, -0.0377,  0.0046,  0.0238, -0.0208, -0.0128,
        -0.0060,  0.0404, -0.0077, -0.0594,  0.0402,  0.0123, -0.0621, -0.0945,
         0.0062, -0.0488, -0.0899, -0.0762, -0.0280, -0.0122, -0.0406,  0.0256,
         0.0012,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0023, -0.0436,  0.0523,  0.0335,  0.0469,  0.2788,  0.4554,  0.0028,
         0.0035, -0.0636,  0.0130,  0.0043,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0478, -0.2818, -0.0325, -0.0673, -0.0912, -0.0221, -0.0117, -0.0044,
        -0.0153, -0.0021, -0.0170, -0.0035, -0.0488, -0.0883, -0.0140, -0.0330,
        -0.0116, -0.0079,  0.0048,  0.0007, -0.0327, -0.0075, -0.0237, -0.0333,
        -0.0175, -0.0467, -0.0119, -0.0111, -0.0101,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0173, -0.4428, -0.0447,  0.0196, -0.0197, -0.1525, -0.1482, -0.0563,
        -0.0067,  0.0375, -0.0266,  0.0282,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0048, -0.3448, -0.0266, -0.1265, -0.0069, -0.1202, -0.2064,  0.0128,
        -0.0808,  0.0095,  0.0078, -0.0529,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0192,  0.0596, -0.1455, -0.0005, -0.0627, -0.0031,  0.0006,  0.0214,
        -0.1506, -0.0248, -0.0163, -0.1797, -0.0291, -0.0191, -0.0323, -0.1419,
         0.0103,  0.0269, -0.0116, -0.0448,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.8260e-02, -2.0242e-01,  8.8796e-03, -8.1210e-02, -1.1432e-01,
        -2.1825e-02, -2.7346e-02,  7.3326e-03, -4.5473e-02, -1.3343e-02,
         1.2448e-03, -4.3884e-02, -2.1092e-02, -2.2072e-02, -4.6429e-04,
        -8.6571e-03, -2.7055e-02,  8.5179e-03, -1.0383e-02, -1.6379e-03,
        -2.6324e-02,  9.6010e-03,  1.7719e-04, -1.6673e-03, -2.7947e-02,
         1.0769e-02, -1.1891e-02, -4.0800e-02, -3.8079e-02,  7.6231e-03,
         3.6722e-03,  6.9169e-04, -2.5665e-02, -6.1689e-02,  1.2488e-02,
        -1.8347e-02,  7.1467e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0168,  0.6392,  0.1224,  0.1118, -0.0109, -0.0070,  0.0014,  0.0418,
        -0.0488,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0484, -0.0077, -0.0215, -0.1278, -0.0095, -0.0365, -0.1087, -0.0675,
         0.0008, -0.0046, -0.0194, -0.0554, -0.1280, -0.1242, -0.0090, -0.0119,
         0.0024,  0.0115, -0.0354, -0.1006,  0.0149, -0.0037,  0.0075, -0.0064,
         0.0115,  0.0046, -0.0207,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0159, -0.2835,  0.0289, -0.0694, -0.0291, -0.0302,  0.0029,  0.0104,
        -0.0055,  0.0023,  0.0164, -0.0428, -0.0084,  0.0091, -0.0112,  0.0160,
        -0.0621,  0.0023,  0.0034,  0.0015,  0.0019, -0.0479, -0.0886,  0.0118,
        -0.0047,  0.0093, -0.0071, -0.0218, -0.0729, -0.0050,  0.0101, -0.0058,
        -0.0100, -0.0516,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 0.0634, -0.0033, -0.0103, -0.0106, -0.0494, -0.0003,  0.0008,  0.0134,
         0.0142,  0.0266,  0.0074, -0.1160,  0.0219,  0.0019,  0.0110, -0.0185,
        -0.0693,  0.0002, -0.0190,  0.0014,  0.0039,  0.0085, -0.0419, -0.0047,
         0.0042,  0.0073,  0.0028, -0.0158, -0.0041, -0.0126,  0.0098, -0.0611,
        -0.1287,  0.0018, -0.0054, -0.0383, -0.0757,  0.0018,  0.0069, -0.0019,
        -0.0209, -0.0601,  0.0085, -0.0049, -0.0094,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0622, -0.0180,  0.0110, -0.0114, -0.0301, -0.2224, -0.3104, -0.0005,
        -0.0100, -0.1082,  0.0316, -0.0073, -0.0910, -0.0397, -0.0123, -0.0340,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0585,  0.0132, -0.0005,  0.0158,  0.0039,  0.0152,  0.0131, -0.0102,
        -0.1499, -0.1590, -0.0468, -0.1496, -0.0137, -0.0493, -0.1318, -0.0120,
         0.0034, -0.0122,  0.0044, -0.0034, -0.0037,  0.0346,  0.0045, -0.0134,
        -0.0029,  0.0455,  0.0295,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0012, -0.0170, -0.0736, -0.0051, -0.0430,  0.0061, -0.0288, -0.0113,
        -0.0223, -0.0138,  0.0040, -0.0020, -0.0390,  0.0018, -0.0330, -0.0065,
        -0.0255,  0.0089,  0.0025, -0.0329,  0.0052, -0.0327, -0.0253, -0.0621,
        -0.0034, -0.0051, -0.0515, -0.0687, -0.0100, -0.0306,  0.0018, -0.0510,
        -0.0499, -0.0208, -0.0203, -0.0467, -0.0006, -0.0420,  0.0008,  0.0047,
        -0.0286, -0.0495, -0.0005, -0.0009,  0.0066,  0.0023], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0171, -0.0104,  0.0036, -0.0010, -0.0021, -0.0012,  0.0155, -0.0976,
         0.0732, -0.0009, -0.0072,  0.0242,  0.0204, -0.0070,  0.0272, -0.0013,
         0.0017,  0.0030, -0.1132, -0.2652, -0.0539,  0.0210,  0.0061,  0.0046,
         0.0113, -0.0060, -0.0476,  0.0789, -0.0324, -0.0449,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0312, -0.0180, -0.0331, -0.0048,  0.0113,  0.0170, -0.0093,  0.5963,
        -0.1324, -0.0807,  0.0659,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0041, -0.3810, -0.0298, -0.0454, -0.0761,  0.0017, -0.0438, -0.0953,
         0.0057,  0.0064, -0.0076,  0.0037, -0.0355, -0.0257, -0.0165, -0.1027,
        -0.0297, -0.0268, -0.0187,  0.0437,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0748, -0.2884, -0.1433, -0.0748,  0.0351, -0.1038, -0.1398,  0.0251,
         0.0532, -0.0515, -0.0102,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0181, -0.2825, -0.1415, -0.1338,  0.0203, -0.0226, -0.0380, -0.0622,
        -0.0751, -0.0204, -0.0076, -0.0157, -0.0191, -0.0306, -0.0076, -0.0394,
        -0.0029, -0.0014,  0.0085,  0.0028,  0.0500,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.3931e-01, -2.2515e-01, -1.4342e-01, -1.2511e-02,  1.4013e-04,
        -2.2029e-01,  1.2938e-02,  1.2284e-02, -2.4649e-02,  1.8680e-04,
        -1.0342e-02, -4.9082e-02, -1.6421e-03, -1.7387e-03,  3.3223e-03,
        -1.0172e-02, -6.6816e-02,  6.6005e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0223, -0.0949, -0.1756, -0.0037, -0.0231, -0.0062,  0.0085,  0.0055,
         0.0036, -0.0054, -0.0034, -0.0082, -0.0022, -0.0645, -0.0790, -0.0287,
        -0.0586, -0.0598, -0.0235,  0.0110, -0.0081, -0.0492, -0.0699, -0.0006,
        -0.0203,  0.0024, -0.0250, -0.0074, -0.0230, -0.0428,  0.0027, -0.0070,
        -0.0102,  0.0091, -0.0120, -0.0036,  0.0093,  0.0086, -0.0011,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0057, -0.3482, -0.0017, -0.0799, -0.0203, -0.0084, -0.0011,  0.0093,
        -0.0175, -0.0085, -0.0310, -0.1160, -0.0126, -0.0038, -0.0022, -0.0037,
        -0.0685,  0.0036,  0.0064, -0.0043, -0.0096, -0.0118,  0.0060,  0.0048,
        -0.0041,  0.0007, -0.0174, -0.0191, -0.0546, -0.0759, -0.0071, -0.0106,
         0.0139, -0.0118,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.0811, -0.0528, -0.2325, -0.2972, -0.0874, -0.1667, -0.0119, -0.0102,
        -0.0601,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0105, -0.1351, -0.0109, -0.0736, -0.0230, -0.0842,  0.0013, -0.0042,
        -0.0604, -0.0208, -0.0327, -0.0274, -0.0062, -0.0090, -0.0092, -0.0720,
        -0.0238, -0.1007, -0.0108,  0.0014, -0.0077,  0.0045,  0.0007, -0.0039,
        -0.0188, -0.0144,  0.0015, -0.0161,  0.0009, -0.0014, -0.0208, -0.0360,
        -0.0003, -0.0348, -0.0077,  0.0025, -0.0092,  0.0138, -0.0216,  0.0661,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0225, -0.0315, -0.1512,  0.0125, -0.0408, -0.0143,  0.0005,  0.0044,
        -0.0500, -0.0701,  0.0015, -0.0003, -0.0010, -0.0062, -0.0044,  0.0175,
         0.0056,  0.0142, -0.0318,  0.0036, -0.0352, -0.0009, -0.0056,  0.0037,
         0.0078, -0.0191, -0.0442,  0.0046, -0.0628, -0.0733, -0.0056, -0.0003,
         0.0015, -0.0011,  0.0059,  0.0059, -0.0313, -0.0623, -0.0085, -0.0031,
         0.0217,  0.0521,  0.0028,  0.0089, -0.0093,  0.0014, -0.0056,  0.0086,
        -0.0012,  0.0003,  0.0113, -0.0102,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0240, -0.2685, -0.0433, -0.0418, -0.0342, -0.0127, -0.0842, -0.0368,
        -0.0389,  0.0030,  0.0074,  0.0210, -0.0940, -0.0134, -0.0202, -0.0614,
        -0.0295, -0.0192, -0.0694,  0.0061,  0.0287,  0.0425,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0725e-01, -4.7329e-01, -2.3425e-02, -3.0039e-02, -7.9769e-04,
        -1.4223e-02,  2.0637e-02, -6.9821e-02, -1.8771e-02, -3.2520e-02,
        -9.4805e-03, -4.4012e-02,  3.2939e-03, -3.7500e-02,  6.7358e-03,
        -6.1030e-02, -5.6913e-03,  3.7827e-04,  1.7267e-02, -1.0424e-02,
        -1.3410e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.8189e-02, -1.9427e-01, -3.3846e-02, -5.1182e-03, -1.0314e-03,
         5.8594e-03, -1.6887e-02,  1.5325e-02, -2.4312e-02,  6.1525e-05,
         5.6712e-03, -2.9605e-03,  1.1150e-03,  1.0676e-02,  3.3018e-04,
         6.6360e-03,  8.9621e-03,  4.1529e-03, -2.3039e-02,  4.8358e-03,
         1.5540e-03,  6.5991e-03, -5.7836e-04,  7.3574e-03,  4.3654e-03,
         2.6903e-03, -3.9520e-03, -2.5596e-02,  3.5488e-03, -3.2413e-02,
        -4.9000e-02, -5.4281e-02, -2.2694e-03, -3.3483e-02, -3.4859e-02,
        -1.5588e-03,  3.3621e-03,  5.3062e-03,  8.4521e-03, -5.4240e-03,
        -6.2323e-02, -9.3699e-04,  1.1361e-04, -5.4351e-03, -3.5551e-02,
         8.5306e-03, -4.6243e-02, -5.5317e-02, -4.6117e-03,  1.7347e-02,
        -1.7358e-02, -1.1780e-02,  1.7385e-03,  5.5835e-03, -2.7209e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0250,  0.0021,  0.0225, -0.0283, -0.0107, -0.0195, -0.0851, -0.0049,
         0.0075,  0.0044, -0.0022, -0.0046,  0.0162, -0.0349, -0.0657, -0.0152,
        -0.0086, -0.0443, -0.0048, -0.1056,  0.0029, -0.0317, -0.0461,  0.0070,
         0.0107,  0.0009,  0.0026, -0.0682, -0.0154,  0.0021,  0.0085,  0.0015,
        -0.0098, -0.0587, -0.0207, -0.0287,  0.0056, -0.0413, -0.0482, -0.0426,
        -0.0113,  0.0036, -0.0003, -0.0015,  0.0180,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0064, -0.2359, -0.0677, -0.0923,  0.0054, -0.0151, -0.0137,  0.0073,
        -0.0435,  0.0044, -0.0230, -0.0276, -0.0101, -0.0740, -0.0540, -0.0185,
        -0.0220, -0.0608, -0.0269, -0.0435, -0.0890,  0.0218,  0.0170, -0.0199,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0222,  0.0014,  0.0064, -0.0048, -0.0210,  0.0004, -0.0435, -0.0546,
         0.0166,  0.0091, -0.0023,  0.0022, -0.0009, -0.0334,  0.0045,  0.0438,
        -0.0661, -0.0441, -0.0661, -0.0675,  0.0047,  0.0076, -0.0592, -0.0846,
        -0.0379, -0.0680, -0.0930, -0.0115, -0.0352, -0.0620,  0.0071, -0.0070,
        -0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0981, -0.4465, -0.0064, -0.0896, -0.0130, -0.0247, -0.0029, -0.0129,
        -0.0883,  0.0122,  0.0133, -0.0186,  0.0237, -0.0752,  0.0111, -0.0157,
        -0.0066,  0.0042,  0.0373,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0216,  0.0077,  0.0322, -0.0138, -0.1206, -0.0242, -0.0407, -0.1193,
        -0.1846, -0.0343, -0.0084, -0.0007,  0.0017,  0.0187,  0.0012, -0.0030,
        -0.0172, -0.0820, -0.0835, -0.0172, -0.0760, -0.0622, -0.0116, -0.0129,
         0.0046,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0911, -0.1951, -0.1419,  0.0069, -0.0205, -0.0797, -0.2376,  0.0376,
        -0.0085, -0.0212, -0.0097,  0.0159, -0.0341,  0.0043, -0.0076,  0.0524,
         0.0359,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0312,  0.0055, -0.0127,  0.0120,  0.0101, -0.0079, -0.0032,  0.0017,
         0.0084, -0.0657, -0.0209,  0.0133,  0.0200,  0.0129,  0.0057, -0.0143,
        -0.0829, -0.2015,  0.0011, -0.0664, -0.0044, -0.1165, -0.1219, -0.0102,
         0.0013, -0.0014,  0.0367,  0.0263, -0.0114,  0.0017,  0.0033,  0.0676,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0333, -0.0913, -0.1143, -0.0157, -0.0152,  0.0018, -0.0653, -0.1528,
        -0.0091,  0.0210,  0.0041, -0.0055, -0.0059, -0.0021, -0.0321, -0.0066,
        -0.0146, -0.0030, -0.0890, -0.0037, -0.0643,  0.0118, -0.0729, -0.0843,
        -0.0183, -0.0311,  0.0158, -0.0150,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5744e-02, -4.8454e-02, -7.7973e-03, -1.2550e-02, -3.6191e-02,
        -9.6818e-02, -1.4068e-01, -5.2515e-02,  8.2285e-03, -3.5404e-03,
         4.9705e-03, -9.0329e-02, -8.2024e-03, -4.9241e-02, -1.6600e-02,
        -1.0927e-01, -1.4016e-02, -1.2520e-01, -1.6133e-02, -1.0963e-02,
        -1.9307e-02,  8.8927e-03, -1.4792e-02, -7.2039e-03, -2.9289e-02,
         3.0598e-03, -1.3866e-04,  4.1226e-02, -8.6439e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0377, -0.1462, -0.0018, -0.0366, -0.1089, -0.0149,  0.0062,  0.0046,
        -0.0036,  0.0128,  0.0196,  0.0028, -0.0411, -0.1363,  0.0158, -0.0607,
         0.0186, -0.0850,  0.0013, -0.0339, -0.0095, -0.0008, -0.0410, -0.0123,
        -0.0289, -0.0681, -0.0363, -0.0038, -0.0008, -0.0008, -0.0029, -0.0017,
        -0.0045,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.5939e-03, -3.0962e-01, -3.6472e-02, -8.2868e-02, -4.4824e-03,
        -6.4723e-02, -1.4894e-03, -2.0542e-02, -8.0724e-03, -6.4384e-02,
        -7.1747e-02, -5.8587e-03, -8.4016e-03,  7.5598e-03, -7.1964e-03,
         1.2715e-02, -2.7337e-02,  3.1951e-03, -1.2416e-01,  2.3782e-02,
        -8.6925e-02,  2.3319e-04, -5.1741e-03, -1.5466e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0126, -0.3023, -0.1308, -0.0019, -0.0549, -0.0024, -0.0039, -0.0504,
         0.0473, -0.0866, -0.0161, -0.0363, -0.0642, -0.0010,  0.0167,  0.0029,
         0.0166, -0.0307, -0.0010, -0.0777, -0.0333, -0.0100, -0.0006,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.3155e-02, -2.3679e-01, -2.7670e-02, -1.8557e-02, -3.1182e-02,
        -1.5028e-02,  2.6911e-03, -4.7109e-04, -4.6557e-03, -7.0809e-04,
         8.0278e-03, -9.4529e-03, -1.2146e-02, -9.2348e-04, -1.0489e-02,
        -7.5084e-03, -8.1060e-02, -3.4699e-02, -2.4971e-02, -1.5012e-02,
         7.6484e-03, -2.7547e-02, -9.5954e-03, -2.0326e-02, -5.1155e-02,
        -1.3216e-02, -9.6541e-03,  4.5807e-05, -1.2649e-02, -6.4845e-02,
        -7.7761e-02,  5.7515e-03, -1.9044e-02, -5.2366e-03, -3.6929e-02,
        -4.7234e-02, -1.4127e-03, -3.0852e-03,  1.1670e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0223, -0.0246, -0.0135, -0.0318, -0.1030, -0.0042,  0.0049,  0.0103,
        -0.0100,  0.0228, -0.0067, -0.0012,  0.0028, -0.0011,  0.0121, -0.0324,
        -0.0130, -0.0026,  0.0105,  0.0089, -0.0455, -0.0080, -0.0044, -0.0407,
        -0.1419, -0.1169, -0.0065, -0.0442, -0.0112, -0.0682, -0.0043,  0.0069,
        -0.0192, -0.0017,  0.0006, -0.0150, -0.0077, -0.0626,  0.0040,  0.0020,
         0.0057, -0.0053,  0.0390,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.7619e-02, -1.8000e-01,  7.8531e-03,  4.0485e-03,  3.4304e-02,
        -3.2692e-03, -2.4138e-02, -8.8263e-03, -3.9144e-02, -1.0042e-02,
         1.8051e-02, -2.5225e-03, -1.7285e-02,  1.9283e-03, -1.2376e-02,
         6.0547e-03, -2.5567e-02, -9.5969e-02, -6.1814e-03, -1.0747e-02,
         6.4959e-03,  4.9988e-03,  2.6475e-03, -7.8327e-03,  9.5747e-03,
        -3.4022e-03,  9.3967e-03, -2.2260e-04,  1.0436e-03,  4.2314e-04,
        -3.7839e-03, -4.7480e-02, -7.3044e-02, -2.0238e-02, -1.6779e-02,
        -3.7036e-03, -5.5650e-03, -3.2501e-02, -3.2278e-02, -1.3452e-04,
        -3.4260e-02, -4.7153e-02, -1.1288e-03, -3.4866e-04, -1.5751e-03,
        -8.4926e-03, -4.4075e-03,  2.1488e-03, -2.4269e-03, -1.6691e-03,
         7.0949e-03, -9.9052e-03,  5.1555e-03, -2.6910e-03,  9.7148e-03,
        -2.4063e-02, -2.0291e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0749,  0.4178,  0.0107, -0.0012,  0.0595, -0.0028,  0.0736,  0.1463,
        -0.0203,  0.0510,  0.0699,  0.0017, -0.0334, -0.0370,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.4950e-02, -1.9359e-01, -4.6344e-02,  1.2678e-03, -7.6124e-03,
        -4.2404e-02, -2.9807e-03,  1.3662e-02, -4.4256e-02, -1.2827e-02,
        -4.8673e-02, -1.0478e-01, -2.1311e-02,  9.7094e-03, -4.5480e-02,
        -1.4002e-03, -3.0549e-02, -3.5121e-02,  1.0460e-02, -5.2624e-02,
        -8.6050e-02, -2.3330e-03, -2.8637e-02,  1.6034e-04,  2.2938e-02,
         1.9747e-02,  4.1397e-03,  6.6000e-03, -1.4261e-02,  3.5131e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1325, -0.2818, -0.0375, -0.2450, -0.1018,  0.0352, -0.0800,  0.0862,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0519, -0.1772, -0.0415, -0.0327, -0.0472, -0.1124,  0.0002,  0.0024,
        -0.0032, -0.0189, -0.0148,  0.0064, -0.0654, -0.0981, -0.0145,  0.0021,
        -0.0045, -0.0062,  0.0002,  0.0067,  0.0019,  0.0002,  0.0070, -0.0097,
        -0.0445,  0.0123, -0.0354, -0.0210, -0.0438, -0.0641,  0.0045,  0.0162,
        -0.0331,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0297, -0.1980, -0.1177, -0.0418, -0.0049, -0.0295, -0.0119, -0.0027,
         0.0169, -0.0227,  0.0023, -0.0117, -0.0150,  0.0009,  0.0138,  0.0032,
         0.0032,  0.0070, -0.0079,  0.0099,  0.0014,  0.0084, -0.0026,  0.0010,
         0.0046, -0.0033, -0.0062, -0.0560, -0.0021,  0.0033,  0.0017, -0.0042,
        -0.0214, -0.0435,  0.0017,  0.0040, -0.0105, -0.0060, -0.0079,  0.0029,
         0.0005, -0.0041, -0.0030, -0.0347, -0.0007,  0.0060, -0.0025,  0.0004,
         0.0013,  0.0005, -0.0038, -0.0445, -0.0104, -0.0138,  0.0003,  0.0089,
        -0.0033,  0.0049,  0.0010, -0.0265, -0.0059, -0.0067, -0.0302,  0.0011,
         0.0043, -0.0033, -0.0035,  0.0154,  0.0122,  0.0016,  0.0014],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7010e-02, -2.3748e-02, -3.0574e-02, -3.4601e-03, -3.3438e-02,
        -9.2570e-03, -9.8936e-02, -3.1833e-06,  1.3897e-03,  1.3778e-02,
         7.1252e-03,  1.7734e-03,  4.7457e-03,  6.8332e-04, -3.2830e-03,
         1.4756e-02,  1.2932e-03,  1.7689e-03, -6.5429e-02, -9.7433e-02,
         1.7931e-03,  4.5451e-03,  6.2445e-04, -2.6828e-02,  2.7098e-02,
        -6.0575e-02, -5.2652e-03,  3.9972e-03,  1.2593e-03, -1.0761e-03,
        -2.2002e-02, -2.9725e-02, -4.4851e-02, -1.4836e-01, -1.5219e-02,
        -1.7028e-02, -8.2518e-02, -3.4798e-02, -2.9052e-03,  5.1555e-03,
         3.2309e-02, -2.1848e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0083,  0.1868,  0.0031,  0.0013,  0.0099, -0.0018, -0.0040, -0.0149,
         0.0318,  0.0581,  0.0188,  0.0420,  0.0667, -0.0008,  0.0049, -0.0011,
         0.0172,  0.0412,  0.0631, -0.0091,  0.0115,  0.0236,  0.0083, -0.0043,
         0.0117,  0.0934,  0.0300,  0.0041,  0.1180, -0.0111,  0.0021,  0.0294,
         0.0077,  0.0012,  0.0040,  0.0410,  0.0134,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0993, -0.5238,  0.0316, -0.0296, -0.0078,  0.0378, -0.0237, -0.0050,
        -0.0401, -0.0169, -0.0276,  0.0356, -0.0805, -0.0322,  0.0086,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.2035e-02, -4.2929e-02,  1.4322e-02, -4.8601e-03, -1.2264e-01,
        -2.0389e-01, -6.5806e-02, -7.8272e-02, -1.9066e-01, -1.6747e-02,
         8.6120e-04,  4.6011e-02, -1.9620e-02,  7.6548e-05,  4.1988e-02,
        -6.4487e-03,  1.2021e-02, -9.7452e-03,  8.9027e-03,  7.2163e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0043, -0.0015, -0.0009,  0.0027, -0.1405,  0.0038,  0.0086, -0.0091,
        -0.0020,  0.0089, -0.0478, -0.0792,  0.0136, -0.0023, -0.0278, -0.0196,
        -0.0028, -0.0312,  0.0095, -0.0234, -0.0016,  0.0021,  0.0017,  0.0008,
         0.0051, -0.0533, -0.0970,  0.0043,  0.0099, -0.0878, -0.0847, -0.0075,
         0.0033, -0.0041, -0.0400, -0.0780,  0.0103, -0.0281,  0.0025, -0.0021,
         0.0003,  0.0177,  0.0183,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.3461e-02, -5.6188e-03, -1.1339e-02, -1.0385e-02, -5.7982e-02,
        -2.0893e-02, -6.1511e-02, -7.6970e-02, -1.8523e-02,  1.3632e-02,
         1.9547e-02,  9.4829e-05, -9.2985e-03, -7.6198e-02, -1.0879e-01,
         6.5806e-03,  9.6332e-03, -1.1320e-03, -1.5292e-03, -3.5919e-03,
        -5.7400e-02, -3.0215e-03,  9.1544e-03,  6.2108e-03,  6.3422e-03,
        -1.8045e-02,  1.4682e-02,  6.1735e-03, -6.6899e-02,  2.7425e-03,
        -1.2010e-02, -4.1602e-02, -4.1364e-03, -4.6963e-02, -6.6907e-02,
        -2.8019e-02,  6.2479e-02, -1.0504e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0032, -0.2526, -0.1165, -0.0998, -0.0048, -0.0955,  0.0045,  0.0093,
        -0.0301, -0.0794, -0.0224, -0.0162, -0.0697, -0.0208, -0.0095, -0.0419,
         0.0036, -0.0657,  0.0006,  0.0220,  0.0115,  0.0207,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0141, -0.1084,  0.0135, -0.0005, -0.0018, -0.0120, -0.0034,  0.0051,
         0.0030, -0.1081, -0.0182,  0.0041, -0.0049,  0.0088, -0.0600, -0.0456,
        -0.0997, -0.0106,  0.0031, -0.0607,  0.0023,  0.0079, -0.0364, -0.0001,
        -0.0175, -0.0076, -0.0028, -0.0028, -0.0053, -0.0558, -0.0028, -0.0312,
         0.0109, -0.0128, -0.0239, -0.0970,  0.0285, -0.0069, -0.0142, -0.0477,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0334, -0.0225, -0.0127, -0.0190,  0.0016, -0.0186, -0.0269, -0.0196,
        -0.0858, -0.0160, -0.0164, -0.0005, -0.0205, -0.1781, -0.0802, -0.0329,
         0.0124,  0.0113, -0.0493,  0.0122, -0.0143,  0.0218,  0.0054,  0.0074,
        -0.0133, -0.0789, -0.0042, -0.0591, -0.1115,  0.0007, -0.0102, -0.0031,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.0481e-02, -3.0828e-01, -1.5783e-04, -1.4653e-02, -1.0818e-02,
        -3.6048e-03, -1.6277e-02, -8.0437e-02, -1.2891e-01, -2.3573e-05,
        -1.7429e-02, -3.7347e-02,  7.0218e-03,  4.7025e-04,  8.2402e-03,
         1.1854e-02, -5.7044e-02, -9.3665e-03,  2.3738e-03, -3.0331e-03,
        -1.1201e-02, -1.5292e-02, -1.7941e-02, -8.0567e-02,  1.5470e-02,
         9.2028e-03, -1.9437e-02, -2.2626e-03, -8.3299e-04, -1.6829e-02,
        -6.0502e-03,  5.6230e-03, -1.3168e-03,  2.0486e-02, -2.9663e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 4.4564e-02, -3.7186e-02, -1.8096e-01, -1.3775e-01, -4.7225e-05,
        -1.1124e-02, -1.9860e-01, -1.3307e-01, -1.2317e-02, -7.4403e-02,
         1.2517e-02, -1.0521e-02,  1.0499e-02, -4.0817e-02, -8.1782e-03,
        -1.4044e-02, -2.3599e-02, -1.1720e-02,  4.7162e-03,  2.4311e-02,
         9.0635e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0396, -0.3266,  0.0320, -0.0793, -0.1109, -0.0537, -0.0233, -0.0866,
        -0.0748,  0.0219, -0.0075,  0.0212,  0.0089, -0.0110, -0.0436, -0.0018,
        -0.0318, -0.0229,  0.0026,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1055, -0.0478, -0.0847,  0.0083, -0.0640, -0.0109,  0.0010, -0.0164,
        -0.0136, -0.0687, -0.2031,  0.0160, -0.0532, -0.1888, -0.0063, -0.0625,
        -0.0007, -0.0252, -0.0093,  0.0140,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0386, -0.3597,  0.0166, -0.0802,  0.0061,  0.0043, -0.0275, -0.0344,
        -0.0055, -0.0034, -0.0133, -0.0134,  0.0179, -0.0526, -0.0180, -0.0423,
        -0.0934, -0.0156, -0.0273, -0.0424,  0.0337,  0.0212,  0.0326,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0591, -0.1597, -0.0573, -0.0245, -0.0292, -0.1094, -0.0335, -0.0158,
        -0.0557, -0.0727, -0.0035, -0.0788, -0.0072, -0.1029,  0.0061,  0.0038,
        -0.0016, -0.0054, -0.0321,  0.0107, -0.0674, -0.0044, -0.0005,  0.0304,
        -0.0284,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1343, -0.2641, -0.0294,  0.0006,  0.0046,  0.0057,  0.0168,  0.0044,
         0.0064,  0.0065, -0.0726, -0.0156, -0.0045,  0.0014,  0.0063, -0.0093,
        -0.0129,  0.0072,  0.0147, -0.0057, -0.1076, -0.0757, -0.0465,  0.0009,
        -0.0051, -0.0359, -0.0120, -0.0025, -0.0171, -0.0135, -0.0167,  0.0070,
        -0.0200,  0.0120,  0.0048,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0163,  0.0071,  0.0037, -0.1131, -0.0007, -0.0146, -0.1347, -0.0247,
        -0.1345, -0.0343, -0.0308, -0.0345,  0.0261, -0.0355, -0.0338, -0.1347,
        -0.0135, -0.1128,  0.0071,  0.0163, -0.0248,  0.0463,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0310, -0.0251,  0.0420, -0.0120, -0.0255, -0.0274, -0.0109, -0.0201,
        -0.0175, -0.4436,  0.0640,  0.0089,  0.0360, -0.0041,  0.0091, -0.1173,
        -0.0070,  0.0231,  0.0519,  0.0094, -0.0141,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0506, -0.4006, -0.0266, -0.0094, -0.0221,  0.0163, -0.1022, -0.1184,
         0.0262,  0.0114, -0.0076,  0.0187,  0.0075, -0.0787, -0.0212, -0.0490,
         0.0145, -0.0053, -0.0135,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0015, -0.0297, -0.0713,  0.0011, -0.0028, -0.0669, -0.0030, -0.0068,
         0.0040, -0.0053, -0.0122, -0.0141,  0.0008, -0.0339,  0.0086, -0.0017,
        -0.0660, -0.0498,  0.0087, -0.0519, -0.0144, -0.0123, -0.0311,  0.0156,
        -0.0492, -0.0051, -0.0066, -0.0305,  0.0003, -0.1084, -0.0134, -0.0170,
        -0.0446, -0.0932,  0.0037, -0.0137, -0.0065, -0.0015, -0.0136, -0.0323,
         0.0037,  0.0007, -0.0011, -0.0049, -0.0018, -0.0057, -0.0288],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0102, -0.2323, -0.1904,  0.0055,  0.0055,  0.0156, -0.0079,  0.0011,
        -0.0384,  0.0045,  0.0004,  0.0041,  0.0197,  0.0195,  0.0178,  0.0017,
        -0.0934, -0.0042,  0.0139,  0.0013,  0.0106, -0.0129, -0.1129, -0.0931,
         0.0033, -0.0103, -0.0139, -0.0027,  0.0099,  0.0035,  0.0120,  0.0091,
        -0.0160, -0.0022,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0457, -0.3188, -0.0231,  0.0077, -0.0054, -0.0410, -0.0101,  0.0131,
        -0.0073, -0.0301, -0.0019, -0.0112, -0.0626, -0.0508,  0.0084, -0.0064,
        -0.0075, -0.0425,  0.0256, -0.0548, -0.0040, -0.0174, -0.0114,  0.0053,
        -0.0120,  0.0064,  0.0037, -0.0190,  0.0023, -0.0311,  0.0097, -0.0091,
        -0.0824,  0.0122,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0463, -0.4080,  0.0077,  0.0251, -0.0100, -0.0913,  0.0242, -0.0078,
         0.0271, -0.0961, -0.0360,  0.0079, -0.0292, -0.1262, -0.0069,  0.0027,
         0.0178,  0.0299,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0047, -0.0086, -0.0025,  0.0186,  0.0008, -0.0085,  0.0092, -0.0185,
        -0.1816, -0.3533,  0.0508,  0.0134,  0.0009, -0.0101,  0.0065, -0.0029,
        -0.0406,  0.0017, -0.0231, -0.1403,  0.0011,  0.0028, -0.0323, -0.0087,
        -0.0084, -0.0364, -0.0137,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0501,  0.0102,  0.0149, -0.0094, -0.0306,  0.0087,  0.0175, -0.0313,
        -0.1762, -0.0075, -0.0047, -0.0525, -0.1866, -0.0046,  0.0065, -0.0051,
        -0.0097,  0.0052,  0.0038, -0.2571,  0.0032,  0.0142, -0.0134,  0.0127,
         0.0268, -0.0373,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0164, -0.2449, -0.0092, -0.0885, -0.1380,  0.0125, -0.0100, -0.0026,
        -0.0529, -0.0250, -0.0767, -0.0390,  0.0064,  0.0048, -0.0294, -0.0732,
        -0.0793, -0.0489, -0.0061,  0.0177,  0.0185,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0482, -0.2925, -0.1711, -0.2503, -0.0267,  0.0170, -0.0580, -0.0071,
        -0.0231, -0.0045,  0.0080, -0.0601, -0.0050,  0.0034, -0.0251,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.5068e-02, -3.0857e-04, -5.1478e-02,  7.0087e-03, -3.9150e-02,
        -2.0980e-03,  5.9269e-02, -9.9034e-03, -2.7438e-03,  2.1220e-02,
        -1.1055e-02, -4.7361e-02, -4.3433e-01, -7.5956e-02, -5.7628e-02,
         4.4400e-02,  2.9370e-02, -3.1655e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0167, -0.3148, -0.0937, -0.1114,  0.0248,  0.0258, -0.0082, -0.0745,
        -0.0230, -0.0281, -0.0672, -0.0262,  0.0170, -0.0378, -0.0073,  0.0111,
         0.0066,  0.0467,  0.0592,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.4707e-02, -1.9436e-01, -1.0493e-01, -4.3882e-03, -2.2407e-02,
        -2.0946e-02,  8.7124e-03,  1.0995e-02, -9.7758e-03, -1.5371e-04,
         3.5686e-03, -1.3992e-02, -1.4728e-02,  6.4713e-03, -4.2451e-03,
        -3.5781e-04, -3.9223e-02,  5.1198e-04,  8.4478e-03, -3.9072e-03,
        -5.7710e-02,  1.1072e-02, -4.8273e-02, -1.0275e-02, -6.9145e-02,
        -1.1282e-02, -3.1371e-02,  2.4031e-02,  2.4812e-03, -1.7061e-02,
        -9.5672e-04,  1.8978e-02, -2.9081e-02, -6.9397e-04, -2.2646e-02,
        -1.9729e-02, -5.0476e-02, -5.8612e-03, -4.4772e-03, -1.8125e-02,
        -4.7805e-03, -7.0387e-03,  2.6495e-02, -2.1127e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0169, -0.1748, -0.0171, -0.0908, -0.0477, -0.0045, -0.0439, -0.0116,
        -0.1006,  0.0022, -0.0257,  0.0204, -0.0110,  0.0033,  0.0057,  0.0082,
         0.0566, -0.0679, -0.0100, -0.0419, -0.0353,  0.0002, -0.0027, -0.0044,
        -0.0028,  0.0056,  0.0021, -0.0819, -0.0038,  0.0113,  0.0006, -0.0232,
         0.0082,  0.0241,  0.0331,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.3049e-02, -2.8798e-01,  7.3416e-03, -9.5510e-02, -2.7013e-02,
        -9.4530e-03, -1.0780e-01, -6.8684e-02,  1.8241e-02,  2.5337e-04,
        -1.4816e-02, -4.1761e-02, -1.3199e-01,  2.3255e-03,  2.0934e-03,
        -6.0139e-02, -1.7994e-02, -2.2305e-03, -1.4517e-02, -7.9475e-05,
         1.5931e-02, -1.4540e-02,  1.3003e-03,  3.4958e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0189, -0.2094, -0.0064, -0.0925, -0.0064, -0.0429, -0.0134, -0.0255,
        -0.0947,  0.0093, -0.0190, -0.0029, -0.0023, -0.0231, -0.0972, -0.0105,
         0.0038,  0.0041, -0.0199, -0.0091, -0.0056, -0.0400,  0.0193, -0.0313,
        -0.0032, -0.0434, -0.0416, -0.0199, -0.0095, -0.0107, -0.0324, -0.0073,
        -0.0140, -0.0104,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.9368e-02,  9.4968e-03,  9.4047e-05, -4.7180e-02, -4.7400e-04,
         1.6419e-02, -1.1620e-01, -1.1712e-02,  2.8637e-02,  2.1123e-02,
         1.0879e-02,  2.4249e-02, -8.2249e-02, -2.7469e-02, -1.3627e-01,
        -1.8358e-01, -1.5677e-03, -2.5101e-02, -8.9075e-03, -3.7432e-02,
         2.9951e-03, -1.2230e-03, -6.8759e-03, -6.6683e-02, -3.8372e-02,
        -3.9238e-03, -4.2463e-02,  1.3410e-03, -1.1318e-03,  6.5866e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0149,  0.0115, -0.0589, -0.3815, -0.0025, -0.0005, -0.0263, -0.0367,
        -0.1534, -0.1661,  0.0520,  0.0385,  0.0571,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0597,  0.0104,  0.0569, -0.0312, -0.0283, -0.0606, -0.0097, -0.1174,
        -0.2011,  0.0207,  0.0044, -0.0933, -0.0253, -0.0473, -0.1308,  0.0129,
         0.0304,  0.0596,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0356, -0.1786, -0.4773, -0.0056, -0.2179, -0.0279,  0.0069, -0.0502,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0210, -0.2087,  0.0014, -0.0347, -0.0297, -0.0014, -0.0122,  0.0016,
         0.0035, -0.0082,  0.0187,  0.0144, -0.0161, -0.0017, -0.0004, -0.0164,
         0.0013,  0.0023,  0.0040, -0.0020,  0.0116,  0.0042, -0.0284, -0.0266,
         0.0069, -0.0224, -0.0043, -0.0089, -0.0959, -0.0154,  0.0160, -0.0484,
        -0.0072, -0.1066, -0.0106, -0.0379, -0.0413,  0.0028,  0.0132,  0.0060,
        -0.0182, -0.0032, -0.0208,  0.0080,  0.0280, -0.0023, -0.0056],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1119, -0.2464,  0.0814, -0.3894, -0.0141,  0.0247, -0.0175, -0.0120,
        -0.0161,  0.0594, -0.0272,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0204, -0.2488, -0.0371, -0.1174, -0.0331, -0.0052, -0.0660, -0.2326,
        -0.0372, -0.0609, -0.0264,  0.0377, -0.0569,  0.0017,  0.0126, -0.0003,
         0.0047, -0.0008,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.8773e-02, -1.4423e-01, -2.8634e-01, -7.1086e-02, -3.5137e-02,
        -2.2836e-02, -1.8763e-01, -6.5421e-03, -7.6760e-02, -1.6073e-02,
        -1.1914e-02, -5.1485e-03,  1.1818e-04,  1.7239e-02,  4.0172e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0249, -0.3208,  0.0097, -0.0199, -0.0783, -0.1032, -0.1189, -0.0572,
        -0.0340, -0.0762, -0.0785,  0.0055, -0.0004, -0.0364,  0.0360,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0446,  0.0074,  0.0147, -0.0288,  0.0037,  0.0498,  0.0443,  0.0410,
        -0.1082, -0.0549, -0.0414, -0.1189, -0.1200,  0.0072,  0.0015, -0.0076,
        -0.0494,  0.0037, -0.0022, -0.0025, -0.0622, -0.1172, -0.0021, -0.0118,
         0.0428,  0.0120,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0184, -0.0356, -0.0264,  0.0195,  0.0144,  0.0192,  0.1151, -0.1988,
        -0.4110, -0.0392,  0.0182,  0.0026, -0.0025,  0.0722, -0.0069,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.2610e-02, -2.7073e-01,  1.2182e-04, -7.3399e-02, -8.0949e-02,
        -7.1311e-03, -5.3086e-02, -2.0198e-02, -5.4212e-04, -1.3431e-01,
         2.0019e-02,  6.4668e-03, -3.2770e-02, -9.1500e-03,  4.1731e-03,
         3.2166e-03, -1.3151e-02,  3.2009e-03, -5.6850e-03,  1.7246e-02,
        -1.3484e-02, -1.5772e-03, -1.7412e-02, -4.6503e-03,  6.9403e-02,
         5.5320e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0008,  0.0094,  0.0195,  0.0050,  0.0105, -0.0190,  0.0031,  0.0213,
        -0.0316, -0.0492, -0.1542, -0.0138, -0.0096,  0.0497, -0.1406, -0.0796,
         0.0068, -0.0461, -0.0271, -0.1035, -0.0170, -0.0435, -0.0222, -0.0561,
        -0.0149, -0.0457,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-7.1265e-02, -3.8494e-01,  9.1866e-03,  7.9370e-03,  1.0683e-03,
         1.4905e-02, -2.4805e-02, -3.9494e-02, -7.2784e-03, -2.0458e-02,
        -1.2520e-02, -2.9995e-02, -4.6400e-02,  8.9259e-03, -2.8290e-02,
        -8.1388e-03, -2.6349e-03, -1.6341e-03, -9.8265e-03, -6.2759e-03,
         2.2196e-02, -1.7739e-02,  5.9624e-03, -1.0322e-02, -7.7254e-02,
         3.7833e-03, -1.6689e-02, -6.4180e-03,  3.5376e-02,  4.4289e-03,
         1.8634e-04, -1.8879e-02, -6.6282e-04, -7.8634e-03,  1.3961e-02,
        -2.2297e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0255, -0.1182, -0.0531,  0.0221, -0.0616,  0.0046, -0.0522, -0.0681,
        -0.0019, -0.0102, -0.0307, -0.0080, -0.0147, -0.0017, -0.0098, -0.0234,
         0.0141,  0.0007,  0.0077,  0.0011, -0.0025, -0.0666,  0.0013, -0.0014,
        -0.0486, -0.0578, -0.0045, -0.0535, -0.0393, -0.0015, -0.0165, -0.0363,
        -0.0532,  0.0043,  0.0653,  0.0179,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.2613e-02,  4.0100e-02,  6.6212e-03, -2.0871e-02, -5.0029e-04,
        -1.7611e-01,  2.6343e-04, -2.7705e-02,  1.8340e-02, -5.0807e-03,
         2.3331e-02, -2.1558e-01, -2.8731e-01, -2.5158e-02,  2.4512e-02,
         2.5630e-02,  3.1783e-02,  1.7361e-02, -3.0717e-02, -4.1662e-04,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0124, -0.1873, -0.0497, -0.0341,  0.0013, -0.0167,  0.0039, -0.0465,
        -0.0094, -0.0326,  0.0055, -0.0136, -0.0026,  0.0043, -0.0932, -0.0652,
        -0.0215, -0.0317,  0.0069, -0.0071, -0.0199, -0.0015,  0.0072, -0.0028,
         0.0033,  0.0050, -0.0031, -0.0146, -0.0064, -0.0546, -0.0582, -0.0153,
        -0.0154, -0.0937, -0.0050,  0.0100, -0.0127, -0.0257,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.1896e-02, -2.3850e-01, -1.1976e-01, -6.2905e-02,  1.7778e-04,
         3.0203e-04, -4.1844e-03,  2.2035e-03,  1.9826e-03, -9.4888e-03,
        -5.0473e-02, -5.6837e-02, -8.0694e-03,  3.7561e-03, -3.4696e-03,
        -5.5680e-02, -2.3885e-03, -7.3867e-02, -6.6431e-03,  6.7818e-03,
         5.2008e-03,  9.4637e-03,  1.6310e-03,  3.3839e-03, -2.9783e-03,
        -1.4693e-03, -2.0883e-02, -2.7600e-02, -6.9296e-02, -2.8646e-03,
        -5.8699e-03, -1.6213e-02, -6.3863e-03,  1.3568e-03, -2.8110e-02,
         4.8111e-04, -5.3301e-03,  9.1619e-03, -5.4830e-03,  1.5562e-02,
         1.1901e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0835, -0.3872,  0.0130,  0.0098, -0.0155, -0.0489,  0.0006,  0.0430,
         0.0012, -0.0049, -0.0090, -0.0492, -0.0857, -0.0104,  0.0027, -0.0599,
        -0.0136, -0.0311, -0.0134,  0.0131,  0.0754, -0.0290,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0488,  0.0098,  0.0031,  0.0021,  0.0188, -0.0280,  0.0023,  0.0100,
        -0.0202,  0.0026, -0.0064, -0.0891, -0.0067, -0.0355,  0.0056, -0.0913,
         0.0063, -0.0467, -0.0157, -0.0788, -0.1087,  0.0083,  0.0150, -0.0415,
        -0.0031,  0.0032, -0.0077, -0.0394, -0.1041, -0.0086, -0.0372, -0.0567,
        -0.0025,  0.0128, -0.0235,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0231, -0.0621, -0.2810, -0.0642, -0.0774, -0.0141,  0.0070,  0.0090,
         0.0118,  0.0032, -0.0237,  0.0098, -0.0115,  0.0161, -0.0009, -0.0097,
         0.0283, -0.0193, -0.0472, -0.0132, -0.0439, -0.0005, -0.0513, -0.0620,
        -0.0121,  0.0043, -0.0383, -0.0333,  0.0217,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0123, -0.3314, -0.0917, -0.0526, -0.0080, -0.0690,  0.0247, -0.0227,
        -0.0374,  0.0084, -0.0416, -0.0427,  0.0108, -0.0162, -0.0080, -0.0707,
        -0.0363,  0.0004, -0.0250,  0.0114, -0.0095, -0.0155, -0.0148, -0.0135,
         0.0253,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.2526e-02, -3.7808e-03, -2.2293e-03, -5.8451e-04, -1.4343e-01,
        -2.2416e-03,  2.9655e-02, -1.2173e-01, -1.6704e-01, -8.7854e-03,
        -3.3979e-03,  3.9044e-02,  1.2408e-02,  3.1406e-02, -1.4988e-01,
        -1.6035e-01,  2.2453e-02, -5.2869e-02, -6.7326e-05, -3.9085e-03,
         6.4281e-03, -1.3212e-02,  2.5779e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.6529e-05,  8.6246e-03,  1.6570e-02,  1.5212e-02,  7.2763e-03,
         9.7079e-03,  5.3530e-03,  9.3485e-03, -8.5572e-04, -4.7246e-03,
         2.5228e-03,  5.3220e-03, -1.1515e-01, -1.3056e-01, -3.4944e-02,
        -1.1476e-02, -7.8872e-03, -4.5605e-02, -1.3551e-02,  1.3082e-02,
         1.5527e-04,  1.3387e-02,  5.3073e-03,  8.3463e-03,  1.6570e-02,
         1.4025e-02, -1.5146e-01, -1.0303e-02, -6.8432e-02,  7.7783e-03,
        -9.2692e-02, -1.1831e-02, -9.7269e-03,  3.1353e-02, -6.1116e-02,
        -2.6960e-04, -9.3881e-03, -2.0842e-03, -2.7929e-02,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0739,  0.0197,  0.0094, -0.1807, -0.3190, -0.0045,  0.0007, -0.1852,
        -0.1570, -0.0296, -0.0142,  0.0063,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-3.7439e-02, -1.0070e-01, -1.7867e-01,  3.2000e-03, -6.6747e-03,
         7.8132e-03, -3.7460e-02, -8.6554e-03, -1.2459e-02, -5.2124e-02,
         2.2790e-03,  6.7059e-03, -1.6345e-02, -5.8611e-03, -5.9651e-02,
        -8.3654e-03, -1.2770e-02, -7.9018e-02, -1.3133e-02,  2.6217e-03,
        -2.3681e-02, -2.7723e-02,  6.3072e-03, -3.8422e-05, -4.5217e-03,
         9.7477e-03,  9.4432e-03, -2.7271e-02,  4.7973e-02, -2.7900e-02,
         2.2590e-04,  6.5781e-03, -4.3114e-02, -7.3525e-03, -1.1066e-03,
        -1.1568e-03, -4.5660e-02, -4.6475e-03, -3.2955e-05, -3.4111e-03,
         3.3542e-03,  1.5850e-03,  1.0250e-02, -2.3804e-02,  1.1162e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0313, -0.4807,  0.0046, -0.1716, -0.0155,  0.0419,  0.0228, -0.0508,
        -0.1321,  0.0281, -0.0127,  0.0078,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0359,  0.4504,  0.1323,  0.1350, -0.0313, -0.0493,  0.0851, -0.0807,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.6825e-02, -2.1012e-01, -1.4431e-02, -1.0564e-03, -5.4869e-02,
        -2.5150e-03, -4.3803e-03, -1.2120e-02, -3.9462e-02, -7.7295e-02,
        -1.2825e-02, -6.6116e-03, -7.8425e-03,  9.8941e-04,  7.0438e-03,
         2.0691e-03,  8.1481e-03, -3.7430e-03,  2.3217e-03, -3.9334e-03,
         4.6911e-03,  5.1158e-03, -4.7781e-03, -6.1760e-03, -7.1704e-03,
         4.8640e-03,  1.5577e-03, -1.4496e-02, -1.0477e-02,  1.6558e-03,
         2.8021e-05, -9.1760e-03,  2.2885e-03,  4.9024e-03, -3.5352e-03,
         5.5550e-03,  2.0119e-03, -8.0614e-03,  6.8638e-03,  6.4355e-04,
        -3.6611e-03,  5.3990e-03,  1.1810e-03, -1.8655e-03, -1.5930e-02,
         1.8679e-02, -6.8561e-02, -6.2534e-03, -4.3076e-02, -1.0266e-01,
        -3.2109e-02,  1.2348e-02, -4.4304e-02, -2.4429e-04, -9.7339e-03,
        -1.8598e-02, -9.8185e-03, -2.4366e-03,  5.2592e-03, -2.6584e-03,
        -2.5794e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0254, -0.0167, -0.0071, -0.0595, -0.0742, -0.0241, -0.0443, -0.0294,
         0.0151,  0.0008, -0.0088, -0.0711, -0.0015, -0.0667, -0.0483, -0.0069,
        -0.0140,  0.0082, -0.0241, -0.0199, -0.0725, -0.1069, -0.0299, -0.0148,
        -0.0486, -0.0383, -0.0199, -0.0182, -0.0260, -0.0236,  0.0007, -0.0314,
         0.0030,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0661, -0.0419, -0.1418, -0.1428, -0.0037,  0.0081, -0.0522, -0.0067,
         0.0006, -0.0130, -0.0051,  0.0158, -0.0449, -0.0054, -0.0242,  0.0781,
         0.0117, -0.0603, -0.1589, -0.0012, -0.0126, -0.0365,  0.0469,  0.0215,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0185, -0.1384, -0.1302, -0.0389, -0.0194, -0.1030,  0.0105, -0.0704,
         0.0038, -0.0614, -0.0735, -0.0745, -0.1066,  0.0094, -0.0132,  0.0094,
        -0.0023,  0.0106,  0.0113, -0.0011, -0.0521, -0.0047, -0.0008, -0.0359,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0410, -0.0136, -0.0540, -0.1214, -0.0279,  0.0106, -0.0087, -0.0657,
         0.0106,  0.0100, -0.0046, -0.0072,  0.0118,  0.0156, -0.0487, -0.0658,
        -0.0802,  0.0087,  0.0049,  0.0085, -0.0924,  0.0019, -0.0393, -0.0972,
         0.0018,  0.0119, -0.0339, -0.0028, -0.0097, -0.0170, -0.0097, -0.0094,
         0.0421, -0.0112,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0057,  0.0109, -0.0102, -0.0051, -0.1301,  0.0092, -0.0117,  0.0066,
        -0.0468, -0.0142,  0.0259,  0.0175, -0.0203, -0.1928, -0.0093, -0.0035,
        -0.1706,  0.0006, -0.0706, -0.0171, -0.0312, -0.1190,  0.0331, -0.0236,
        -0.0143,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0742, -0.0090,  0.0304,  0.0122,  0.0151, -0.0080,  0.1203,  0.1045,
         0.1188,  0.3359,  0.0208,  0.0605,  0.0279,  0.0382, -0.0166, -0.0075,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0063, -0.2517, -0.0093, -0.0994, -0.0180, -0.0821, -0.2122, -0.0256,
         0.0214,  0.0051,  0.0073,  0.0072, -0.0449,  0.0077,  0.0074, -0.0106,
        -0.0197, -0.0788, -0.0459, -0.0150, -0.0028, -0.0135,  0.0081,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0365, -0.0175, -0.0980, -0.1656, -0.3870, -0.0555,  0.0795,  0.0165,
        -0.0027,  0.0407,  0.0309,  0.0119,  0.0417, -0.0162,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-5.3330e-03, -9.6743e-02, -1.4953e-01, -1.8012e-01,  1.8339e-05,
         7.9661e-03, -4.0552e-02,  1.0209e-02, -2.7278e-02, -2.0361e-02,
        -8.6178e-02, -8.1003e-02,  1.0384e-02,  6.7896e-04, -1.2630e-02,
         4.2175e-02, -1.3729e-02, -9.5275e-02, -6.8657e-03, -6.4060e-02,
        -9.3397e-03, -1.2480e-02,  1.9184e-02,  7.9013e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1750, -0.0023, -0.1484,  0.0152, -0.1247, -0.0390, -0.3028,  0.0229,
         0.0443,  0.0897, -0.0357,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0194,  0.0305,  0.0195,  0.0171,  0.0243, -0.0744, -0.2010,  0.0018,
        -0.0859,  0.0110, -0.1439,  0.0042, -0.0748,  0.0262, -0.0763, -0.0233,
        -0.0386, -0.0619,  0.0226,  0.0161, -0.0272,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0729, -0.0148,  0.0114,  0.0183,  0.0284,  0.0125, -0.0190, -0.2631,
        -0.0335, -0.3174, -0.1029,  0.0380,  0.0246, -0.0432,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0329,  0.0154, -0.0160, -0.0056,  0.0158,  0.0025, -0.0029, -0.0436,
        -0.0195,  0.0312,  0.0661,  0.0407,  0.4842, -0.0356, -0.0175,  0.0218,
         0.0441, -0.0021, -0.0350, -0.0481, -0.0194,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0072, -0.3118,  0.0181, -0.1285,  0.0453, -0.1558, -0.0065, -0.1367,
        -0.0123, -0.0953, -0.0395,  0.0131,  0.0300,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.1139e-02, -1.0036e-01,  5.1767e-03,  1.1883e-02, -6.5718e-03,
        -3.1241e-03, -2.9837e-02, -4.5548e-02, -1.6863e-02, -3.4894e-03,
         1.0264e-03,  2.1855e-02,  3.5808e-03,  4.8621e-03,  5.8049e-03,
        -7.3177e-03,  2.8471e-03, -8.1503e-03,  6.9538e-04,  2.0443e-02,
         1.4561e-03,  1.3916e-02, -9.3990e-03, -2.2124e-05, -1.6567e-01,
        -3.8072e-03, -8.6352e-02, -1.0429e-01, -1.0619e-01, -1.5405e-03,
        -1.6850e-02,  2.5932e-03, -1.0448e-02, -1.1120e-02, -4.3673e-03,
         9.7257e-03, -4.5053e-02, -1.1204e-03,  6.2231e-04, -1.8620e-02,
        -2.0539e-02, -3.8503e-03, -2.2446e-03, -3.2456e-03,  6.3839e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0486,  0.0243,  0.0751,  0.0196,  0.0031, -0.0142, -0.0110,  0.0118,
         0.0171,  0.0025,  0.0845,  0.0094,  0.0040,  0.0097,  0.0752,  0.0125,
         0.0980, -0.0025,  0.0048,  0.0053, -0.0205,  0.0603, -0.0300,  0.0896,
         0.0130,  0.0131,  0.0038,  0.0167,  0.0590,  0.0105, -0.0100,  0.0175,
         0.1133,  0.0092,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0138, -0.0095,  0.0265, -0.0489, -0.0052, -0.0296,  0.0034, -0.0012,
         0.0062, -0.0093, -0.0125,  0.0014, -0.0096, -0.0050, -0.0058,  0.0103,
        -0.0173, -0.0174, -0.0379, -0.2118, -0.3579,  0.0411,  0.0259, -0.0551,
         0.0047, -0.0329,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0065, -0.3698, -0.0016, -0.1118,  0.0334,  0.0047, -0.0038, -0.0655,
        -0.0209, -0.0094, -0.0097,  0.0032,  0.0127,  0.0149, -0.0083, -0.0479,
        -0.1060, -0.0018, -0.0229, -0.0020, -0.0695,  0.0188,  0.0550,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.3824e-02, -1.5078e-01, -3.4836e-02, -4.8815e-02, -4.6605e-03,
         3.3925e-03, -7.7639e-03, -4.8480e-02, -7.7429e-02, -2.3653e-03,
         1.1792e-02,  6.2224e-03,  5.9048e-03,  2.2466e-03, -1.2287e-01,
         1.8860e-02, -1.3006e-02, -4.2981e-02, -7.6388e-02, -1.1390e-02,
         1.7174e-03, -1.2379e-02, -2.4358e-03,  1.2076e-02, -7.6623e-04,
        -2.2831e-02, -5.1600e-02, -8.5221e-02, -6.8567e-03, -5.2823e-03,
         1.5944e-02, -4.5649e-03,  2.0208e-03,  8.8346e-04,  4.5286e-04,
         1.1838e-02, -2.9770e-02, -6.9854e-05, -2.1171e-02, -8.1131e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0513,  0.0033, -0.0011, -0.0256, -0.0739, -0.0110, -0.0168, -0.0884,
         0.0138, -0.0035, -0.0051,  0.0047, -0.0332, -0.0033, -0.0364, -0.0499,
        -0.0817, -0.0069, -0.0095, -0.0771,  0.0162, -0.0923, -0.0769,  0.0216,
        -0.0042,  0.0017, -0.0087, -0.0653,  0.0092, -0.0167, -0.0725, -0.0098,
         0.0031, -0.0052,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0659, -0.0364, -0.0203, -0.0145, -0.0429, -0.0026,  0.0180, -0.0087,
         0.0083,  0.0068, -0.0026, -0.0205,  0.0174, -0.0135, -0.1497, -0.0007,
        -0.0035, -0.0019, -0.0330, -0.0019,  0.0111,  0.0002, -0.0022,  0.0154,
        -0.0471, -0.0738,  0.0369, -0.0097, -0.0031,  0.0029, -0.0079,  0.0051,
         0.0093, -0.0062, -0.0420,  0.0120, -0.0099, -0.0354, -0.0056, -0.0422,
        -0.0171, -0.0028, -0.0590, -0.0170, -0.0076,  0.0005, -0.0034, -0.0046,
        -0.0181,  0.0038, -0.0151, -0.0038], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0516, -0.0041, -0.0028, -0.0066, -0.1575,  0.0038,  0.0100,  0.0451,
         0.0109, -0.0113, -0.0010,  0.0060,  0.0036, -0.1825, -0.0287, -0.0584,
        -0.1337, -0.0716, -0.0432,  0.0064, -0.0094,  0.0074, -0.0158, -0.0737,
        -0.0014,  0.0105,  0.0120,  0.0035,  0.0113, -0.0007, -0.0157,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0294, -0.1830, -0.0009, -0.0492,  0.0210,  0.0149, -0.0211,  0.0026,
        -0.0081,  0.0158,  0.0157, -0.0431,  0.0007,  0.0049, -0.0723, -0.0644,
        -0.0236, -0.0697, -0.1043, -0.1224, -0.0274, -0.0008, -0.0164, -0.0272,
         0.0611,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0329,  0.0018,  0.0185, -0.0638,  0.0569, -0.2099, -0.4189,  0.0321,
        -0.0465, -0.0343, -0.0517, -0.0327,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0151, -0.2683, -0.0189, -0.0499, -0.1025, -0.0211, -0.0006,  0.0005,
        -0.0156,  0.0078,  0.0044, -0.0161, -0.0726, -0.1038,  0.0017, -0.0557,
        -0.0037, -0.0165,  0.0064,  0.0010, -0.0256, -0.0012, -0.0214,  0.0199,
         0.0148, -0.0492,  0.0267, -0.0214,  0.0375,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0452, -0.3225,  0.0056, -0.0048, -0.0349, -0.1282, -0.2583,  0.0344,
        -0.0869,  0.0236,  0.0299, -0.0256,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0909, -0.3515,  0.0300, -0.1085,  0.0007, -0.1846, -0.1116, -0.0218,
        -0.0545,  0.0145,  0.0108,  0.0207,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0808,  0.0099, -0.1915,  0.0094, -0.1119, -0.0117, -0.0005, -0.0138,
        -0.1432, -0.0160, -0.0274, -0.1860, -0.0174, -0.0073, -0.0147, -0.1136,
         0.0020,  0.0245,  0.0062, -0.0122,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0385, -0.1986, -0.0117, -0.0816, -0.1103,  0.0129, -0.0202,  0.0018,
        -0.0375, -0.0033, -0.0086, -0.0586, -0.0361,  0.0032,  0.0112,  0.0010,
        -0.0236,  0.0089,  0.0081, -0.0039, -0.0238,  0.0133,  0.0112, -0.0072,
        -0.0072,  0.0118, -0.0115, -0.0538, -0.0445,  0.0119,  0.0107, -0.0064,
        -0.0357, -0.0369,  0.0028,  0.0152, -0.0167,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0406,  0.8724,  0.0104, -0.0272,  0.0012,  0.0069,  0.0156,  0.0139,
         0.0118,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0034, -0.0301,  0.0323, -0.1385, -0.0188,  0.0185, -0.0570, -0.0818,
        -0.0177,  0.0054, -0.0028, -0.0083, -0.1289, -0.1363,  0.0025,  0.0008,
         0.0008,  0.0379, -0.0720, -0.1165,  0.0018, -0.0002,  0.0024, -0.0076,
        -0.0037,  0.0311, -0.0430,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0221, -0.1911,  0.0196, -0.0839, -0.0135, -0.0405, -0.0006,  0.0017,
         0.0026,  0.0140, -0.0026, -0.0753, -0.0314, -0.0021,  0.0007,  0.0198,
        -0.0635,  0.0076, -0.0040,  0.0034,  0.0012, -0.0695, -0.0904, -0.0026,
        -0.0077,  0.0124,  0.0031, -0.0399, -0.1001, -0.0144,  0.0053, -0.0084,
         0.0077, -0.0373,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0131, -0.0089, -0.0043, -0.0088, -0.0667,  0.0014,  0.0013,  0.0016,
         0.0023,  0.0022, -0.0064, -0.0733,  0.0087, -0.0023,  0.0059, -0.0490,
        -0.1385,  0.0179, -0.0332,  0.0015,  0.0148,  0.0003, -0.0455, -0.0133,
         0.0042,  0.0087, -0.0059,  0.0114, -0.0041, -0.0027,  0.0228, -0.0949,
        -0.1153, -0.0010, -0.0143, -0.0389, -0.0544, -0.0248, -0.0019, -0.0035,
        -0.0139, -0.0388, -0.0006,  0.0130,  0.0040,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0209, -0.0017,  0.0081, -0.0011, -0.0505, -0.1804, -0.2661, -0.0548,
         0.0279, -0.1606, -0.0400, -0.0065, -0.0977,  0.0269, -0.0093, -0.0475,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0194,  0.0174,  0.0118, -0.0044, -0.0148, -0.0081, -0.0194, -0.0118,
        -0.1149, -0.1927, -0.0557, -0.1465, -0.0030, -0.0915, -0.1669,  0.0004,
        -0.0086, -0.0041,  0.0061, -0.0067,  0.0023,  0.0082, -0.0097, -0.0303,
        -0.0066,  0.0349,  0.0038,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0240,  0.0024, -0.0576,  0.0068, -0.0458, -0.0003, -0.0294, -0.0040,
        -0.0168, -0.0014,  0.0037, -0.0050, -0.0317, -0.0040, -0.0353, -0.0072,
        -0.0181,  0.0094,  0.0412, -0.0582,  0.0153, -0.0308, -0.0233, -0.0842,
        -0.0071,  0.0009, -0.0500, -0.0489,  0.0038, -0.0204,  0.0048, -0.0338,
        -0.0418,  0.0113, -0.0017, -0.0164, -0.0119, -0.0659,  0.0092,  0.0094,
        -0.0191, -0.0331, -0.0042,  0.0090, -0.0391, -0.0026], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0308,  0.0432, -0.0039,  0.0176, -0.0065, -0.0285,  0.0117, -0.1349,
         0.0131,  0.0095, -0.0020,  0.0037,  0.0058, -0.0027, -0.0148, -0.0038,
        -0.0008, -0.0310, -0.1199, -0.3165, -0.0025, -0.0127, -0.0053, -0.0182,
        -0.0050,  0.0145, -0.1078,  0.0030, -0.0240, -0.0062,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1561,  0.0661,  0.0252, -0.0047,  0.0147, -0.0667, -0.0439, -0.3819,
        -0.1275, -0.0744, -0.0389,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0108, -0.4525,  0.0396, -0.0258, -0.0558, -0.0138, -0.0461, -0.0766,
        -0.0095,  0.0091,  0.0167,  0.0161, -0.1032, -0.0163,  0.0039, -0.0436,
        -0.0135, -0.0031, -0.0185,  0.0255,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1856, -0.2511, -0.1458, -0.0816,  0.0292, -0.0877, -0.0887,  0.0127,
        -0.0014, -0.0675,  0.0487,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0189, -0.2228, -0.1109, -0.1682, -0.0168, -0.0223, -0.0326, -0.0811,
        -0.0954, -0.0106, -0.0159, -0.0126, -0.0040, -0.0096, -0.0136, -0.0570,
         0.0150, -0.0129, -0.0121,  0.0077,  0.0599,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.0935e-02, -2.6879e-01, -2.2490e-01, -3.2919e-02, -4.2591e-02,
        -2.2834e-01, -4.3165e-03,  5.4057e-05, -3.9458e-02, -1.4529e-02,
        -4.6477e-03, -4.0976e-02, -2.6224e-03, -6.0077e-03, -4.4625e-03,
        -2.9899e-04, -2.5451e-02,  1.8697e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0202, -0.1696, -0.1264, -0.0228, -0.0111, -0.0048, -0.0082,  0.0015,
         0.0009, -0.0088,  0.0020,  0.0012, -0.0100, -0.0600, -0.0453, -0.0059,
        -0.0350, -0.0545, -0.0300, -0.0052, -0.0062, -0.0492, -0.0989,  0.0005,
        -0.0392,  0.0159, -0.0191, -0.0135, -0.0202, -0.0496, -0.0019, -0.0088,
         0.0050,  0.0068, -0.0118, -0.0083,  0.0021, -0.0134, -0.0063,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0420, -0.4313,  0.0085, -0.0202,  0.0047,  0.0026,  0.0128,  0.0153,
         0.0301,  0.0014, -0.0228, -0.1090, -0.0146,  0.0059,  0.0113,  0.0074,
        -0.0255,  0.0020,  0.0101,  0.0067, -0.0090, -0.0038,  0.0038,  0.0013,
        -0.0080, -0.0016, -0.0241, -0.0129, -0.0488, -0.0560,  0.0030, -0.0093,
         0.0015, -0.0329,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0094, -0.0908, -0.2452, -0.2645, -0.1566, -0.1224, -0.0534,  0.0158,
         0.0420,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0281, -0.2183, -0.0112, -0.0921, -0.0337, -0.0709, -0.0148, -0.0375,
        -0.0721, -0.0004, -0.0413, -0.0198,  0.0006, -0.0061, -0.0138, -0.0483,
        -0.0086, -0.0466, -0.0098,  0.0020, -0.0084,  0.0051,  0.0108, -0.0084,
        -0.0176, -0.0012,  0.0032, -0.0023, -0.0089, -0.0035, -0.0183, -0.0532,
         0.0025, -0.0288,  0.0089, -0.0101, -0.0191,  0.0077,  0.0031,  0.0027,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0491, -0.0511, -0.1309, -0.0020, -0.0172, -0.0085, -0.0065,  0.0007,
        -0.0655, -0.0603, -0.0032,  0.0068,  0.0012,  0.0031, -0.0015,  0.0089,
        -0.0008,  0.0121, -0.0471,  0.0016, -0.0308,  0.0159,  0.0056, -0.0105,
         0.0034, -0.0248, -0.0566,  0.0083, -0.0564, -0.0831,  0.0009,  0.0014,
         0.0049,  0.0003,  0.0047,  0.0031, -0.0392, -0.0640,  0.0058, -0.0029,
         0.0109, -0.0134,  0.0072,  0.0063,  0.0043, -0.0177, -0.0023,  0.0023,
         0.0055, -0.0016, -0.0120, -0.0160,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0174, -0.2669, -0.0549, -0.0349,  0.0169, -0.0072, -0.0749, -0.0162,
        -0.0450, -0.0048, -0.0015, -0.0425, -0.1211, -0.0006, -0.0774, -0.0905,
        -0.0613, -0.0008, -0.0502,  0.0021,  0.0079, -0.0049,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.4105e-02, -4.6534e-01,  3.9421e-02,  1.4175e-02,  1.0272e-02,
         3.4475e-02, -2.3232e-02, -8.7718e-02, -2.3126e-02, -4.4437e-05,
        -1.7089e-02, -4.4480e-02, -2.0443e-03, -1.2624e-02, -1.3861e-02,
        -5.5442e-02, -2.2348e-02, -1.5178e-02, -4.7295e-03,  8.9778e-02,
        -5.1771e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0003, -0.1595, -0.0150, -0.0230, -0.0045,  0.0014, -0.0299,  0.0174,
        -0.0198, -0.0026,  0.0028, -0.0104,  0.0019,  0.0045,  0.0037,  0.0053,
         0.0023,  0.0023, -0.0191,  0.0007,  0.0081, -0.0027, -0.0051,  0.0082,
         0.0024,  0.0064,  0.0007, -0.0359, -0.0024, -0.0387, -0.0680, -0.0654,
        -0.0088, -0.0552, -0.0509, -0.0078, -0.0078,  0.0087,  0.0095, -0.0055,
        -0.0627,  0.0024,  0.0121,  0.0073, -0.0375,  0.0038, -0.0477, -0.0264,
        -0.0046,  0.0130, -0.0263, -0.0084,  0.0020,  0.0157, -0.0052],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0114, -0.0005,  0.0175, -0.0135,  0.0005, -0.0356, -0.0771, -0.0106,
         0.0016, -0.0014, -0.0062,  0.0043, -0.0085, -0.0138, -0.0698, -0.0278,
        -0.0095, -0.0323, -0.0049, -0.1066,  0.0091, -0.0253, -0.0438, -0.0029,
         0.0114, -0.0025, -0.0108, -0.0860, -0.0273, -0.0095, -0.0127, -0.0016,
        -0.0184, -0.1017, -0.0084, -0.0450, -0.0059, -0.0224, -0.0350, -0.0339,
        -0.0035, -0.0054,  0.0150, -0.0016, -0.0070,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0095, -0.2661, -0.1006, -0.0961, -0.0205, -0.0239, -0.0269, -0.0195,
        -0.0591,  0.0016,  0.0081, -0.0169, -0.0154, -0.0748, -0.0466,  0.0320,
        -0.0126, -0.0440, -0.0095, -0.0339, -0.0468,  0.0059, -0.0210, -0.0087,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0101,  0.0182,  0.0021,  0.0004, -0.0231, -0.0075, -0.0491, -0.0633,
         0.0196, -0.0191, -0.0007, -0.0073, -0.0217, -0.0475,  0.0078,  0.0260,
        -0.0652, -0.0423, -0.0693, -0.0459,  0.0026,  0.0107, -0.0513, -0.0811,
        -0.0014, -0.0433, -0.0648, -0.0090, -0.0202, -0.0988, -0.0186,  0.0167,
        -0.0350,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.9433e-02, -4.4675e-01, -2.7543e-02, -8.6652e-02, -1.9595e-02,
        -9.6325e-03,  6.9500e-03, -2.7977e-02, -1.3504e-01, -1.4833e-02,
         5.3057e-03, -1.3867e-02, -1.7743e-03, -1.0082e-01, -1.9974e-04,
         7.3193e-03, -7.5416e-03, -1.9703e-02,  9.0697e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0960, -0.0195,  0.0417, -0.0238, -0.1653, -0.0138, -0.0780, -0.0942,
        -0.1175, -0.0180, -0.0086,  0.0396, -0.0006,  0.0096, -0.0051,  0.0014,
        -0.0191, -0.0556, -0.0310,  0.0023, -0.0350, -0.0931, -0.0030,  0.0030,
        -0.0252,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0073, -0.2935, -0.1600,  0.0048, -0.0506, -0.1385, -0.1262,  0.0014,
         0.0017, -0.0515, -0.0080,  0.0225, -0.0614, -0.0072,  0.0125, -0.0050,
         0.0480,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-2.3081e-02,  1.1417e-04, -1.4316e-02, -2.2957e-03,  3.1730e-03,
        -1.7727e-02,  4.3609e-04, -1.2959e-02, -1.4147e-02, -8.3929e-02,
        -1.9091e-02, -6.9977e-03,  1.7633e-02,  1.0309e-02, -4.5380e-02,
        -2.2760e-02, -1.1882e-01, -1.5401e-01, -1.5603e-02, -9.5348e-02,
         3.0713e-02, -7.4872e-02, -8.6153e-02, -7.4771e-03, -5.9753e-03,
         1.0225e-02, -1.1113e-02, -3.1662e-03,  3.5719e-02, -1.9804e-02,
        -2.4591e-02,  1.2066e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0028, -0.0944, -0.1921, -0.0283,  0.0161, -0.0055, -0.0499, -0.1611,
        -0.0126,  0.0050,  0.0098,  0.0033, -0.0084,  0.0080, -0.0289,  0.0081,
        -0.0314, -0.0111, -0.0583,  0.0069, -0.0475,  0.0020, -0.0482, -0.1193,
        -0.0030, -0.0126,  0.0033, -0.0221,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0112, -0.0469,  0.0011,  0.0212,  0.0106, -0.1613, -0.1398, -0.0286,
        -0.0227, -0.0291, -0.0200, -0.1004,  0.0018,  0.0085, -0.0068, -0.0593,
         0.0039, -0.1001, -0.0007,  0.0229, -0.0154, -0.0072, -0.0072, -0.0308,
        -0.1002,  0.0050,  0.0291,  0.0009, -0.0072,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.5970e-02, -1.4805e-01, -2.2918e-02, -4.9508e-02, -1.2733e-01,
         9.0103e-03, -1.0807e-02,  2.8923e-03, -9.2839e-03,  1.8373e-02,
         5.1904e-02, -2.9359e-02, -4.6182e-02, -1.2842e-01, -2.4379e-03,
        -2.5387e-02,  5.0330e-03, -5.5492e-02, -1.6342e-03, -1.3158e-02,
        -3.3517e-03,  9.4613e-03, -3.0857e-02, -8.5746e-03,  5.2747e-03,
        -3.3081e-02, -4.9614e-02,  4.5406e-03, -8.7135e-03, -4.3896e-03,
        -1.3627e-04,  3.3789e-02, -5.0705e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.2966e-02, -2.3035e-01, -1.0360e-02, -9.0676e-02, -6.7630e-03,
        -4.5570e-02,  2.8048e-05, -2.8184e-02, -2.9959e-02, -6.0075e-02,
        -1.1942e-01, -2.4663e-02, -1.7707e-02, -2.5503e-03,  1.3325e-03,
         1.3817e-02, -3.9118e-02,  1.6378e-03, -1.3377e-01, -8.3751e-03,
        -7.5098e-02, -5.8218e-04,  1.1694e-02,  1.5305e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0358, -0.2029, -0.1620,  0.0121, -0.0614, -0.0123, -0.0370, -0.0456,
         0.0093, -0.0529, -0.0084, -0.0188, -0.0711,  0.0032, -0.0102,  0.0027,
         0.0013, -0.0443,  0.0015, -0.0987, -0.0254,  0.0415,  0.0416,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0023, -0.1380,  0.0159, -0.0054, -0.0379, -0.0648,  0.0185,  0.0019,
        -0.0076, -0.0066,  0.0096,  0.0032, -0.0167,  0.0080, -0.0091,  0.0177,
        -0.1010, -0.0455, -0.0502, -0.0004, -0.0087, -0.0360, -0.0182, -0.0035,
        -0.0652, -0.0177, -0.0103,  0.0046,  0.0027, -0.0432, -0.0756, -0.0027,
        -0.0328,  0.0044, -0.0418, -0.0543, -0.0008, -0.0029,  0.0142,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0283, -0.0054, -0.0036, -0.0340, -0.1944,  0.0009,  0.0085,  0.0113,
        -0.0080, -0.0005, -0.0026, -0.0020, -0.0035, -0.0063,  0.0008, -0.0792,
        -0.0462, -0.0047,  0.0077, -0.0221, -0.0752,  0.0027,  0.0145, -0.0005,
        -0.0889, -0.0584, -0.0008, -0.0307, -0.0042, -0.0459, -0.0156, -0.0049,
        -0.0048, -0.0097, -0.0082,  0.0038,  0.0022, -0.0920,  0.0171,  0.0006,
         0.0010,  0.0256, -0.0226,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.6174e-03, -2.3561e-01, -2.5312e-02, -6.8207e-03, -3.3848e-04,
         5.1992e-03, -1.1843e-02,  1.1262e-02, -6.9305e-02, -2.1987e-03,
         1.1935e-02, -1.5809e-04, -1.9984e-02, -4.0284e-03, -2.4370e-04,
         3.3106e-03, -2.1659e-02, -7.1616e-02, -7.7468e-03,  2.4830e-03,
         3.6797e-03,  4.1979e-03, -8.5585e-03,  5.2273e-03,  9.4913e-03,
         4.8775e-03,  1.7965e-03,  9.8829e-03, -7.9694e-04,  2.8659e-03,
         4.3639e-03, -3.3911e-02, -6.9776e-02,  3.3153e-02, -1.2288e-02,
         5.6583e-03,  7.6560e-03, -4.1479e-02, -3.7234e-02, -1.7734e-03,
        -3.9859e-02, -4.1408e-02, -3.3712e-03, -9.3604e-03, -8.8759e-03,
         1.4652e-02, -3.7965e-03,  7.8304e-04, -1.6348e-03,  7.1835e-04,
         7.6302e-03,  9.2345e-03,  2.9054e-04,  4.0053e-03,  2.8780e-03,
         2.4467e-02, -1.5694e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1748, -0.3805, -0.0134, -0.0144, -0.0448,  0.0108, -0.0648, -0.1402,
        -0.0391, -0.0005, -0.0588,  0.0133, -0.0009,  0.0437,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0235, -0.1961, -0.0969, -0.0131,  0.0031, -0.0462,  0.0048,  0.0030,
        -0.0425,  0.0074, -0.0577, -0.0716, -0.0092, -0.0343, -0.0409, -0.0012,
        -0.0260, -0.0269, -0.0072, -0.0477, -0.1296, -0.0219, -0.0209, -0.0015,
        -0.0022, -0.0020,  0.0109,  0.0212, -0.0296,  0.0010,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0408, -0.2640,  0.0087, -0.3071, -0.2266, -0.0110, -0.1167, -0.0252,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0109, -0.1715, -0.0089, -0.0280, -0.0587, -0.0578, -0.0019, -0.0099,
        -0.0051, -0.0124,  0.0118, -0.0058, -0.1127, -0.0923, -0.0037,  0.0013,
        -0.0004,  0.0033,  0.0033,  0.0034,  0.0119,  0.0078,  0.0024, -0.0263,
        -0.0401,  0.0042, -0.0400, -0.0187, -0.0556, -0.1253, -0.0030, -0.0155,
        -0.0458,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.0026e-03, -1.9869e-01, -7.3892e-02, -1.4577e-02, -1.1968e-02,
        -3.6607e-02, -2.3503e-02, -1.0023e-02,  2.0682e-03, -2.0670e-02,
        -2.1739e-03, -1.8796e-02, -1.4603e-02, -5.2805e-03,  5.7536e-03,
        -7.3454e-03, -3.8985e-03, -2.8173e-03, -2.7871e-02,  3.2862e-03,
         5.3721e-04,  4.4140e-03, -6.6998e-03, -2.5051e-03,  6.6919e-03,
        -2.1174e-03,  1.0908e-03, -4.3797e-02,  4.8933e-03,  2.2327e-03,
        -2.5998e-03, -3.0089e-04, -4.0668e-02, -4.8192e-02,  9.9863e-03,
        -7.7883e-03, -3.2899e-02,  9.9452e-03, -2.9335e-03,  1.8615e-04,
         4.3115e-03,  8.4256e-03,  3.4279e-04, -3.2433e-02, -4.1520e-03,
        -1.2487e-03, -8.0329e-04, -6.5802e-03,  5.5841e-04, -5.8771e-03,
         2.2486e-03, -3.3533e-02, -8.9372e-03, -1.1235e-02, -7.8169e-03,
         4.0876e-03, -2.2134e-04,  3.0391e-03, -5.1860e-03, -2.2567e-02,
        -1.1401e-02, -1.7205e-02, -5.3259e-02,  9.0183e-04, -1.4970e-03,
        -3.2561e-03, -9.3589e-03, -1.9776e-03,  3.6386e-03,  1.6654e-03,
        -1.0937e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0380, -0.0103, -0.0395, -0.0026, -0.0558, -0.0011, -0.1449, -0.0220,
         0.0038,  0.0109, -0.0057, -0.0106,  0.0070,  0.0113,  0.0238,  0.0104,
         0.0020,  0.0016,  0.0189, -0.0670, -0.0103,  0.0086,  0.0103, -0.0397,
         0.0099, -0.0413, -0.0052,  0.0031, -0.0014, -0.0142,  0.0131, -0.0029,
        -0.0630, -0.0972, -0.0267,  0.0015, -0.0860, -0.0327,  0.0026,  0.0073,
         0.0317, -0.0040,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0391, -0.2959, -0.0102,  0.0044,  0.0036,  0.0062,  0.0063,  0.0106,
        -0.0086, -0.0519, -0.0153, -0.0383, -0.0555,  0.0010, -0.0063, -0.0073,
         0.0019, -0.0774, -0.0889, -0.0031, -0.0192, -0.0411,  0.0070,  0.0067,
         0.0028, -0.0596, -0.0015, -0.0025, -0.0488,  0.0144,  0.0076, -0.0024,
         0.0055,  0.0051, -0.0060,  0.0185, -0.0195,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0244, -0.5344, -0.0247, -0.1135,  0.0108,  0.0574, -0.0167, -0.0099,
        -0.0166, -0.0040, -0.0241,  0.0611, -0.0362, -0.0322,  0.0340,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0011, -0.0533,  0.0021, -0.0067, -0.1537, -0.2336, -0.0046, -0.0698,
        -0.2436, -0.0323,  0.0087, -0.0333, -0.0282, -0.0352,  0.0211, -0.0043,
        -0.0032, -0.0291,  0.0197,  0.0162,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0111,  0.0160,  0.0039, -0.0019, -0.1232,  0.0024,  0.0035, -0.0192,
        -0.0028,  0.0152, -0.0520, -0.0690,  0.0051,  0.0037, -0.0331, -0.0259,
         0.0056, -0.0259,  0.0182, -0.0476, -0.0127,  0.0008, -0.0021, -0.0006,
        -0.0054, -0.0401, -0.1260, -0.0113,  0.0043, -0.0353, -0.0646, -0.0037,
         0.0008,  0.0028, -0.0434, -0.0708,  0.0143, -0.0155, -0.0027, -0.0045,
         0.0176,  0.0226, -0.0127,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0229, -0.0019,  0.0028,  0.0011, -0.0495,  0.0019, -0.0291, -0.0761,
        -0.0266, -0.0158,  0.0194,  0.0002,  0.0091, -0.1329, -0.1683, -0.0011,
        -0.0010,  0.0047, -0.0011,  0.0183, -0.0976, -0.0163,  0.0253,  0.0113,
         0.0064, -0.0034,  0.0036,  0.0011, -0.0892, -0.0091,  0.0014, -0.0347,
         0.0018, -0.0325, -0.0387,  0.0037,  0.0162, -0.0236,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.1270e-03, -2.7571e-01, -1.3568e-01, -1.1757e-01,  1.0750e-02,
        -8.6067e-02,  2.3870e-02,  1.7632e-05, -6.6379e-02, -6.5648e-02,
         7.2133e-03,  5.2139e-03, -4.8822e-02,  4.7917e-03,  1.2819e-02,
        -3.0890e-03,  1.4528e-03, -7.2952e-02,  1.5901e-02,  1.5187e-02,
         1.4047e-03,  2.4331e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.5475e-02, -1.4069e-01, -2.8426e-02,  2.7207e-03,  1.4120e-02,
         3.6407e-03, -2.7261e-03,  3.0666e-03, -1.6163e-02, -1.4228e-01,
        -5.0527e-02,  6.2528e-03, -1.2308e-02, -1.3672e-02, -5.2218e-02,
        -3.6024e-02, -8.0434e-02, -2.1027e-02, -6.4879e-03, -3.7354e-02,
         1.9178e-02, -5.5490e-05, -3.0050e-02,  1.9252e-03,  1.9433e-03,
         3.4306e-03, -6.8884e-03, -3.1495e-03,  2.0089e-03, -3.5068e-02,
        -1.7591e-03,  6.7347e-03,  8.0287e-03, -2.0385e-02, -2.5585e-02,
        -4.1336e-02, -2.0405e-02,  4.5281e-03,  2.9396e-02, -1.2538e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0406, -0.0156, -0.0076,  0.0073, -0.0179, -0.0153, -0.0024, -0.0091,
        -0.0635, -0.0176, -0.0074, -0.0030, -0.0231, -0.1876, -0.0976,  0.0092,
         0.0215, -0.0089, -0.1235,  0.0070,  0.0111, -0.0038,  0.0105,  0.0003,
         0.0128, -0.0654,  0.0153, -0.0629, -0.0595, -0.0089,  0.0305,  0.0335,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.0738e-03, -4.4588e-01, -1.8862e-03, -1.3357e-03, -1.6923e-02,
         6.9160e-03, -6.5866e-03, -5.4432e-02, -7.7488e-02, -1.3814e-04,
         2.2889e-02, -3.1941e-02, -1.9726e-03, -3.0197e-03, -5.7108e-03,
        -3.7285e-03, -5.0497e-02, -1.3287e-02,  2.2327e-03,  9.5815e-04,
         2.0559e-03,  1.7597e-03, -5.4835e-03, -8.3108e-02, -1.4737e-02,
         1.9201e-03, -5.6022e-02, -8.4805e-03, -1.4102e-02, -2.4349e-02,
         6.3713e-03, -7.7833e-03, -6.9222e-03,  1.8792e-03,  1.1136e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0112, -0.0292, -0.1329, -0.1171, -0.0181, -0.0325, -0.1378, -0.1367,
        -0.0171, -0.1037, -0.0234, -0.0550,  0.0011, -0.0571, -0.0245, -0.0183,
        -0.0350, -0.0097,  0.0088,  0.0112,  0.0195,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0200, -0.3542,  0.0078, -0.0601, -0.1021, -0.0796, -0.0261, -0.0410,
        -0.0280,  0.0244,  0.0094,  0.0029, -0.0024, -0.0161, -0.0433,  0.0181,
        -0.0408, -0.0803,  0.0434,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0432, -0.0913, -0.1085,  0.0033, -0.0260, -0.0039,  0.0618, -0.0044,
         0.0166, -0.1554, -0.1104,  0.0009, -0.0591, -0.1574, -0.0226, -0.0292,
        -0.0132, -0.0276,  0.0363,  0.0289,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0335, -0.4288, -0.0114, -0.0871, -0.0033,  0.0015, -0.0158, -0.0487,
         0.0007, -0.0033, -0.0088, -0.0179, -0.0024, -0.0418, -0.0053, -0.0432,
        -0.0800, -0.0270, -0.0295, -0.0505,  0.0368,  0.0045, -0.0183,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0164, -0.2071, -0.0665, -0.0683, -0.0095, -0.0805, -0.0104, -0.0784,
        -0.0799, -0.0635, -0.0273, -0.0582,  0.0004, -0.0648,  0.0108,  0.0050,
        -0.0042,  0.0012, -0.0163, -0.0211, -0.0455,  0.0033,  0.0046, -0.0367,
         0.0202,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7633e-02, -4.0703e-01,  7.9213e-03, -5.6417e-03, -2.3459e-03,
        -4.2159e-02,  1.3797e-02, -2.9360e-03,  8.8509e-03,  1.7941e-03,
        -8.9036e-02, -6.9694e-03,  6.1814e-03,  1.4631e-02,  1.3101e-02,
        -3.1290e-03, -2.1809e-03, -1.1626e-03, -8.4474e-04, -1.1070e-02,
        -9.4728e-02, -6.0218e-02, -5.3374e-02, -5.4761e-03, -4.1117e-03,
        -8.9689e-03,  1.2058e-02,  4.4047e-04,  5.1409e-03, -1.5890e-02,
        -2.6126e-02, -3.3439e-04, -3.9151e-02,  8.1620e-03,  7.4068e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0959, -0.0107, -0.0144, -0.0706, -0.0121, -0.0143, -0.0539,  0.0016,
        -0.1059, -0.0079, -0.0290, -0.0369, -0.0315, -0.0519, -0.0875, -0.1368,
        -0.0375, -0.0884,  0.0328, -0.0033,  0.0547, -0.0221,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0846, -0.0609,  0.0893, -0.0076,  0.0071, -0.0568, -0.0116, -0.0348,
        -0.0034, -0.1548,  0.0442, -0.0013, -0.0127, -0.0356, -0.0398, -0.1423,
        -0.0750,  0.0230, -0.0182, -0.0239, -0.0732,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1298, -0.3597, -0.0236,  0.0064, -0.0091,  0.0249, -0.0561, -0.1003,
         0.0358,  0.0078,  0.0236,  0.0092,  0.0298, -0.0461,  0.0022, -0.0268,
        -0.0082,  0.0784, -0.0221,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0034,  0.0018, -0.0405,  0.0118,  0.0049, -0.0699,  0.0110, -0.0093,
        -0.0061, -0.0003, -0.0134, -0.0395, -0.0026, -0.0421,  0.0066,  0.0051,
        -0.0484, -0.0617,  0.0145, -0.0402, -0.0042, -0.0010, -0.0358, -0.0093,
        -0.0647, -0.0129, -0.0257, -0.0466, -0.0239, -0.1197,  0.0049,  0.0077,
        -0.0383, -0.0995,  0.0046, -0.0089,  0.0050,  0.0005,  0.0003, -0.0112,
        -0.0038,  0.0052,  0.0075, -0.0062,  0.0040,  0.0036,  0.0119],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0395, -0.2996, -0.1266, -0.0123,  0.0040, -0.0037,  0.0107,  0.0017,
        -0.0214,  0.0127,  0.0086, -0.0024,  0.0089,  0.0270,  0.0055, -0.0089,
        -0.0946,  0.0003,  0.0051, -0.0176, -0.0092,  0.0095, -0.0699, -0.0903,
         0.0087,  0.0011, -0.0024, -0.0010, -0.0075,  0.0071,  0.0060,  0.0061,
         0.0437,  0.0261,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.5042e-02,  2.9596e-01, -2.0739e-02,  6.1092e-03,  3.4861e-03,
         4.9572e-02, -8.1295e-04, -1.5595e-02,  6.0922e-03,  1.5718e-02,
        -2.7074e-02, -8.2511e-03,  5.5161e-02,  8.4005e-02, -2.0048e-05,
         8.1873e-03,  2.8726e-02,  1.4138e-02,  3.6422e-02,  8.1123e-02,
         9.8570e-04,  1.1765e-02,  1.4162e-02,  1.0409e-02,  4.6687e-03,
        -1.2192e-02, -2.0917e-02,  1.7718e-02, -3.1833e-03,  3.6084e-02,
         1.9084e-03, -2.0217e-02,  2.4039e-02, -3.9511e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-0.0214, -0.4198, -0.0032, -0.0094, -0.0619, -0.0931, -0.0231,  0.0014,
        -0.0039, -0.1062, -0.0145, -0.0134, -0.0505, -0.1408, -0.0153,  0.0013,
        -0.0016,  0.0190,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0368,  0.0056,  0.0144,  0.0345,  0.0047, -0.0088,  0.0085, -0.0007,
        -0.1314, -0.2552, -0.0055,  0.0076, -0.0410, -0.0128,  0.0126,  0.0084,
        -0.0070,  0.0247,  0.0139, -0.1793,  0.0074,  0.0162, -0.0433, -0.0267,
        -0.0077,  0.0440, -0.0410,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 9.3980e-02, -1.8398e-02,  1.1077e-02, -4.6759e-03, -4.6097e-04,
         1.7738e-02,  1.0217e-02,  4.3787e-02,  1.6453e-01,  1.9793e-02,
         3.1588e-03,  4.0198e-02,  1.9422e-01,  7.3249e-03, -1.7019e-04,
        -1.1525e-03,  5.7340e-03,  6.5689e-03,  1.9477e-02,  2.1599e-01,
         9.1179e-03,  5.9086e-04,  3.2208e-02,  1.6602e-04, -6.2900e-02,
         1.6370e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0125, -0.1875, -0.0081, -0.0966, -0.0787,  0.0294, -0.0022, -0.0110,
        -0.0722, -0.0184, -0.0767, -0.0536,  0.0067, -0.0090, -0.0777, -0.0728,
        -0.0572, -0.0676,  0.0046, -0.0078,  0.0497,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0307, -0.2782, -0.1641, -0.1845, -0.0368,  0.0158, -0.1104, -0.0071,
        -0.0275,  0.0136,  0.0125, -0.0874,  0.0068,  0.0140,  0.0105,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0712, -0.0303, -0.0102,  0.0062, -0.0605,  0.0051,  0.0126, -0.0373,
        -0.0091,  0.0325, -0.0063,  0.0280, -0.5182, -0.0541, -0.0395, -0.0356,
        -0.0289,  0.0144,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0349, -0.2562, -0.0836, -0.1111, -0.0154, -0.0079,  0.0010, -0.0205,
        -0.0029, -0.0596, -0.1051, -0.0655, -0.0150, -0.0567,  0.0023, -0.0038,
         0.0205, -0.0637, -0.0743,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0053, -0.1449, -0.1023,  0.0048, -0.0084, -0.0297,  0.0168, -0.0080,
        -0.0185,  0.0047, -0.0091, -0.0034, -0.0288, -0.0036, -0.0097,  0.0073,
        -0.0483, -0.0016,  0.0006,  0.0174, -0.0621, -0.0028, -0.0429, -0.0009,
        -0.0371,  0.0036, -0.0460,  0.0197,  0.0034, -0.0450, -0.0072,  0.0282,
        -0.0527, -0.0011, -0.0453, -0.0223, -0.0634,  0.0077, -0.0099, -0.0037,
         0.0008, -0.0056,  0.0101,  0.0052], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0423, -0.1986, -0.0165, -0.0718, -0.0510, -0.0111, -0.0222, -0.0028,
        -0.0686, -0.0178, -0.0159,  0.0143,  0.0013, -0.0017,  0.0194, -0.0024,
         0.0737, -0.0608,  0.0005, -0.0300, -0.0381,  0.0009, -0.0035,  0.0040,
        -0.0114,  0.0076, -0.0101, -0.0672,  0.0088,  0.0120,  0.0030, -0.0412,
        -0.0028, -0.0528, -0.0139,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.4846e-02, -4.1085e-01, -1.6290e-02, -1.0638e-01, -2.5406e-02,
        -8.0941e-03, -8.2976e-02, -6.5406e-02,  2.1944e-04,  3.8902e-03,
         2.4141e-03, -4.7222e-02, -6.9723e-02, -1.0212e-02,  1.6057e-03,
        -5.4212e-02, -2.0551e-03, -4.1828e-03,  1.0843e-02,  7.0840e-03,
         6.8877e-03,  7.3255e-03, -1.9796e-02,  1.2081e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.3854e-02, -2.4380e-01, -1.9624e-02, -8.0028e-02, -3.9257e-02,
        -2.4045e-02, -1.1939e-03, -2.9631e-02, -7.6214e-02,  1.0229e-02,
        -3.7086e-02,  7.1683e-03, -7.0134e-04, -4.6644e-02, -5.4072e-02,
        -1.5536e-03,  6.0837e-03, -8.1846e-03, -5.1593e-02,  7.3240e-03,
         1.4881e-04, -3.6837e-02,  2.1465e-03, -2.6354e-02,  4.2470e-03,
        -3.4107e-02, -2.7323e-02, -2.6167e-02, -4.4834e-03, -9.1303e-03,
        -3.4606e-02,  4.1460e-03, -6.4404e-03,  2.5573e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0766,  0.0069, -0.0195, -0.0274,  0.0110,  0.0044, -0.1338, -0.0165,
         0.0183,  0.0035,  0.0200, -0.0156, -0.0958, -0.0080, -0.1218, -0.1824,
        -0.0111,  0.0058, -0.0040, -0.0005,  0.0097,  0.0021,  0.0059, -0.0946,
        -0.0503, -0.0070, -0.0295, -0.0084,  0.0021, -0.0075,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0462, -0.0244, -0.0805, -0.3249, -0.0393, -0.0163, -0.0259,  0.0020,
        -0.1784, -0.2030, -0.0037, -0.0518,  0.0036,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0430,  0.0108, -0.0134,  0.0073, -0.0112, -0.0390,  0.0029, -0.1996,
        -0.2327, -0.0019, -0.0175, -0.0937, -0.0361, -0.0559, -0.2186, -0.0032,
         0.0091, -0.0039,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0676, -0.2196, -0.3980,  0.1157, -0.1167,  0.0341, -0.0408, -0.0075,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.5833e-02, -2.2086e-01, -5.9176e-03, -4.0440e-02, -2.9007e-02,
        -1.6124e-02, -2.1638e-02, -8.4963e-03, -2.5507e-03, -1.8597e-02,
        -3.2295e-03,  1.8813e-02, -1.7570e-02, -3.1045e-03,  3.6891e-03,
        -1.0047e-02,  4.7806e-03,  6.0638e-03, -4.1023e-03, -6.7400e-05,
        -3.8415e-03,  9.8840e-03, -2.3398e-02, -2.6314e-02,  9.5870e-04,
        -2.4045e-02, -5.1474e-03,  7.9044e-03, -7.8073e-02, -1.8522e-02,
         1.1362e-02, -5.2910e-02, -7.9299e-03, -1.0516e-01, -1.6574e-02,
        -2.2982e-02, -1.8923e-02, -8.4635e-03,  1.1009e-02,  6.0396e-03,
        -7.5116e-03, -1.4365e-02, -1.2926e-02, -2.8679e-03,  8.6220e-03,
        -7.4319e-03, -1.5908e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0342,  0.1870, -0.0801,  0.1917,  0.0815, -0.0607, -0.0646,  0.0316,
         0.1121,  0.1130, -0.0435,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.6293e-02, -3.8259e-01, -1.1107e-02, -4.3715e-02,  4.2929e-03,
         1.0477e-04, -3.7039e-02, -1.4877e-01,  2.7672e-02, -1.1571e-01,
        -4.2833e-02,  5.3200e-03, -8.7883e-02, -1.0525e-02, -3.2589e-03,
         1.1501e-02,  6.0715e-03,  2.5313e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1074, -0.1401, -0.3266,  0.0076,  0.0838,  0.0168, -0.1469,  0.0114,
        -0.0502,  0.0008,  0.0088,  0.0176,  0.0091,  0.0090,  0.0639,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0843, -0.4171, -0.0212,  0.0201,  0.0166, -0.0605, -0.1334, -0.0280,
        -0.0195, -0.0569, -0.0763, -0.0070, -0.0032,  0.0021,  0.0539,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0343,  0.0058, -0.0082,  0.0634,  0.0073,  0.0240, -0.0013, -0.0044,
        -0.0748, -0.0541, -0.0476, -0.1195, -0.1713, -0.0104, -0.0008, -0.0192,
        -0.0302, -0.0055,  0.0068,  0.0086, -0.0944, -0.1606,  0.0097,  0.0167,
         0.0031, -0.0183,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0980, -0.0039,  0.0360,  0.0012,  0.0310, -0.0010, -0.0453, -0.1766,
        -0.3827, -0.0230, -0.0686,  0.0024, -0.0208,  0.0741,  0.0352,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0081, -0.3800,  0.0011, -0.0470, -0.0989, -0.0262, -0.0511, -0.0009,
        -0.0033, -0.1237,  0.0223, -0.0022, -0.0224, -0.0067, -0.0057, -0.0034,
        -0.0108,  0.0022, -0.0466,  0.0082, -0.0282,  0.0106,  0.0188,  0.0174,
         0.0061,  0.0480,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0487,  0.0215,  0.0140,  0.0273, -0.0138, -0.0473, -0.0141,  0.0088,
        -0.0524, -0.0681, -0.1267, -0.0175, -0.0164, -0.0129, -0.1431, -0.0751,
         0.0274, -0.0551,  0.0051, -0.0586, -0.0258, -0.0048, -0.0269, -0.0266,
         0.0486, -0.0135,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-9.0444e-02, -2.9807e-01,  1.0147e-02,  2.0689e-02, -4.4453e-03,
         1.0545e-02, -3.4775e-02, -4.2960e-02, -4.5272e-05, -2.8278e-02,
        -8.3672e-03, -3.1350e-02, -6.3947e-02,  1.8242e-03, -2.7317e-02,
        -5.6755e-03, -1.4485e-03, -4.0172e-03, -1.3578e-03,  2.4204e-03,
         1.8342e-02, -3.0698e-03,  1.6107e-02, -2.9676e-02, -5.8138e-02,
         1.7653e-02, -1.9798e-02, -8.6766e-03,  2.3497e-02, -1.6124e-03,
         1.6298e-03, -1.6795e-02, -4.5418e-03, -1.2230e-02,  5.7262e-02,
        -2.2847e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0295, -0.1402, -0.0486,  0.0059, -0.0516,  0.0084, -0.0419, -0.0882,
         0.0096,  0.0129, -0.0350,  0.0047, -0.0036,  0.0062, -0.0079, -0.0196,
         0.0031, -0.0015,  0.0112, -0.0006, -0.0093, -0.0918, -0.0032, -0.0070,
        -0.0395, -0.0726, -0.0066, -0.0411, -0.0446,  0.0090, -0.0034, -0.0473,
        -0.0468,  0.0088, -0.0176, -0.0213,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0368,  0.0082,  0.0195, -0.0343, -0.0153, -0.2186, -0.0456, -0.0262,
         0.0007, -0.0111, -0.0343, -0.0854, -0.2327,  0.1154,  0.0089, -0.0153,
         0.0169,  0.0139, -0.0567,  0.0043,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4602e-02, -1.6574e-01, -3.2747e-02, -3.9743e-02,  1.2919e-02,
        -2.6727e-04,  2.9971e-03, -8.0175e-02, -5.4219e-03, -2.1114e-02,
         8.6490e-03, -5.9384e-03,  5.5287e-03,  1.0467e-02, -8.4807e-02,
        -6.5239e-02, -1.2521e-02, -5.9716e-02, -1.1255e-02, -8.7951e-03,
        -1.7149e-02, -2.0831e-03, -1.9166e-03,  2.3332e-02,  6.3189e-03,
        -3.3082e-03,  1.4597e-04, -5.6261e-03,  3.8405e-03, -2.6463e-02,
        -1.1242e-01,  1.1444e-02,  6.0128e-03, -1.0141e-01, -1.5031e-03,
        -1.0816e-02, -4.0464e-03,  1.3514e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5842e-02, -2.4470e-01, -1.2018e-01, -9.4072e-02, -2.2135e-04,
         3.7864e-03, -3.0832e-03, -5.8389e-03, -6.8265e-03, -2.2830e-02,
        -5.1584e-02, -8.2008e-02, -7.1243e-03, -1.7598e-03,  1.4743e-03,
        -4.0834e-02,  7.1331e-03, -3.2375e-02, -4.1975e-04,  6.2599e-03,
         7.1507e-03,  5.4217e-03,  1.3632e-02,  1.4127e-02, -1.0632e-03,
        -5.9076e-03, -1.3974e-02, -3.6093e-02, -6.5009e-02,  4.4335e-03,
        -7.3979e-03, -4.3168e-03,  1.1015e-02,  4.3861e-04, -2.9010e-02,
         7.5381e-03,  6.9557e-03,  6.1754e-04,  5.8488e-05, -4.2346e-03,
         1.3254e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0051, -0.4451, -0.0179, -0.0428, -0.0123, -0.0903, -0.0005, -0.0060,
        -0.0192,  0.0105,  0.0031, -0.0914, -0.0835, -0.0114,  0.0175, -0.0717,
        -0.0011, -0.0494, -0.0023,  0.0084,  0.0098,  0.0008,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0214,  0.0109,  0.0039,  0.0032,  0.0122, -0.0393, -0.0120,  0.0168,
         0.0130,  0.0111,  0.0143, -0.0473, -0.0046, -0.0427,  0.0030, -0.1104,
         0.0166, -0.0414, -0.0009, -0.0670, -0.0986,  0.0085,  0.0328, -0.0468,
         0.0071, -0.0052, -0.0101, -0.0310, -0.1062,  0.0189, -0.0655, -0.0370,
        -0.0235, -0.0056, -0.0113,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0111, -0.0147, -0.2553, -0.0211, -0.0568,  0.0012,  0.0150, -0.0125,
         0.0176, -0.0020, -0.0537,  0.0095, -0.0134,  0.0130, -0.0013, -0.0209,
         0.0425, -0.0221, -0.0926, -0.0105, -0.0635,  0.0243, -0.0148, -0.0731,
        -0.0238, -0.0109, -0.0484,  0.0194, -0.0349,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.7041e-02, -3.5861e-01, -8.0042e-02, -5.0435e-02, -2.3896e-02,
        -6.5615e-02,  6.4375e-03, -1.4338e-05, -2.5551e-02, -3.1175e-02,
        -4.0868e-02, -4.1312e-02, -4.5910e-03, -4.3196e-03,  4.9604e-03,
        -3.5576e-02, -3.3046e-02, -1.4951e-02, -3.5889e-02,  9.6954e-03,
        -7.7475e-03, -2.0760e-02, -1.2298e-02, -1.7052e-02,  1.8118e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0456,  0.0097, -0.0072,  0.0064, -0.1921,  0.0103, -0.0337, -0.1004,
        -0.1723, -0.0233, -0.0120,  0.0236, -0.0045,  0.0411, -0.0913, -0.0866,
         0.0195, -0.0702, -0.0036,  0.0036,  0.0015,  0.0103,  0.0311,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0220,  0.0100,  0.0070,  0.0160,  0.0165,  0.0077, -0.0104, -0.0111,
        -0.0045, -0.0096,  0.0016, -0.0011, -0.0843, -0.0677, -0.0064, -0.0338,
        -0.0036, -0.0794,  0.0060,  0.0064,  0.0214,  0.0137,  0.0082,  0.0016,
        -0.0163, -0.0197, -0.1444,  0.0130, -0.0928,  0.0124, -0.0871, -0.0160,
        -0.0183, -0.0171, -0.0589, -0.0062, -0.0079,  0.0216, -0.0180,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.6125e-02, -2.3932e-02, -1.6326e-05, -1.4075e-01, -3.0382e-01,
         3.7016e-02, -8.8739e-02, -4.3491e-02, -1.4984e-01,  8.1104e-02,
        -3.1368e-02,  3.3794e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-2.5115e-02, -1.6124e-01, -2.3742e-01, -2.2631e-02, -2.8616e-03,
         2.1626e-02, -4.3703e-02,  1.0939e-02, -4.5671e-03, -4.6819e-02,
        -9.8556e-04,  2.6669e-03, -2.7549e-02, -4.5148e-03, -5.1632e-02,
         2.7464e-03, -7.6166e-03, -3.8538e-02,  6.1664e-04, -7.7892e-03,
        -1.1597e-02, -2.6172e-02,  5.7105e-03, -6.7468e-03, -3.2270e-03,
         6.2030e-04,  1.0591e-02, -7.1122e-03, -1.8888e-02,  9.8351e-04,
        -7.2735e-03, -1.2635e-03, -6.8092e-02,  3.8404e-03,  7.1061e-03,
         9.0276e-03, -4.6556e-02, -8.8168e-05,  4.5706e-04, -1.6455e-03,
         7.7794e-04,  1.7104e-03,  1.4518e-02, -1.1072e-02, -1.3354e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0042, -0.5657, -0.0392, -0.0797,  0.0039, -0.0060,  0.0762, -0.0357,
        -0.1091, -0.0181, -0.0582,  0.0040,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1692, -0.6000,  0.0209, -0.1468, -0.0131, -0.0026, -0.0232, -0.0241,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0306, -0.1781, -0.0073,  0.0017, -0.0520, -0.0118,  0.0014, -0.0038,
        -0.0179, -0.0731, -0.0129,  0.0036, -0.0037, -0.0008,  0.0042,  0.0019,
        -0.0008, -0.0042, -0.0016, -0.0048,  0.0034,  0.0065, -0.0043, -0.0009,
        -0.0081,  0.0024,  0.0008, -0.0005, -0.0349, -0.0042, -0.0016, -0.0101,
         0.0015,  0.0015,  0.0006,  0.0044,  0.0125,  0.0031, -0.0018,  0.0021,
        -0.0059, -0.0028,  0.0082, -0.0085, -0.0119,  0.0134, -0.0762,  0.0041,
        -0.0730, -0.1174, -0.0392, -0.0002, -0.0318,  0.0082, -0.0177, -0.0068,
        -0.0195, -0.0054,  0.0037, -0.0047,  0.0201], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0262, -0.0069, -0.0037, -0.0112, -0.0535,  0.0206, -0.0347,  0.0042,
        -0.0250,  0.0004, -0.0092, -0.0965,  0.0126, -0.0929, -0.0900,  0.0036,
        -0.0222, -0.0079, -0.0380, -0.0253, -0.0666, -0.0626, -0.0209,  0.0119,
        -0.0402, -0.0444, -0.0062, -0.0116, -0.0142, -0.0245, -0.0128,  0.0255,
        -0.0738,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.0766e-02, -2.4991e-02, -1.2256e-01, -1.3654e-01, -1.8492e-02,
         1.1328e-02, -8.1035e-02, -1.4203e-04,  3.0177e-03, -1.1499e-02,
        -2.8076e-03, -1.9607e-03, -4.3795e-02, -1.7624e-02, -9.8567e-03,
         2.6686e-02, -1.1867e-02, -9.1990e-02, -2.4645e-01,  3.1809e-02,
        -1.1995e-02, -4.3579e-02,  1.8522e-02, -1.0685e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0070, -0.0884, -0.0370, -0.0520, -0.0035, -0.1160,  0.0077, -0.0618,
         0.0066, -0.0962, -0.0761, -0.1136, -0.1579, -0.0090,  0.0176,  0.0096,
         0.0052, -0.0254,  0.0172,  0.0007, -0.0305, -0.0327,  0.0245,  0.0037,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0224,  0.0234, -0.0833, -0.0833, -0.0060, -0.0040,  0.0159, -0.0433,
        -0.0016,  0.0121,  0.0058, -0.0010, -0.0018, -0.0030, -0.0446, -0.1301,
        -0.0832,  0.0108,  0.0034, -0.0154, -0.0682,  0.0025, -0.0539, -0.0921,
         0.0185,  0.0118, -0.0531, -0.0041, -0.0085, -0.0173, -0.0099, -0.0097,
         0.0473,  0.0087,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0068, -0.0206, -0.0039,  0.0168, -0.1124, -0.0005, -0.0146, -0.0014,
         0.0129, -0.0107, -0.0021,  0.0054, -0.0192, -0.1361,  0.0218,  0.0340,
        -0.1466,  0.0244, -0.0803,  0.0080, -0.1005, -0.1567,  0.0405,  0.0069,
         0.0168,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0896, -0.0166, -0.0436, -0.0023,  0.0132,  0.0481, -0.0994, -0.0512,
        -0.1310, -0.2079,  0.0426, -0.0478, -0.1025, -0.0371,  0.0540, -0.0133,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-9.7740e-03, -2.0965e-01, -5.1567e-03, -1.6058e-01, -3.9470e-02,
        -6.7525e-02, -1.3863e-01, -6.8656e-03,  7.8030e-03, -4.9723e-05,
        -3.2722e-03, -6.5236e-03, -4.5867e-02,  1.5456e-03,  5.2939e-03,
        -1.9036e-02, -4.3013e-02, -9.5915e-02, -5.5121e-02, -2.5136e-02,
        -5.9970e-03,  2.9413e-02,  1.8369e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0161, -0.0334, -0.0399, -0.2307, -0.4475, -0.0374,  0.0382, -0.0079,
        -0.0347,  0.0381, -0.0205,  0.0056,  0.0222, -0.0279,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0730, -0.0846, -0.0698, -0.1151, -0.0156,  0.0175, -0.0616, -0.0113,
        -0.0385, -0.0119, -0.1030, -0.0990, -0.0113, -0.0121,  0.0036,  0.0029,
         0.0289, -0.1015, -0.0121, -0.0699, -0.0203, -0.0099,  0.0061, -0.0206,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0766,  0.0218, -0.1532,  0.0202, -0.2403,  0.0246, -0.4062,  0.0052,
         0.0361, -0.0007,  0.0151,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0075,  0.0046,  0.0220, -0.0013,  0.0315, -0.0732, -0.1673, -0.0594,
        -0.1248, -0.0223, -0.0883,  0.0135, -0.0864,  0.0011, -0.0757, -0.0357,
        -0.0292, -0.0978, -0.0114,  0.0114, -0.0354,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0691,  0.0444, -0.0089,  0.0206,  0.0480,  0.0074, -0.0450, -0.3031,
        -0.0731, -0.2652, -0.0301,  0.0251,  0.0221, -0.0379,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0165, -0.0824,  0.0050, -0.0216,  0.0019, -0.0016, -0.0199,  0.0531,
        -0.0048, -0.0064, -0.1165, -0.0366, -0.3858, -0.0058, -0.0161, -0.0178,
        -0.0123, -0.0221,  0.0450,  0.0699,  0.0589,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0690,  0.3566,  0.0168,  0.1516, -0.0236,  0.1086,  0.0399,  0.0727,
         0.0342,  0.0861,  0.0321,  0.0041,  0.0047,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0187, -0.1459, -0.0076, -0.0117, -0.0138, -0.0114, -0.0528, -0.0637,
        -0.0204,  0.0053,  0.0016,  0.0030,  0.0077,  0.0029, -0.0019, -0.0205,
        -0.0015,  0.0031, -0.0061,  0.0156, -0.0129,  0.0107,  0.0031, -0.0076,
        -0.1034,  0.0209, -0.0631, -0.0848, -0.0631, -0.0140, -0.0122,  0.0068,
        -0.0031, -0.0028,  0.0092,  0.0026, -0.0358, -0.0040,  0.0075, -0.0166,
        -0.0358, -0.0063, -0.0096,  0.0281,  0.0207], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0230, -0.0445, -0.0903, -0.0165, -0.0066,  0.0121,  0.0015, -0.0387,
        -0.0058, -0.0095, -0.0795,  0.0073, -0.0125,  0.0219, -0.1210, -0.0238,
        -0.1328, -0.0016,  0.0116,  0.0090,  0.0221, -0.0839,  0.0066, -0.0857,
         0.0143,  0.0194, -0.0059, -0.0110, -0.0428, -0.0089, -0.0035,  0.0039,
        -0.0171,  0.0052,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0418,  0.0011,  0.0114, -0.0290, -0.0013, -0.0054,  0.0214, -0.0007,
         0.0488,  0.0064, -0.0066, -0.0121, -0.0042,  0.0072,  0.0046,  0.0240,
         0.0137,  0.0370, -0.0293, -0.2101, -0.2973,  0.0366,  0.0172, -0.0952,
        -0.0095, -0.0280,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0071, -0.3276, -0.0108, -0.0791,  0.0264, -0.0052, -0.0038, -0.0997,
        -0.0081, -0.0033,  0.0080,  0.0024,  0.0014,  0.0074, -0.0162, -0.1034,
        -0.0771,  0.0257, -0.0569, -0.0273, -0.0645,  0.0097,  0.0289,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0447, -0.1503, -0.0279, -0.0363, -0.0029,  0.0004, -0.0138, -0.0579,
        -0.0957, -0.0070,  0.0022, -0.0018,  0.0117, -0.0007, -0.1066, -0.0020,
         0.0122, -0.0427, -0.0589, -0.0092, -0.0014,  0.0022,  0.0033,  0.0208,
         0.0003, -0.0041, -0.0684, -0.0953, -0.0099, -0.0124,  0.0067, -0.0039,
        -0.0025,  0.0041, -0.0099, -0.0010, -0.0277,  0.0063, -0.0270,  0.0080,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0223,  0.0184,  0.0117, -0.0357, -0.0995, -0.0005, -0.0279, -0.0614,
         0.0057, -0.0089,  0.0022, -0.0124, -0.0481, -0.0057, -0.0255, -0.0688,
        -0.0702, -0.0035, -0.0282, -0.0450,  0.0013, -0.0810, -0.0964, -0.0062,
         0.0033,  0.0074, -0.0087, -0.0714, -0.0085, -0.0249, -0.0526, -0.0048,
        -0.0030,  0.0291,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0099, -0.0059, -0.0114, -0.0202, -0.0714, -0.0103,  0.0046, -0.0089,
         0.0053,  0.0014,  0.0018, -0.0266,  0.0094, -0.0120, -0.1245,  0.0250,
         0.0068, -0.0024, -0.0467, -0.0019,  0.0044,  0.0034, -0.0003,  0.0075,
        -0.0822, -0.1100,  0.0351, -0.0249, -0.0031, -0.0107, -0.0225,  0.0008,
        -0.0008, -0.0028, -0.0362,  0.0026, -0.0273, -0.0407, -0.0068, -0.0309,
        -0.0278, -0.0173, -0.0431, -0.0111,  0.0070,  0.0007, -0.0018,  0.0034,
        -0.0075, -0.0024, -0.0079, -0.0106], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.7214e-02,  2.0559e-03, -1.3356e-02, -1.6399e-02, -1.5543e-01,
        -5.3304e-03, -3.4104e-02,  3.3758e-02,  7.8436e-03,  1.2659e-03,
         2.6433e-02,  8.5249e-03, -2.2498e-02, -1.1588e-01, -7.6927e-03,
        -4.2918e-02, -1.2530e-01, -4.9054e-02, -1.1080e-01, -1.1222e-02,
        -3.7904e-02,  6.5601e-03, -3.5001e-02, -5.8306e-02, -2.4720e-03,
        -2.0820e-03,  9.5893e-03, -6.5611e-05,  1.4287e-04,  5.4502e-03,
        -5.3547e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0424, -0.1916,  0.0013, -0.0224,  0.0116,  0.0259, -0.0248, -0.0010,
        -0.0024,  0.0176,  0.0081, -0.0595,  0.0277, -0.0251, -0.1026, -0.0762,
        -0.0012, -0.0826, -0.1181, -0.0851, -0.0316, -0.0094,  0.0087, -0.0112,
        -0.0119,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0152,  0.0060, -0.0191,  0.0283, -0.0429, -0.3616, -0.3934, -0.0616,
         0.0211, -0.0039, -0.0149,  0.0321,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0270, -0.2530, -0.0161, -0.0679, -0.1051,  0.0027,  0.0177,  0.0048,
         0.0081,  0.0050,  0.0214, -0.0049, -0.0413, -0.1101, -0.0187, -0.0807,
         0.0042,  0.0118,  0.0034,  0.0131, -0.0241, -0.0027, -0.0005, -0.0240,
         0.0099, -0.0276,  0.0244, -0.0517, -0.0182,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0905, -0.4413,  0.0372, -0.0585,  0.0446, -0.1224, -0.1255,  0.0454,
         0.0012, -0.0130,  0.0159, -0.0045,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0782, -0.4008, -0.0438, -0.1315,  0.0004, -0.0868, -0.1299, -0.0211,
        -0.0467, -0.0085, -0.0226,  0.0296,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0146,  0.0435, -0.2693, -0.0253, -0.1567,  0.0146,  0.0067, -0.0076,
        -0.1161, -0.0192,  0.0198, -0.0862,  0.0060,  0.0294, -0.0113, -0.0805,
        -0.0088,  0.0049, -0.0239,  0.0554,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0010, -0.2144, -0.0218, -0.0648, -0.1168, -0.0159, -0.0281, -0.0153,
        -0.0156, -0.0189, -0.0182, -0.0708, -0.0392, -0.0086, -0.0146,  0.0058,
        -0.0376,  0.0098,  0.0005,  0.0010, -0.0529,  0.0203,  0.0142,  0.0098,
         0.0057,  0.0111,  0.0063, -0.0222, -0.0205, -0.0008,  0.0045,  0.0024,
        -0.0332, -0.0487, -0.0091,  0.0133,  0.0062,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0207,  0.7492,  0.0581,  0.0411,  0.0255,  0.0212, -0.0418,  0.0203,
        -0.0220,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0141,  0.0222,  0.0073, -0.1158, -0.0051,  0.0501, -0.0749, -0.0764,
         0.0310,  0.0066, -0.0125, -0.0172, -0.1576, -0.1475, -0.0011,  0.0106,
        -0.0062, -0.0200, -0.0589, -0.1005,  0.0066,  0.0132,  0.0026,  0.0027,
        -0.0099,  0.0229,  0.0065,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0084, -0.2908, -0.0069, -0.0691, -0.0312, -0.0319, -0.0067,  0.0027,
        -0.0073,  0.0038, -0.0057, -0.0840, -0.0288,  0.0089, -0.0070,  0.0042,
        -0.0640, -0.0006,  0.0046,  0.0030,  0.0068, -0.0636, -0.0581,  0.0024,
        -0.0015, -0.0017, -0.0010, -0.0175, -0.1043, -0.0128, -0.0028,  0.0097,
         0.0069,  0.0414,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-6.6690e-03, -6.1287e-04,  9.7575e-03,  2.2010e-03, -8.7282e-02,
        -9.2631e-03,  1.4199e-02,  1.2414e-02,  1.7318e-02,  9.8193e-03,
        -2.3435e-03, -7.0494e-02,  6.7529e-04,  2.0564e-03,  2.5333e-03,
        -7.8704e-02, -9.5076e-02, -4.3638e-03, -3.0203e-02, -7.5973e-03,
         9.6339e-03,  2.7889e-03, -3.1839e-02,  9.3263e-03, -1.0454e-02,
         3.0816e-03, -2.2623e-05, -1.0995e-02,  3.2289e-03, -8.9212e-03,
         4.8144e-03, -7.0994e-02, -1.2691e-01,  2.3779e-02, -1.0500e-02,
        -4.2472e-02, -3.6845e-02, -1.2385e-02,  1.7447e-02,  1.8047e-03,
        -1.9484e-02, -4.2806e-02,  7.4927e-03,  2.1366e-02, -7.0290e-03,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.3793e-03, -2.4243e-02,  4.3455e-03, -3.3711e-02, -8.9371e-03,
        -2.5010e-01, -2.1489e-01,  1.4477e-02, -2.3103e-02, -1.3649e-01,
        -3.0743e-02, -2.4568e-04, -1.2251e-01, -1.2933e-02,  2.9244e-02,
        -8.6651e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0146,  0.0131, -0.0109,  0.0202,  0.0140, -0.0072, -0.0038, -0.0136,
        -0.1200, -0.1717, -0.0604, -0.0795, -0.0178, -0.1145, -0.1489, -0.0139,
        -0.0115, -0.0129,  0.0022, -0.0097, -0.0008,  0.0225, -0.0077, -0.0272,
        -0.0234, -0.0281, -0.0297,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0548, -0.0086, -0.0760,  0.0094, -0.0322,  0.0044, -0.0230, -0.0088,
        -0.0261, -0.0127, -0.0016, -0.0123, -0.0261,  0.0015, -0.0194, -0.0015,
        -0.0202, -0.0018,  0.0072, -0.0262,  0.0048, -0.0259, -0.0095, -0.0698,
         0.0117, -0.0007, -0.0301, -0.0541,  0.0100, -0.0384, -0.0060, -0.0392,
        -0.0437, -0.0110, -0.0027, -0.0507, -0.0169, -0.0705,  0.0108,  0.0030,
        -0.0258, -0.0429, -0.0083,  0.0090,  0.0184, -0.0122], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.9112e-02, -4.0249e-02,  5.5983e-03, -1.2249e-02, -8.1681e-03,
        -1.7981e-02,  1.2374e-03,  1.3380e-01,  2.2241e-02,  2.5851e-03,
        -6.5774e-03, -9.3227e-03, -2.0762e-03, -9.3016e-06,  2.6579e-02,
         7.5433e-03, -1.0283e-02,  2.2835e-02,  1.3221e-01,  3.0464e-01,
        -4.1778e-04,  1.1589e-02,  1.5395e-02,  1.4511e-02, -1.6384e-04,
         1.0944e-02,  7.3285e-02, -1.9137e-02,  7.0747e-03, -5.2189e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0715,  0.0153,  0.0058,  0.0572, -0.0005,  0.0864,  0.1398,  0.4824,
         0.1227, -0.0016, -0.0168,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0156, -0.4272,  0.0476, -0.0204, -0.0486, -0.0173, -0.0460, -0.1289,
         0.0098,  0.0049,  0.0162,  0.0129, -0.0185, -0.0116, -0.0142, -0.0634,
        -0.0058, -0.0492, -0.0269,  0.0149,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0859, -0.3752, -0.2056, -0.0926,  0.0335, -0.0703, -0.0751,  0.0015,
        -0.0043, -0.0228,  0.0333,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0784, -0.2769, -0.1007, -0.0942, -0.0161,  0.0057, -0.0059, -0.1344,
        -0.0970, -0.0126, -0.0125, -0.0242, -0.0239, -0.0252, -0.0079, -0.0429,
         0.0134,  0.0026, -0.0054, -0.0050,  0.0150,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.6950e-03, -3.4432e-01, -1.9022e-01, -1.4893e-02,  7.2597e-03,
        -2.5516e-01, -3.9068e-02,  2.1542e-03, -2.7252e-02, -1.1818e-02,
        -6.6255e-03, -5.7045e-02,  9.1639e-03, -5.1018e-03,  6.3216e-03,
        -1.4149e-04,  1.2873e-03,  2.0476e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0221, -0.1125, -0.1664, -0.0033, -0.0108,  0.0186, -0.0104,  0.0097,
         0.0044, -0.0110,  0.0090,  0.0036, -0.0141, -0.0530, -0.0438,  0.0023,
        -0.0289, -0.0385, -0.0151,  0.0020, -0.0125, -0.0507, -0.1007, -0.0111,
        -0.0358, -0.0133, -0.0564, -0.0120, -0.0462, -0.0311,  0.0033, -0.0024,
        -0.0004,  0.0132, -0.0074,  0.0044,  0.0065, -0.0069, -0.0062,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0568, -0.3618,  0.0031, -0.0952, -0.0042, -0.0193, -0.0094, -0.0008,
        -0.0345, -0.0212, -0.0072, -0.1214,  0.0026,  0.0021,  0.0142,  0.0077,
        -0.0766,  0.0030, -0.0074, -0.0035, -0.0033, -0.0031,  0.0034,  0.0020,
         0.0056, -0.0008, -0.0186, -0.0035, -0.0288, -0.0450, -0.0007,  0.0107,
         0.0119, -0.0106,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.0567,  0.0228, -0.2071, -0.3101,  0.0483, -0.1528, -0.0009, -0.0781,
        -0.1232,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0096, -0.1666, -0.0389, -0.0643, -0.0249, -0.0816, -0.0042, -0.0506,
        -0.0370, -0.0046, -0.0448, -0.0217, -0.0030, -0.0196, -0.0030, -0.0890,
        -0.0199, -0.0359, -0.0048, -0.0049,  0.0038,  0.0082,  0.0033,  0.0064,
        -0.0614, -0.0007, -0.0048,  0.0023, -0.0041, -0.0029, -0.0242, -0.0581,
         0.0010, -0.0315,  0.0239, -0.0054,  0.0034,  0.0019, -0.0102,  0.0137,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0118, -0.0663, -0.1990, -0.0109, -0.0615, -0.0224, -0.0102, -0.0107,
        -0.0602, -0.0494, -0.0049, -0.0004,  0.0006, -0.0013, -0.0192,  0.0013,
         0.0109,  0.0181, -0.0586,  0.0074, -0.0389, -0.0020,  0.0136, -0.0040,
        -0.0029, -0.0182, -0.0213,  0.0147, -0.0371, -0.0547,  0.0128,  0.0018,
        -0.0011, -0.0056,  0.0077,  0.0022, -0.0303, -0.0230, -0.0024, -0.0006,
         0.0113,  0.0157,  0.0088, -0.0048, -0.0011,  0.0012, -0.0078,  0.0056,
         0.0043, -0.0044,  0.0133,  0.0018,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0108, -0.2094, -0.0226, -0.0587, -0.0326, -0.0061, -0.1136, -0.0164,
        -0.0238, -0.0023,  0.0263, -0.0240, -0.1047, -0.0163, -0.0752, -0.0686,
        -0.0471, -0.0119, -0.0631, -0.0081, -0.0395, -0.0188,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.1332e-02, -5.1030e-01,  2.8969e-04, -3.6033e-03,  1.0362e-02,
         4.6204e-04,  1.9671e-02, -6.8928e-02,  9.2853e-03, -1.8182e-02,
         6.7608e-03, -6.0113e-02, -2.7616e-03, -2.1336e-02, -2.5073e-04,
        -1.0572e-01, -3.8982e-02, -2.3353e-02,  2.2557e-03,  3.2669e-02,
        -3.3389e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0011, -0.1878, -0.0349, -0.0265,  0.0012, -0.0039, -0.0351,  0.0017,
        -0.0229, -0.0067,  0.0026, -0.0091,  0.0062, -0.0095,  0.0024,  0.0049,
         0.0096,  0.0006, -0.0469,  0.0071,  0.0050,  0.0011, -0.0069,  0.0018,
        -0.0036,  0.0014, -0.0019, -0.0417,  0.0033, -0.0310, -0.0792, -0.0677,
         0.0132, -0.0372, -0.0299, -0.0023,  0.0051,  0.0028, -0.0076, -0.0002,
        -0.0495,  0.0023,  0.0074, -0.0017, -0.0482,  0.0096, -0.0465, -0.0284,
         0.0019,  0.0005, -0.0209, -0.0063, -0.0073,  0.0037,  0.0023],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0288, -0.0048,  0.0165, -0.0172, -0.0002, -0.0299, -0.0857, -0.0131,
         0.0127,  0.0010, -0.0002,  0.0049, -0.0052, -0.0240, -0.0633, -0.0079,
        -0.0093, -0.0294, -0.0299, -0.0879, -0.0119, -0.0425, -0.0487,  0.0020,
         0.0059, -0.0062,  0.0093, -0.0703, -0.0150,  0.0133,  0.0014,  0.0003,
        -0.0163, -0.0793, -0.0016, -0.0370, -0.0089, -0.0429, -0.0419, -0.0370,
        -0.0066,  0.0009,  0.0119,  0.0031,  0.0141,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.2328e-02, -2.4613e-01, -1.1519e-01, -1.1129e-01, -1.3692e-03,
        -9.9120e-03, -2.6260e-02, -1.9206e-02, -2.8049e-02, -1.2107e-04,
         1.8044e-02, -2.8944e-02,  9.8300e-03, -4.5694e-02, -6.6448e-02,
        -1.7165e-05, -1.3629e-02, -6.2167e-02,  2.1792e-02, -4.9638e-02,
        -5.7045e-02,  3.5769e-02,  6.8003e-03, -4.3276e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0427,  0.0226,  0.0044,  0.0058, -0.0118, -0.0051, -0.0539, -0.0458,
         0.0109, -0.0043,  0.0023,  0.0028, -0.0072, -0.0377,  0.0079,  0.0050,
        -0.0413, -0.0497, -0.0566, -0.0686, -0.0149,  0.0109, -0.0394, -0.0962,
         0.0020, -0.0439, -0.0491, -0.0333, -0.0540, -0.0801, -0.0025, -0.0457,
        -0.0416,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0743, -0.4163, -0.0269, -0.1243, -0.0035,  0.0031,  0.0022, -0.0218,
        -0.0975, -0.0016,  0.0061, -0.0087,  0.0070, -0.1135,  0.0143, -0.0230,
        -0.0062, -0.0308,  0.0189,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0258,  0.0046,  0.0367, -0.0233, -0.1150, -0.0448, -0.0935, -0.0912,
        -0.1285, -0.0039,  0.0224, -0.0345, -0.0080, -0.0189,  0.0037, -0.0010,
         0.0007, -0.0606, -0.0523, -0.0098, -0.0563, -0.0861, -0.0047, -0.0010,
         0.0728,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0348, -0.2338, -0.0960,  0.0215, -0.0428, -0.1658, -0.1456,  0.0223,
        -0.0277, -0.0758, -0.0132,  0.0017, -0.0801,  0.0040, -0.0134, -0.0048,
         0.0168,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-0.0159, -0.0064,  0.0076,  0.0133, -0.0002,  0.0100, -0.0073, -0.0186,
        -0.0104,  0.1016,  0.0338, -0.0153, -0.0199, -0.0085, -0.0051,  0.0130,
         0.0977,  0.1013, -0.0131,  0.0950, -0.0288,  0.0907,  0.1268,  0.0007,
        -0.0102, -0.0232, -0.0118, -0.0172, -0.0087,  0.0173,  0.0079, -0.0626,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.4158e-02, -1.2594e-01, -2.0275e-01,  2.5840e-02, -1.3647e-03,
        -5.1248e-03, -8.9920e-02, -1.2579e-01, -1.2258e-02,  1.6361e-02,
        -1.9256e-04, -4.8073e-03, -1.8065e-02,  9.4817e-03, -1.1864e-02,
         1.7166e-02,  1.5951e-03,  1.1587e-02, -3.7189e-02, -2.0103e-03,
        -4.2220e-02,  8.3057e-03, -4.5294e-02, -1.0313e-01, -8.8602e-03,
        -1.4813e-02,  5.3766e-03, -1.8535e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.1408e-03, -6.2258e-03,  4.9157e-05,  8.3151e-03,  9.8985e-03,
        -1.2475e-01, -1.7412e-01, -4.5089e-02,  1.3871e-02, -2.1268e-02,
         1.3722e-02, -6.0866e-02,  9.5776e-03,  9.6308e-04,  1.2521e-02,
        -6.1879e-02,  1.3983e-03, -1.0124e-01,  1.0846e-02,  1.4038e-02,
        -1.8677e-02,  3.7875e-02, -4.5899e-03,  1.5072e-02, -5.9027e-02,
         1.6208e-02, -5.2298e-03,  8.3886e-02,  5.9658e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0084, -0.1621, -0.0215, -0.0415, -0.1162,  0.0110, -0.0004,  0.0009,
        -0.0031,  0.0063,  0.0056, -0.0187, -0.0363, -0.0842,  0.0232, -0.0492,
         0.0166, -0.1065, -0.0058, -0.0261, -0.0034, -0.0063, -0.0402, -0.0038,
        -0.0194, -0.0782, -0.0478, -0.0047,  0.0018, -0.0053, -0.0060, -0.0068,
        -0.0326,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0066, -0.2357, -0.0110, -0.1189, -0.0118, -0.0653, -0.0160, -0.0395,
        -0.0146, -0.0500, -0.1029, -0.0028, -0.0098,  0.0016,  0.0072,  0.0071,
        -0.0329,  0.0066, -0.1209,  0.0153, -0.0827, -0.0046, -0.0075, -0.0286,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.1629e-02, -1.7971e-01, -1.7825e-01, -1.5524e-02, -8.4596e-02,
        -2.5343e-03, -1.1051e-02, -6.2517e-02,  8.7444e-03, -3.7003e-02,
         2.3670e-02, -3.9955e-02, -7.9341e-02, -1.8955e-02,  5.6575e-03,
        -1.1613e-04,  5.5870e-03, -4.8464e-02, -8.7325e-03, -4.4401e-02,
        -3.2239e-02, -5.6301e-02,  5.0223e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.3879e-02, -2.2514e-01,  8.8771e-03,  2.9083e-04, -2.7631e-02,
        -4.5093e-02,  9.4891e-03, -1.4224e-03,  2.8318e-04,  8.6918e-03,
         2.8822e-03, -1.0638e-02, -3.7291e-02, -7.3313e-03, -1.0641e-02,
        -2.7738e-02, -8.7368e-02, -5.7816e-02, -4.1234e-02, -8.7081e-03,
        -1.0616e-03, -1.1831e-02, -1.0581e-02, -5.0938e-03, -5.3751e-02,
        -1.2958e-02, -1.8315e-04,  4.0410e-03,  3.0827e-04, -3.6374e-02,
        -3.9273e-02, -1.0980e-02, -2.6784e-02, -6.7098e-03, -3.0625e-02,
        -4.0347e-02, -4.7489e-03, -6.0625e-02,  1.1283e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0463, -0.0491, -0.0118, -0.0210, -0.1396,  0.0028,  0.0053,  0.0035,
         0.0001,  0.0083, -0.0015, -0.0031,  0.0027,  0.0035, -0.0082, -0.0519,
         0.0096,  0.0092,  0.0019,  0.0108, -0.0475,  0.0021,  0.0139, -0.0070,
        -0.0906, -0.1158, -0.0119, -0.0246, -0.0216, -0.0701, -0.0080,  0.0032,
        -0.0042, -0.0068, -0.0108, -0.0016, -0.0187, -0.0942, -0.0038, -0.0026,
        -0.0112,  0.0035, -0.0361,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.4463e-03, -1.5521e-01,  6.0568e-03,  3.9116e-03,  3.4621e-02,
         9.8136e-03, -8.4665e-03, -2.5000e-02, -6.3886e-02,  1.1305e-02,
         6.0419e-03,  3.4277e-04, -2.5627e-02,  1.4912e-03, -2.6151e-04,
        -6.1464e-03, -3.6450e-02, -5.6996e-02,  1.7080e-03,  1.8409e-03,
         7.1919e-03, -2.2858e-02, -7.7032e-03, -5.6163e-03,  1.2568e-02,
        -1.1341e-02,  3.7457e-03, -3.7847e-03, -1.8002e-03, -2.1497e-03,
        -1.3577e-02, -4.4354e-02, -9.5088e-02, -5.9631e-03, -5.9749e-03,
        -2.0235e-03, -5.5414e-03, -9.0868e-03, -3.7455e-02, -1.1882e-02,
        -3.7133e-02, -6.8890e-02, -6.7351e-03,  6.7273e-03, -5.8627e-03,
        -1.9394e-02, -2.8082e-03,  3.4716e-03, -1.9180e-03, -9.3740e-04,
        -1.0004e-04, -1.2496e-02,  1.4242e-02,  7.6461e-04,  1.1974e-03,
        -2.6981e-02, -1.6018e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0264, -0.4731,  0.0120,  0.0045, -0.0619, -0.0048, -0.1150, -0.1341,
        -0.0363, -0.0192, -0.0478,  0.0089, -0.0028,  0.0531,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.6132e-03, -2.0578e-01, -6.4335e-02,  1.8209e-03, -1.5287e-02,
        -2.8671e-02,  8.0586e-03,  5.3451e-05, -3.7980e-02, -9.2973e-04,
        -3.9393e-02, -7.7789e-02,  2.2874e-03, -1.5324e-02, -5.7454e-02,
        -1.1006e-02, -3.1816e-02, -2.7504e-02,  1.0521e-02, -5.2440e-02,
        -7.9934e-02, -5.7861e-03, -3.4057e-02,  7.7654e-03,  1.6141e-02,
         2.9394e-02,  6.3093e-03,  1.5658e-02, -5.1757e-02, -6.1135e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0937, -0.3765, -0.0528, -0.2901, -0.1321,  0.0131,  0.0311,  0.0105,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0094, -0.2259, -0.0154, -0.0230, -0.0538, -0.0678, -0.0048, -0.0104,
        -0.0026, -0.0058, -0.0137,  0.0082, -0.0321, -0.0832, -0.0113, -0.0027,
        -0.0056, -0.0003,  0.0064,  0.0036,  0.0121,  0.0034,  0.0090, -0.0599,
        -0.0338,  0.0015, -0.0358, -0.0063, -0.0654, -0.0793, -0.0031, -0.0545,
        -0.0498,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0017, -0.1564, -0.0783, -0.0276, -0.0145, -0.0320, -0.0134, -0.0144,
         0.0068, -0.0193, -0.0078, -0.0103, -0.0259, -0.0021,  0.0051,  0.0055,
         0.0005,  0.0024, -0.0230, -0.0009, -0.0058,  0.0056, -0.0083, -0.0067,
        -0.0021,  0.0031, -0.0110, -0.0662, -0.0035,  0.0025, -0.0065, -0.0002,
        -0.0388, -0.0736, -0.0233,  0.0123, -0.0153,  0.0073,  0.0064, -0.0002,
        -0.0023,  0.0010, -0.0012, -0.0364, -0.0125,  0.0045, -0.0025, -0.0002,
        -0.0016, -0.0034,  0.0032, -0.0266, -0.0099, -0.0128,  0.0014, -0.0013,
        -0.0023,  0.0064,  0.0004, -0.0162, -0.0019, -0.0128, -0.0401,  0.0007,
         0.0022, -0.0038, -0.0049,  0.0097,  0.0178, -0.0003, -0.0135],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.1594e-02, -2.1317e-02, -4.0285e-02,  8.8127e-03, -5.8002e-02,
         5.4946e-03, -8.9798e-02, -9.8987e-03,  6.1318e-03, -2.0025e-04,
        -9.3137e-03, -6.3944e-03, -2.8930e-03,  7.1671e-03, -2.7052e-03,
         7.8498e-03, -8.8518e-03,  1.6632e-02, -5.1621e-03, -1.2219e-01,
         2.9219e-03,  4.8862e-04,  1.2188e-02, -4.4530e-02,  1.6098e-02,
        -6.9047e-02, -1.1647e-02,  2.9880e-03, -4.1382e-03, -2.1129e-03,
         2.4880e-04, -8.0226e-03, -7.6245e-02, -8.6625e-02,  2.5792e-03,
        -3.3127e-02, -5.5868e-02, -2.3129e-02,  2.8779e-06, -1.3141e-03,
         1.5247e-02,  2.0742e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5205e-02, -2.5408e-01,  3.8122e-05, -3.9233e-03,  1.9949e-04,
        -9.2854e-03,  1.4178e-02, -2.3944e-02,  5.3048e-03, -4.7590e-02,
        -1.2649e-02, -5.6166e-02, -8.3843e-02, -7.8747e-03, -1.1230e-02,
        -1.8465e-02, -6.5759e-03, -5.8531e-02, -8.9150e-02, -1.7125e-02,
        -9.2333e-03, -3.3530e-02, -4.8969e-03, -4.1072e-03, -1.3852e-03,
        -5.0570e-02, -4.0887e-03,  2.3785e-03, -6.2927e-02,  2.1356e-02,
        -1.3360e-03, -6.1653e-03,  4.7803e-03, -1.0487e-02, -9.9387e-03,
        -1.6811e-02, -2.0650e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0586,  0.4960,  0.0332,  0.0404,  0.0056, -0.0393, -0.0411, -0.0390,
        -0.0150, -0.0189,  0.0587, -0.0421,  0.0725, -0.0092,  0.0304,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.1137e-02, -1.6290e-02,  5.6386e-05, -6.4053e-03, -1.4888e-01,
        -1.8779e-01,  1.5010e-02, -1.3946e-01, -2.2322e-01, -2.8175e-02,
        -8.0072e-03,  1.5375e-02, -4.8634e-02, -4.8356e-03,  2.6472e-02,
        -1.0744e-02,  2.2744e-03,  7.0681e-02,  3.6868e-03,  1.2876e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0159,  0.0112,  0.0026, -0.0078, -0.1796, -0.0213,  0.0045, -0.0356,
        -0.0042,  0.0038, -0.0561, -0.0650, -0.0021, -0.0123, -0.0465, -0.0423,
        -0.0011, -0.0245,  0.0006, -0.0216, -0.0041,  0.0002, -0.0033,  0.0014,
         0.0028, -0.0424, -0.0311, -0.0007, -0.0068, -0.0558, -0.0603, -0.0054,
         0.0020, -0.0084, -0.0424, -0.0811,  0.0068, -0.0300, -0.0100,  0.0066,
         0.0017,  0.0167, -0.0213,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0218, -0.0019,  0.0152, -0.0033, -0.0586, -0.0056, -0.0445, -0.1171,
         0.0194, -0.0221, -0.0078, -0.0014, -0.0129, -0.0827, -0.1399, -0.0035,
         0.0118, -0.0013, -0.0062, -0.0066, -0.0537, -0.0054,  0.0072, -0.0070,
         0.0134,  0.0003, -0.0053,  0.0060, -0.0792, -0.0084,  0.0027, -0.0489,
        -0.0172, -0.0510, -0.0731,  0.0149,  0.0117,  0.0108,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0170, -0.2584, -0.0913, -0.1265,  0.0217, -0.0792, -0.0069,  0.0055,
        -0.0300, -0.0811,  0.0114,  0.0018, -0.0504, -0.0017,  0.0161,  0.0085,
         0.0191, -0.0612, -0.0046,  0.0304,  0.0404,  0.0368,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.0513e-03, -2.2073e-01, -4.5188e-02,  1.1585e-02,  1.0016e-02,
        -2.9592e-02, -8.1009e-03,  1.4842e-02, -1.0671e-02, -8.6362e-02,
         1.5320e-03,  8.7441e-03, -1.8531e-04, -1.3905e-03, -2.8376e-02,
        -4.3635e-02, -6.6699e-02, -5.4128e-03,  7.5593e-04, -5.4181e-02,
         1.3179e-02, -4.6882e-03, -4.1714e-02, -1.3741e-02, -6.0320e-03,
        -1.2016e-03, -6.3659e-04,  9.8681e-03,  4.3410e-03, -2.9800e-02,
         9.6906e-03, -2.2054e-02,  2.3506e-04, -3.9741e-03, -3.3289e-02,
        -9.3420e-02, -8.2140e-03, -1.1260e-02, -3.2742e-04, -4.3286e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0206, -0.0088,  0.0058,  0.0025,  0.0117, -0.0298, -0.0114, -0.0015,
        -0.0629,  0.0071, -0.0091, -0.0099, -0.0285, -0.1645, -0.1326,  0.0140,
         0.0067,  0.0070, -0.0628,  0.0109,  0.0048,  0.0116, -0.0096,  0.0084,
        -0.0203, -0.0928, -0.0061, -0.0658, -0.1107, -0.0215, -0.0220, -0.0183,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.6384e-03, -3.4384e-01, -2.7236e-03,  1.9479e-02, -2.9688e-03,
        -7.1236e-03,  5.4818e-04, -8.4746e-02, -1.2106e-01, -1.2008e-02,
        -1.0701e-02, -4.0220e-02, -9.6622e-03,  1.3708e-02,  1.0881e-02,
         2.1552e-03, -5.3985e-02, -1.0411e-02, -2.7352e-03, -4.4906e-03,
        -4.5838e-03,  2.2890e-02, -6.0909e-03, -7.6685e-02, -1.0307e-02,
        -1.0644e-02, -3.3098e-02, -1.2124e-04,  7.7837e-03, -1.3946e-02,
         4.8175e-03,  5.6636e-03,  9.2239e-04,  3.1065e-02, -1.3291e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0474, -0.0689, -0.2404, -0.0667, -0.0135,  0.0261, -0.1149, -0.1356,
        -0.0456, -0.0482, -0.0060, -0.0377,  0.0162, -0.0237, -0.0134, -0.0057,
        -0.0050, -0.0068, -0.0053,  0.0504, -0.0226,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0240, -0.4352, -0.0062, -0.0884, -0.0902, -0.0475, -0.0184, -0.0614,
        -0.0767,  0.0071, -0.0031,  0.0076,  0.0122, -0.0242, -0.0525, -0.0031,
        -0.0239, -0.0081,  0.0103,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0494,  0.0216, -0.1072,  0.0400, -0.0899, -0.0090,  0.0131,  0.0080,
        -0.0165, -0.0718, -0.1581, -0.0297, -0.0552, -0.2002, -0.0033, -0.0509,
         0.0167, -0.0299, -0.0073, -0.0221,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0185, -0.3793, -0.0017, -0.0973,  0.0022,  0.0058,  0.0409, -0.0694,
         0.0080,  0.0023, -0.0086, -0.0119,  0.0077, -0.0665, -0.0203, -0.0571,
        -0.0665, -0.0036, -0.0377, -0.0442,  0.0117, -0.0225, -0.0165,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0092, -0.1906, -0.0562, -0.0372, -0.0057, -0.0510,  0.0007, -0.0393,
        -0.1223, -0.1251, -0.0192, -0.0421,  0.0046, -0.0869,  0.0203,  0.0133,
        -0.0124,  0.0080, -0.0146,  0.0057, -0.0468, -0.0094,  0.0202, -0.0551,
         0.0040,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1046, -0.3328, -0.0066, -0.0023,  0.0089, -0.0088, -0.0020, -0.0294,
         0.0094, -0.0203, -0.0613, -0.0129,  0.0004,  0.0055,  0.0106, -0.0048,
         0.0018,  0.0038,  0.0185,  0.0107, -0.0715, -0.0729, -0.0431, -0.0276,
        -0.0160, -0.0073,  0.0059, -0.0021, -0.0099, -0.0128, -0.0202,  0.0020,
        -0.0119,  0.0262, -0.0151,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0108, -0.0023,  0.0203, -0.0820, -0.0030, -0.0398, -0.1141,  0.0072,
        -0.1034,  0.0016,  0.0310, -0.0274, -0.0053, -0.0367, -0.0643, -0.1362,
        -0.0334, -0.1784, -0.0226, -0.0372, -0.0385,  0.0045,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0715,  0.0064, -0.0230, -0.0128, -0.0137, -0.0210,  0.0031, -0.0467,
         0.0247, -0.1197,  0.0662, -0.1722,  0.0952,  0.0440, -0.1009, -0.0390,
        -0.0043,  0.0085, -0.0198, -0.0673, -0.0399,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0936,  0.4270,  0.0176,  0.0123,  0.0424,  0.0121,  0.0625,  0.0714,
        -0.0283, -0.0023, -0.0115, -0.0102, -0.0326,  0.0877, -0.0099,  0.0503,
        -0.0142,  0.0056, -0.0084,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0293,  0.0040, -0.0171, -0.0002, -0.0051, -0.0614,  0.0076, -0.0015,
         0.0058,  0.0023, -0.0207, -0.0132, -0.0044, -0.0408,  0.0003,  0.0021,
        -0.0531, -0.0676, -0.0042, -0.0566, -0.0254, -0.0143, -0.0539,  0.0074,
        -0.0733, -0.0012,  0.0015, -0.0134, -0.0067, -0.1015, -0.0053, -0.0185,
        -0.0443, -0.0723,  0.0007, -0.0049, -0.0172,  0.0036, -0.0053, -0.0507,
        -0.0072, -0.0012, -0.0032, -0.0008, -0.0099, -0.0156, -0.0434],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0142, -0.2732, -0.1715, -0.0227,  0.0016, -0.0035, -0.0049,  0.0023,
        -0.0447,  0.0020,  0.0101, -0.0092,  0.0206, -0.0023,  0.0202,  0.0125,
        -0.0640,  0.0018,  0.0046, -0.0010,  0.0262, -0.0169, -0.0844, -0.0780,
         0.0223,  0.0236,  0.0058,  0.0107, -0.0044, -0.0029, -0.0084,  0.0164,
        -0.0057, -0.0076,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0654, -0.2692, -0.0049,  0.0026,  0.0257, -0.0407, -0.0122, -0.0006,
         0.0074, -0.0485, -0.0080, -0.0029, -0.0741, -0.1066,  0.0151, -0.0034,
         0.0149,  0.0071,  0.0248, -0.0775,  0.0071, -0.0359, -0.0088, -0.0180,
        -0.0021, -0.0010, -0.0089, -0.0220, -0.0077, -0.0281,  0.0015,  0.0027,
        -0.0304, -0.0138,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0401, -0.4862, -0.0215, -0.0223, -0.0056, -0.0668,  0.0204, -0.0006,
        -0.0120, -0.0928, -0.0145, -0.0264, -0.0343, -0.1285,  0.0046, -0.0098,
        -0.0120,  0.0014,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0389,  0.0144,  0.0053, -0.0113,  0.0061,  0.0032,  0.0056,  0.0064,
         0.1826,  0.3280,  0.0508, -0.0076,  0.0211,  0.0325, -0.0230, -0.0127,
        -0.0006, -0.0216, -0.0254,  0.1201, -0.0124, -0.0170, -0.0096, -0.0048,
        -0.0056, -0.0307,  0.0026,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0574, -0.0082, -0.0041, -0.0129,  0.0109, -0.0044, -0.0037, -0.0515,
        -0.1728, -0.0135,  0.0051, -0.0512, -0.1669, -0.0412, -0.0177,  0.0166,
         0.0015,  0.0070,  0.0085, -0.2004, -0.0131,  0.0071, -0.0667, -0.0130,
         0.0123, -0.0325,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0846, -0.1939,  0.0166, -0.1095, -0.1338,  0.0020, -0.0017,  0.0320,
        -0.0186, -0.0326, -0.0710, -0.0354,  0.0118,  0.0145,  0.0162, -0.0829,
        -0.0338, -0.0227, -0.0288,  0.0374,  0.0201,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0364, -0.1879, -0.1649, -0.2010, -0.0429,  0.0284, -0.1529, -0.0047,
        -0.0285,  0.0181, -0.0137, -0.0630, -0.0176, -0.0202,  0.0198,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.2864e-01,  2.1692e-04, -6.2267e-03, -1.1892e-02,  1.2053e-02,
         2.4833e-02,  4.8686e-02, -2.8117e-02,  4.0933e-03,  2.9068e-02,
        -2.0344e-02, -6.0281e-02, -5.1379e-01, -3.2996e-02,  1.2867e-02,
        -2.7212e-02,  2.4137e-02, -1.4549e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0940, -0.2640, -0.0730, -0.0973, -0.0021, -0.0049, -0.0277, -0.0718,
        -0.0098, -0.1141, -0.0830, -0.0297,  0.0137, -0.0689,  0.0083,  0.0014,
         0.0102, -0.0259, -0.0004,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 9.7398e-03, -1.4863e-01, -1.3784e-01, -2.1629e-02, -7.2423e-03,
        -5.7450e-02,  1.6898e-02,  2.2159e-03, -1.6542e-02, -2.5259e-03,
         7.8462e-04, -3.5189e-06, -3.0706e-02,  2.0280e-04, -2.3620e-03,
        -3.1529e-03, -5.8932e-02, -3.9724e-03,  8.8526e-03,  1.0635e-02,
        -6.0135e-02,  8.2270e-04, -4.4269e-02,  6.4231e-04, -7.0833e-02,
         8.9646e-03, -4.3701e-02,  2.9746e-03, -4.4324e-03, -9.7820e-03,
         7.9778e-03,  7.7070e-03, -4.3574e-02,  1.8066e-03, -2.4233e-02,
        -2.0569e-02, -5.7690e-02, -9.3843e-06, -1.4485e-02, -9.3998e-03,
        -5.9004e-03,  4.3289e-03,  4.1326e-03, -1.1316e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.7958e-02, -2.6415e-01, -2.0219e-02, -1.2327e-01, -1.0601e-01,
        -3.2614e-03, -4.1838e-02, -2.2558e-02, -8.7852e-02, -1.2552e-03,
        -3.5034e-02,  1.2386e-03,  9.0435e-03, -5.0191e-03,  2.5170e-03,
        -1.0985e-03,  2.8406e-02, -3.0325e-02,  4.0797e-03, -1.9064e-02,
        -2.8278e-02, -3.2115e-03, -4.0057e-03, -1.6054e-03,  5.8600e-03,
         5.3550e-03, -4.0639e-03, -5.2965e-02,  4.4858e-03, -3.2729e-03,
         1.6430e-04, -2.0239e-02,  1.4964e-02, -5.1390e-03,  2.1864e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.8752e-02, -4.6203e-01, -4.6877e-03, -7.6830e-03, -1.9311e-02,
         1.3268e-02, -6.6338e-02, -5.3275e-02, -2.3661e-02,  2.8971e-03,
         6.7708e-03, -4.5123e-02, -1.1787e-01,  4.3111e-06,  4.5141e-04,
        -3.6176e-02,  2.4366e-03, -7.0415e-03,  1.9544e-02,  6.0195e-03,
         2.7953e-03, -5.7644e-03,  6.4846e-02,  1.3253e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0224, -0.2055, -0.0004, -0.0418, -0.0105, -0.0364,  0.0227, -0.0137,
        -0.1169, -0.0099, -0.0643,  0.0136,  0.0161, -0.0208, -0.0607,  0.0043,
         0.0064, -0.0109, -0.0696, -0.0086, -0.0099, -0.0756, -0.0036, -0.0430,
        -0.0156, -0.0311, -0.0087, -0.0203,  0.0014, -0.0020, -0.0127,  0.0047,
         0.0030, -0.0130,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0492,  0.0228,  0.0041, -0.0151,  0.0179,  0.0225, -0.1383, -0.0126,
         0.0112,  0.0050, -0.0040, -0.0118, -0.0912, -0.0158, -0.1354, -0.1523,
        -0.0256,  0.0119, -0.0068, -0.0161,  0.0069, -0.0091,  0.0104, -0.0890,
        -0.0410, -0.0123, -0.0404,  0.0007,  0.0006, -0.0199,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0638,  0.0494, -0.0116, -0.2874, -0.0381,  0.0109, -0.0205, -0.0287,
        -0.1341, -0.2036,  0.0181,  0.0780, -0.0558,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1466,  0.0177, -0.0068,  0.0209, -0.0031, -0.0331,  0.0103, -0.1452,
        -0.1689,  0.0094, -0.0259, -0.0937,  0.0336, -0.0468, -0.1785, -0.0073,
         0.0102,  0.0421,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0057, -0.2326, -0.4929, -0.0423, -0.1721, -0.0027, -0.0365, -0.0152,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.3644e-02, -2.2490e-01, -9.8791e-03, -4.5960e-02, -4.5040e-02,
        -8.4150e-03, -2.2431e-02, -7.0674e-03, -4.3018e-03, -1.3254e-02,
         6.7979e-03,  1.9679e-03, -1.0415e-02, -8.3263e-03, -1.5908e-03,
        -6.9599e-03,  3.1938e-03,  6.4139e-03,  4.7537e-03, -3.4222e-04,
         1.0849e-02, -6.5479e-05, -3.8404e-02, -4.1299e-02, -5.9959e-03,
        -2.1021e-02, -1.3581e-03, -5.3312e-03, -8.2999e-02, -2.8502e-02,
         3.5253e-03, -4.9934e-02, -9.5622e-03, -9.9167e-02, -5.3304e-03,
        -4.5564e-02, -3.1474e-02,  1.7644e-03,  1.7342e-02,  3.2970e-03,
        -1.5934e-02, -8.1060e-03, -7.7504e-03,  5.5281e-04,  5.4657e-03,
        -4.1769e-03,  9.5731e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0032, -0.2855, -0.0622, -0.4927, -0.0230, -0.0128, -0.0387, -0.0188,
        -0.0067, -0.0486, -0.0078,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0326, -0.2689, -0.0232, -0.1323, -0.0285, -0.0061, -0.0496, -0.2118,
         0.0013, -0.0766, -0.0292,  0.0279, -0.0593,  0.0008, -0.0092, -0.0104,
        -0.0185, -0.0138,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0677, -0.1300, -0.4424,  0.0292, -0.0161,  0.0131, -0.1190,  0.0038,
        -0.0763, -0.0113, -0.0147,  0.0093,  0.0169, -0.0260, -0.0240,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1241, -0.3390,  0.0090,  0.0094,  0.0273, -0.0711, -0.1356, -0.0131,
         0.0373, -0.0839, -0.0819,  0.0105, -0.0007,  0.0341,  0.0230,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0423,  0.0174,  0.0081, -0.0008,  0.0222,  0.0240, -0.0054, -0.0088,
        -0.1136, -0.0292, -0.0205, -0.1607, -0.1615, -0.0287,  0.0043, -0.0008,
        -0.0258, -0.0067, -0.0145,  0.0092, -0.0872, -0.1371,  0.0345, -0.0109,
        -0.0111, -0.0146,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0075, -0.0160, -0.0202, -0.0390, -0.0081, -0.0146,  0.0104,  0.2790,
         0.3937,  0.0481, -0.0053, -0.0031, -0.0011, -0.1266,  0.0275,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.3457e-02, -2.9982e-01, -8.4080e-03, -6.9945e-02, -4.9638e-02,
        -3.0600e-02, -9.1157e-02, -7.9857e-03,  1.4673e-03, -1.2630e-01,
         4.7515e-03, -7.9422e-03, -4.8116e-02, -5.3106e-03, -1.6658e-02,
        -8.4621e-03, -2.4958e-02, -1.3746e-04, -4.0157e-02, -1.6848e-02,
        -3.9403e-02,  7.9699e-03,  9.0106e-03,  1.2482e-02, -2.1952e-02,
        -1.7059e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0408,  0.0038,  0.0024,  0.0096,  0.0137, -0.0198, -0.0028,  0.0197,
        -0.0400, -0.0438, -0.1682, -0.0025,  0.0150,  0.0674, -0.1640, -0.0851,
         0.0025, -0.0341,  0.0013, -0.0774, -0.0285, -0.0371, -0.0125, -0.0704,
        -0.0067, -0.0309,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-5.7061e-02, -3.2784e-01,  6.3201e-03,  1.5000e-03, -2.0681e-02,
        -1.7256e-02, -4.0665e-02, -4.1275e-02,  2.0730e-03, -2.2815e-02,
        -1.5574e-02, -2.9131e-02, -7.2280e-02,  6.7946e-03, -2.2889e-02,
        -1.3249e-02,  4.5223e-03, -2.0478e-03, -2.8024e-02, -6.1664e-03,
         1.5849e-02, -1.5453e-02,  5.1941e-03, -2.7554e-02, -7.1218e-02,
         8.8965e-03, -1.9445e-02, -8.3351e-03,  2.5369e-02, -8.8924e-04,
        -8.8877e-03, -2.6955e-02, -1.0473e-02,  2.4447e-06, -7.0396e-03,
         1.0279e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0469, -0.0866, -0.0248, -0.0057, -0.0670, -0.0085, -0.0674, -0.0719,
         0.0128,  0.0014, -0.0438, -0.0031, -0.0005,  0.0003, -0.0059, -0.0195,
         0.0005, -0.0013,  0.0025,  0.0099,  0.0002, -0.0707, -0.0125, -0.0021,
        -0.0530, -0.0635, -0.0009, -0.0552, -0.0563, -0.0032, -0.0117, -0.0548,
        -0.0810,  0.0167, -0.0138,  0.0239,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0408,  0.0394, -0.0041, -0.0108, -0.0007, -0.2917, -0.0415, -0.0061,
        -0.0145,  0.0416, -0.0215, -0.0880, -0.2756,  0.0435,  0.0256, -0.0208,
        -0.0016,  0.0086,  0.0166,  0.0072,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0172, -0.1242, -0.0450, -0.0151,  0.0095,  0.0024,  0.0056, -0.0773,
        -0.0172, -0.0366, -0.0069,  0.0050, -0.0066, -0.0067, -0.0844, -0.1093,
        -0.0175, -0.0700, -0.0002,  0.0038, -0.0175,  0.0053,  0.0044, -0.0085,
        -0.0062,  0.0113,  0.0018,  0.0013, -0.0013, -0.0498, -0.0913, -0.0248,
         0.0006, -0.0411,  0.0011,  0.0461, -0.0032,  0.0239,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0009, -0.1681, -0.0890, -0.0785, -0.0066, -0.0025, -0.0211, -0.0035,
        -0.0180,  0.0124, -0.0609, -0.1118, -0.0048, -0.0073, -0.0130, -0.0730,
         0.0122, -0.0629,  0.0020, -0.0022,  0.0011,  0.0033,  0.0030, -0.0022,
        -0.0007,  0.0026, -0.0147, -0.0373, -0.0597,  0.0029, -0.0068, -0.0178,
        -0.0037, -0.0030, -0.0216,  0.0116, -0.0211,  0.0011,  0.0027, -0.0181,
        -0.0145], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0366, -0.4072,  0.0179, -0.0071, -0.0034, -0.0960, -0.0136,  0.0182,
        -0.0287,  0.0080, -0.0011, -0.0657, -0.0928,  0.0030,  0.0067, -0.0515,
        -0.0061, -0.0818, -0.0057, -0.0146,  0.0163, -0.0179,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0784,  0.0091,  0.0056,  0.0048,  0.0087, -0.0488,  0.0006,  0.0083,
        -0.0200,  0.0020,  0.0045, -0.0654,  0.0101, -0.0587, -0.0003, -0.1171,
        -0.0116, -0.0500, -0.0013, -0.0481, -0.1052, -0.0037, -0.0054, -0.0293,
         0.0031, -0.0044, -0.0013, -0.0466, -0.0912,  0.0062, -0.0451, -0.0418,
         0.0173, -0.0019, -0.0441,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0725, -0.0168, -0.1480, -0.0374, -0.0966,  0.0046,  0.0065,  0.0013,
        -0.0265,  0.0039, -0.0421,  0.0090,  0.0119,  0.0023, -0.0064, -0.0284,
         0.0259,  0.0119, -0.0793, -0.0291, -0.0822,  0.0407,  0.0017, -0.0590,
        -0.0090, -0.0363, -0.0814, -0.0068, -0.0226,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0595, -0.2688, -0.0451, -0.0440,  0.0168, -0.0467,  0.0268, -0.0163,
        -0.0397, -0.0029, -0.0458, -0.0577,  0.0127, -0.0099, -0.0012, -0.0873,
        -0.0374, -0.0227, -0.0492,  0.0092, -0.0075, -0.0092, -0.0261,  0.0014,
         0.0559,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0446, -0.0201, -0.0308,  0.0133, -0.2176, -0.0373, -0.0124, -0.1017,
        -0.1046, -0.0041,  0.0225, -0.0016,  0.0045, -0.0202, -0.0838, -0.0977,
         0.0089, -0.0676, -0.0164, -0.0229, -0.0079,  0.0305,  0.0290,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.9935e-02, -3.8497e-04, -3.1285e-03,  4.6334e-03,  1.2028e-02,
         1.4393e-02, -5.4356e-04, -2.4791e-03,  8.0426e-04, -5.5417e-03,
        -4.7749e-03, -5.3913e-03, -8.2594e-02, -1.3627e-01, -1.0712e-04,
        -1.1237e-02, -1.7610e-03, -4.0213e-02,  9.9920e-03,  4.4039e-03,
         7.7235e-03,  1.3646e-02,  1.4824e-02, -1.0456e-02, -3.3430e-02,
        -2.8262e-02, -1.7404e-01, -2.6729e-02, -1.4036e-01,  2.3458e-02,
        -2.7370e-02,  1.2958e-02,  8.3912e-03,  5.2210e-03, -2.4455e-02,
         9.2336e-04, -1.4950e-02, -5.5602e-03, -2.6628e-02,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0027,  0.0003,  0.0319, -0.1776, -0.2304,  0.0648,  0.0286, -0.1088,
        -0.2378,  0.0868, -0.0213,  0.0091,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0414, -0.1653, -0.2449, -0.0057, -0.0040, -0.0480, -0.0327,  0.0260,
        -0.0007, -0.0251,  0.0010, -0.0069,  0.0034, -0.0052, -0.0379, -0.0014,
        -0.0238, -0.0464,  0.0006, -0.0159, -0.0245, -0.0218,  0.0085, -0.0091,
         0.0068,  0.0039, -0.0126,  0.0166, -0.0078,  0.0145,  0.0032,  0.0066,
        -0.0196, -0.0044, -0.0063, -0.0200, -0.0444, -0.0071, -0.0011, -0.0031,
         0.0025, -0.0008,  0.0042,  0.0113, -0.0030,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0007, -0.3889, -0.0653, -0.1477,  0.0025, -0.0346,  0.0243, -0.0041,
        -0.1629,  0.0856, -0.0574,  0.0260,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0396, -0.5570, -0.0299, -0.2077, -0.0465, -0.0379,  0.0087,  0.0727,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.0472e-02, -2.2084e-01, -2.4035e-02, -5.3598e-02, -6.8859e-02,
        -1.2036e-02,  1.7136e-02, -1.4695e-02, -5.1960e-02, -6.8443e-02,
         3.8374e-03,  1.4194e-03,  3.7894e-03,  3.0212e-03,  2.9892e-03,
         3.0371e-03,  7.3398e-04,  7.4728e-03, -1.4254e-04, -4.0442e-04,
         4.6654e-04,  5.4822e-03, -1.3974e-03, -9.8950e-03, -6.8151e-04,
         3.1612e-03, -1.3650e-03, -6.0111e-03, -1.9660e-02, -5.1587e-03,
         3.1303e-03, -3.2599e-03,  3.5909e-03,  2.3177e-05, -2.4281e-03,
         2.2783e-03,  2.1813e-03, -2.4464e-03,  3.8836e-03,  8.2566e-04,
        -1.1980e-03, -1.0521e-03, -5.1007e-03, -7.4518e-04,  5.7753e-03,
         5.6988e-03, -7.4393e-02, -6.1184e-03, -4.1853e-02, -1.1347e-01,
        -7.5317e-03, -4.1345e-03, -4.3951e-02,  1.1846e-04, -1.3157e-02,
        -2.3411e-03, -9.8829e-03, -4.4527e-03,  3.3517e-03,  8.2731e-03,
        -1.1441e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0189,  0.0037,  0.0102, -0.0488, -0.0669,  0.0114, -0.0554,  0.0039,
         0.0017,  0.0200,  0.0163, -0.0720,  0.0020, -0.0568, -0.1130, -0.0016,
         0.0030,  0.0126, -0.0554, -0.0277, -0.0894, -0.0448, -0.0164,  0.0446,
        -0.0251, -0.0316, -0.0127, -0.0035, -0.0348, -0.0275, -0.0103,  0.0370,
        -0.0210,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0089,  0.0366,  0.2018,  0.1037, -0.0290, -0.0065,  0.0793, -0.0067,
         0.0083, -0.0104,  0.0043,  0.0041,  0.0405,  0.0037,  0.0158, -0.0197,
         0.0422,  0.0900,  0.1740,  0.0067, -0.0035,  0.0273, -0.0354,  0.0415,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0022, -0.0953, -0.1616, -0.0559,  0.0261, -0.1029,  0.0108, -0.0465,
        -0.0205, -0.0833, -0.0648, -0.0907, -0.1300, -0.0141,  0.0032,  0.0030,
        -0.0166,  0.0047,  0.0145, -0.0053, -0.0248, -0.0077, -0.0140,  0.0017,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0093,  0.0021, -0.1131, -0.0909, -0.0044, -0.0002,  0.0064, -0.0666,
        -0.0119,  0.0155,  0.0077,  0.0007, -0.0022,  0.0071, -0.0560, -0.1163,
        -0.0787, -0.0116, -0.0098,  0.0219, -0.0686,  0.0036, -0.0482, -0.0692,
         0.0043,  0.0064, -0.0390, -0.0063, -0.0011, -0.0134, -0.0016, -0.0176,
         0.0616,  0.0265,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0221,  0.0062, -0.0136,  0.0120, -0.0771, -0.0077, -0.0088,  0.0070,
         0.0038,  0.0016,  0.0090,  0.0462,  0.0131, -0.1732,  0.0056,  0.0037,
        -0.1900,  0.0048, -0.1105, -0.0143, -0.1219, -0.1148,  0.0126, -0.0099,
        -0.0107,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0934, -0.0148, -0.0152, -0.0364, -0.0342, -0.0046, -0.1396, -0.0701,
        -0.1037, -0.2958, -0.0979, -0.0195, -0.0454, -0.0052, -0.0180, -0.0062,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0212, -0.2366, -0.0087, -0.1006, -0.0044, -0.0481, -0.2184, -0.0253,
         0.0107,  0.0105,  0.0009,  0.0040, -0.0260,  0.0074, -0.0032,  0.0182,
        -0.0415, -0.0788, -0.0502, -0.0047, -0.0021,  0.0478,  0.0308,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0408,  0.0304,  0.0285,  0.1948,  0.4063, -0.0111, -0.0512,  0.0048,
         0.0208, -0.0530, -0.0203, -0.0809, -0.0228, -0.0346,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0683, -0.1219, -0.1200, -0.1437,  0.0082, -0.0314, -0.0299,  0.0134,
        -0.0213, -0.0044, -0.0737, -0.0952, -0.0037, -0.0042, -0.0044, -0.0023,
         0.0143, -0.1031,  0.0040, -0.0595, -0.0008,  0.0105,  0.0151,  0.0470,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1214, -0.0775, -0.1734, -0.0194, -0.1895,  0.0256, -0.3318,  0.0260,
         0.0025, -0.0173, -0.0155,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0769,  0.0458,  0.0395,  0.0044, -0.0018, -0.0745, -0.0772,  0.0157,
        -0.1078, -0.0118, -0.0873,  0.0130, -0.0623, -0.0040, -0.0688, -0.0341,
        -0.0502, -0.1194, -0.0029, -0.0489, -0.0536,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0591,  0.0434,  0.0105, -0.0274,  0.0044, -0.0116,  0.0397, -0.2993,
        -0.1244, -0.2109, -0.0612,  0.0044,  0.0354, -0.0684,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0127,  0.0341,  0.0158, -0.0270, -0.0154,  0.0106,  0.0209, -0.0166,
        -0.0010, -0.0136,  0.1125,  0.0828,  0.4231,  0.0202,  0.0653,  0.0307,
         0.0250, -0.0153, -0.0091, -0.0224, -0.0260,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0269, -0.3255, -0.0011, -0.1177, -0.0216, -0.0972, -0.0254, -0.1910,
        -0.0278, -0.0458, -0.0496,  0.0394, -0.0310,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0309, -0.1484,  0.0054, -0.0032, -0.0036, -0.0043, -0.0267, -0.0734,
        -0.0162, -0.0009,  0.0083,  0.0005, -0.0044, -0.0034, -0.0027, -0.0274,
         0.0027,  0.0016,  0.0040, -0.0367,  0.0021,  0.0157,  0.0008, -0.0160,
        -0.1270,  0.0146, -0.0954, -0.0866, -0.0866,  0.0127, -0.0091, -0.0216,
        -0.0020, -0.0022,  0.0086,  0.0070, -0.0263,  0.0035,  0.0024, -0.0156,
        -0.0161,  0.0015,  0.0073,  0.0068, -0.0078], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0420, -0.0487, -0.1033, -0.0164, -0.0002,  0.0010, -0.0024, -0.0225,
        -0.0037, -0.0027, -0.0634, -0.0138, -0.0312, -0.0119, -0.0634, -0.0120,
        -0.1303, -0.0019, -0.0121,  0.0092, -0.0007, -0.0714, -0.0119, -0.0618,
         0.0043,  0.0049, -0.0091, -0.0048, -0.1022, -0.0192, -0.0228, -0.0182,
        -0.0535, -0.0230,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.9656e-04,  6.0825e-02,  1.7287e-02, -1.6318e-02,  2.4222e-04,
         4.9209e-03, -1.1012e-02,  8.8785e-03,  2.1058e-02,  6.1724e-03,
        -2.8046e-02, -1.4761e-02, -4.6032e-03,  2.8560e-03,  1.3813e-02,
         1.7731e-02,  3.3162e-03,  2.8950e-02, -2.5181e-02, -1.9086e-01,
        -3.7391e-01,  5.8891e-02,  1.6361e-02,  2.9360e-02,  3.0974e-03,
        -4.0755e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4832e-02, -4.0182e-01,  2.2877e-02, -1.2923e-01,  2.5541e-02,
        -2.4085e-03, -4.1691e-03, -6.7973e-02,  4.7147e-03, -1.4357e-02,
         1.8068e-02,  2.2830e-04,  6.1176e-03,  5.6054e-03, -1.3098e-02,
        -6.4139e-02, -9.9765e-02, -5.9719e-03, -4.1299e-02,  5.6674e-03,
        -2.8235e-02,  4.8414e-03,  1.9046e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.4792e-02, -1.3196e-01, -1.6331e-02, -5.3459e-02, -7.6036e-03,
        -4.9657e-03, -8.5601e-03, -5.4454e-02, -8.5453e-02, -6.6492e-03,
         1.5551e-04, -3.0726e-03,  1.9790e-03, -4.4614e-03, -1.8454e-01,
         1.8935e-02,  8.0512e-03, -2.8113e-02, -4.3588e-02, -1.9477e-02,
        -1.3110e-03, -3.8524e-03,  2.4651e-03,  1.2687e-02, -6.8517e-04,
        -3.7521e-03, -5.7625e-02, -7.1295e-02, -7.6735e-03, -1.4673e-02,
         1.1618e-03, -2.5643e-03, -1.0997e-02,  3.0216e-03, -4.8532e-03,
         4.7887e-03, -3.7929e-02,  1.2034e-02, -1.3363e-02,  6.6635e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0295, -0.0406, -0.0015, -0.0331, -0.0680,  0.0036, -0.0021, -0.1011,
         0.0080, -0.0093,  0.0011, -0.0041, -0.0380,  0.0101, -0.0318, -0.1019,
        -0.0931, -0.0252, -0.0230, -0.0562, -0.0075, -0.0876, -0.0576,  0.0159,
        -0.0074,  0.0016, -0.0076, -0.0624, -0.0034, -0.0157, -0.0275, -0.0051,
        -0.0179, -0.0015,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-6.2577e-03, -1.3834e-02, -5.7121e-02, -4.8943e-02, -9.6469e-02,
        -1.8407e-03,  2.2093e-02,  1.4460e-02, -2.3224e-03,  2.4424e-03,
        -4.2837e-03, -1.4512e-02,  2.2554e-03, -5.5384e-03, -7.7043e-02,
        -1.7854e-02, -1.9043e-03, -7.7670e-03, -5.4339e-02, -1.0827e-03,
         3.4235e-03,  2.0173e-03, -5.1015e-03,  1.2111e-02, -3.0698e-02,
        -6.0988e-02,  1.2526e-02,  1.4139e-02,  2.0984e-03,  1.2653e-02,
         1.3397e-02,  3.5953e-03, -1.6079e-03,  4.8192e-03, -3.1035e-02,
         1.5792e-03, -2.3984e-02, -4.4179e-02, -1.2012e-02, -4.9338e-02,
        -3.8476e-02, -2.4186e-02, -4.2928e-02, -6.0501e-03,  7.7523e-03,
         2.7611e-05, -3.5748e-03, -2.2414e-03, -1.6396e-02, -1.1386e-03,
        -5.6553e-02, -7.0086e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0298, -0.0199, -0.0137, -0.0079, -0.0970,  0.0208, -0.0245,  0.0389,
         0.0036, -0.0300,  0.0039,  0.0102, -0.0142, -0.1285, -0.0383, -0.0609,
        -0.1051, -0.0395, -0.0698, -0.0147, -0.0206, -0.0318, -0.0052, -0.0545,
        -0.0018,  0.0089, -0.0202, -0.0005,  0.0295, -0.0095,  0.0463,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0144, -0.2351,  0.0027, -0.0160,  0.0109,  0.0130, -0.0060,  0.0085,
        -0.0115,  0.0346, -0.0142, -0.0675,  0.0015, -0.0054, -0.0684, -0.0737,
        -0.0198, -0.0408, -0.1244, -0.1215, -0.0409, -0.0280, -0.0167, -0.0101,
         0.0145,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0200, -0.0185, -0.0062, -0.0049,  0.0351, -0.2819, -0.4316, -0.0861,
         0.0102, -0.0262,  0.0323, -0.0471,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0387,  0.3370,  0.0342,  0.0603,  0.0869,  0.0195, -0.0127,  0.0010,
         0.0143, -0.0008,  0.0120, -0.0051,  0.0353,  0.0823, -0.0020,  0.0561,
         0.0178,  0.0143, -0.0079, -0.0012,  0.0256,  0.0049,  0.0184,  0.0117,
         0.0094,  0.0415, -0.0128,  0.0093, -0.0269,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0791, -0.4666,  0.0030, -0.0194,  0.0288, -0.1144, -0.2341,  0.0154,
         0.0026, -0.0100,  0.0140, -0.0127,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0284, -0.4146,  0.0069, -0.1195,  0.0168, -0.1057, -0.2087,  0.0047,
        -0.0547,  0.0032,  0.0069,  0.0299,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0144,  0.0837, -0.2650, -0.0412, -0.1098, -0.0039, -0.0029,  0.0284,
        -0.1010,  0.0196, -0.0009, -0.1365, -0.0061,  0.0072,  0.0162, -0.1153,
        -0.0082,  0.0340,  0.0044,  0.0014,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0117, -0.2902, -0.0120, -0.1032, -0.1298,  0.0055, -0.0493, -0.0040,
        -0.0393, -0.0134, -0.0108, -0.0589, -0.0239, -0.0059, -0.0032,  0.0020,
        -0.0214,  0.0145,  0.0114,  0.0027, -0.0079,  0.0252,  0.0038,  0.0026,
        -0.0094,  0.0030, -0.0094, -0.0225, -0.0209,  0.0068,  0.0005, -0.0018,
        -0.0154, -0.0226, -0.0134,  0.0140, -0.0080,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1679,  0.6178, -0.0246, -0.0735, -0.0307,  0.0360, -0.0112, -0.0017,
         0.0365,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0023,  0.0070, -0.0086, -0.1737,  0.0059, -0.0159, -0.0565, -0.0600,
         0.0133, -0.0044, -0.0060, -0.0012, -0.1364, -0.1246, -0.0100,  0.0200,
         0.0011,  0.0210, -0.0775, -0.0607,  0.0276, -0.0063, -0.0057, -0.0065,
         0.0022,  0.0755, -0.0700,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0127, -0.2871,  0.0210, -0.0829, -0.0013, -0.0419, -0.0045,  0.0072,
         0.0043,  0.0057, -0.0044, -0.0567, -0.0129,  0.0068, -0.0112,  0.0007,
        -0.0755, -0.0080,  0.0125,  0.0044,  0.0075, -0.0276, -0.0891,  0.0089,
        -0.0072, -0.0156, -0.0116, -0.0256, -0.0787, -0.0062, -0.0026,  0.0108,
        -0.0065, -0.0404,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([ 4.9267e-02,  2.8281e-03, -3.6846e-03, -2.8861e-03, -4.7821e-02,
        -7.0218e-04,  8.3918e-03,  5.1259e-03,  4.0722e-03,  2.2474e-02,
        -3.4361e-04, -7.7548e-02, -7.3840e-05, -3.5497e-03, -1.3491e-02,
        -9.7773e-02, -1.1932e-01,  8.7200e-03, -5.0680e-02,  5.1469e-03,
         1.6414e-02, -6.8225e-03, -3.3508e-02,  9.3521e-04,  9.5382e-03,
        -6.2689e-03,  2.8280e-04, -1.1913e-02,  5.4929e-03, -4.5533e-03,
        -8.6898e-03, -6.3283e-02, -1.1336e-01,  3.6501e-03, -1.4231e-02,
        -2.1093e-02, -3.7336e-02, -3.3439e-03,  1.4302e-02, -6.0578e-03,
        -2.5108e-02, -3.4383e-02,  7.0281e-03, -1.0840e-02, -1.7660e-02,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0741,  0.0098, -0.0222, -0.0480, -0.0331, -0.1689, -0.2646, -0.0221,
        -0.0142, -0.1631,  0.0196, -0.0089, -0.1011, -0.0096, -0.0116, -0.0290,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0850, -0.0044,  0.0041,  0.0138,  0.0094, -0.0125,  0.0395,  0.0497,
         0.1620,  0.1461,  0.0176,  0.0980,  0.0135,  0.0638,  0.1473, -0.0076,
         0.0119,  0.0055, -0.0123,  0.0102,  0.0129, -0.0449, -0.0010,  0.0056,
        -0.0041, -0.0123,  0.0051,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0195, -0.0020, -0.0585,  0.0012, -0.0403, -0.0092, -0.0296, -0.0122,
        -0.0176, -0.0012,  0.0023, -0.0046, -0.0404, -0.0022, -0.0199, -0.0059,
        -0.0276,  0.0014,  0.0031, -0.0233,  0.0150, -0.0180, -0.0181, -0.0720,
        -0.0021,  0.0005, -0.0335, -0.0719, -0.0048, -0.0366, -0.0045, -0.0441,
        -0.0657,  0.0012,  0.0189, -0.0585, -0.0253, -0.0652, -0.0188, -0.0023,
        -0.0292, -0.0421, -0.0128,  0.0092,  0.0050, -0.0028], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0168,  0.0266, -0.0079, -0.0075, -0.0159, -0.0373, -0.0022, -0.1324,
         0.0241,  0.0118, -0.0187,  0.0084,  0.0002,  0.0012, -0.0289,  0.0026,
         0.0056, -0.0325, -0.1947, -0.2251, -0.0284, -0.0195, -0.0037, -0.0127,
         0.0148,  0.0032, -0.0787,  0.0165, -0.0188, -0.0033,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1580, -0.0970,  0.0145,  0.0156,  0.0180,  0.0467,  0.0461,  0.4846,
        -0.0191,  0.0473,  0.0533,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0704, -0.3074,  0.0212, -0.0246, -0.0605, -0.0026, -0.0917, -0.1257,
         0.0026, -0.0132,  0.0131, -0.0004, -0.0093, -0.0224, -0.0208, -0.0922,
        -0.0281, -0.0326, -0.0074, -0.0538,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 6.2258e-02, -4.6275e-01, -1.8081e-01, -2.3546e-02,  4.5181e-02,
        -8.3685e-02, -9.8215e-02,  2.0665e-02,  5.0867e-03, -1.7363e-02,
         4.3851e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0418, -0.3850, -0.1295, -0.1226,  0.0134, -0.0135, -0.0025, -0.0588,
        -0.0947,  0.0055,  0.0036, -0.0048,  0.0140, -0.0110,  0.0012, -0.0568,
         0.0104,  0.0077, -0.0110, -0.0024,  0.0097,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0542, -0.2374, -0.1593,  0.0040, -0.0393, -0.2112,  0.0098, -0.0350,
        -0.0306, -0.0362, -0.0076, -0.0656, -0.0005, -0.0097, -0.0202,  0.0024,
        -0.0195,  0.0572,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0112, -0.1399, -0.1624, -0.0098, -0.0254, -0.0101,  0.0108,  0.0034,
        -0.0018, -0.0171,  0.0072,  0.0042, -0.0050, -0.0496, -0.0650,  0.0107,
        -0.0334, -0.0711, -0.0528,  0.0004, -0.0197, -0.0481, -0.0313, -0.0008,
        -0.0509, -0.0041, -0.0257, -0.0007, -0.0158, -0.0305,  0.0110,  0.0051,
        -0.0049,  0.0100, -0.0117, -0.0012,  0.0032,  0.0222, -0.0117,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0521, -0.3655, -0.0087, -0.1015, -0.0221,  0.0014,  0.0047,  0.0228,
        -0.0429, -0.0085, -0.0118, -0.0759, -0.0186,  0.0006,  0.0012,  0.0057,
        -0.0841, -0.0099,  0.0014,  0.0058, -0.0063, -0.0011,  0.0033, -0.0011,
         0.0056,  0.0037, -0.0252, -0.0022, -0.0670, -0.0232,  0.0040,  0.0012,
        -0.0050, -0.0059,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.0601, -0.0434, -0.2230, -0.4202, -0.0038, -0.1559,  0.0269, -0.0608,
        -0.0059,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0204, -0.1991, -0.0132, -0.0682, -0.0070, -0.0975, -0.0072, -0.0328,
        -0.0465,  0.0025, -0.0479, -0.0234, -0.0108, -0.0015,  0.0083, -0.0617,
        -0.0045, -0.1133, -0.0145, -0.0107, -0.0014,  0.0064,  0.0002,  0.0093,
        -0.0283, -0.0015,  0.0061, -0.0039, -0.0061, -0.0074, -0.0117, -0.0352,
         0.0118, -0.0430,  0.0142, -0.0026, -0.0042,  0.0046,  0.0034,  0.0076,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0395, -0.1158, -0.1426,  0.0031, -0.0610, -0.0095, -0.0066, -0.0063,
        -0.0389, -0.0668,  0.0044, -0.0030,  0.0023,  0.0015, -0.0019,  0.0045,
         0.0033,  0.0012, -0.0460,  0.0041, -0.0509, -0.0118,  0.0024, -0.0091,
         0.0073, -0.0125, -0.0417,  0.0034, -0.0301, -0.0889,  0.0021, -0.0098,
        -0.0013,  0.0030,  0.0066, -0.0044, -0.0406, -0.0588,  0.0087, -0.0018,
        -0.0032, -0.0061,  0.0043, -0.0003,  0.0009, -0.0007, -0.0069, -0.0042,
        -0.0012,  0.0005, -0.0018,  0.0127,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0196, -0.2515, -0.0045, -0.0660, -0.0196, -0.0275, -0.0844, -0.0237,
        -0.0621,  0.0045,  0.0311, -0.0180, -0.0623,  0.0021, -0.0476, -0.0619,
        -0.0608,  0.0026, -0.1176,  0.0062, -0.0045, -0.0217,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.8460e-02, -5.8561e-01, -1.6989e-02,  2.9226e-03, -1.3126e-02,
        -4.9226e-03,  2.5239e-02, -4.0925e-02, -3.3125e-03, -1.8501e-02,
        -1.1730e-02, -6.7698e-02, -6.1603e-03,  2.4671e-03, -3.3689e-03,
        -8.5576e-02, -3.9539e-03, -2.1968e-02, -3.5899e-04, -9.3564e-03,
         7.3578e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.6905e-03, -1.7046e-01, -3.6511e-02, -2.2625e-02, -7.3776e-03,
         6.4864e-03, -2.6785e-02,  1.1534e-02, -2.8348e-02, -1.3993e-03,
         9.3994e-03, -1.5063e-02,  3.0590e-03,  6.2471e-03,  2.6885e-03,
        -5.2729e-04,  7.2532e-03, -1.5988e-03, -2.3228e-02, -6.9568e-04,
         5.6857e-03,  6.6204e-03, -3.3493e-03,  2.1273e-03,  2.4436e-03,
         2.9369e-03, -1.5248e-02, -5.4456e-02,  2.1207e-04, -4.5186e-02,
        -7.7899e-02, -5.1531e-02, -2.3115e-03, -3.6723e-02, -3.0119e-02,
         1.0600e-03, -6.6394e-03, -5.1518e-03,  6.1052e-03, -9.9085e-03,
        -7.8008e-02,  9.6614e-04, -4.0133e-03, -3.2908e-04, -2.9162e-02,
         1.0342e-04, -2.9851e-02, -2.4560e-02,  8.2406e-03,  8.8263e-03,
        -2.2544e-02, -1.6064e-03, -1.6376e-02, -1.1913e-02, -1.3811e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0221,  0.0029,  0.0210, -0.0202, -0.0107, -0.0075, -0.0595,  0.0117,
         0.0088,  0.0044,  0.0080,  0.0083, -0.0073, -0.0264, -0.0554, -0.0187,
         0.0035, -0.0379,  0.0054, -0.0826, -0.0076, -0.0287, -0.0205,  0.0025,
         0.0067,  0.0063,  0.0050, -0.0796,  0.0163,  0.0123,  0.0171,  0.0083,
        -0.0184, -0.0787,  0.0011, -0.0505, -0.0028, -0.0714, -0.0460, -0.0560,
        -0.0007, -0.0228,  0.0075, -0.0085, -0.0026,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0133, -0.1563, -0.0841, -0.0956,  0.0084, -0.0126, -0.0214,  0.0150,
        -0.0572, -0.0053,  0.0061, -0.0297,  0.0093, -0.1156, -0.0947, -0.0025,
        -0.0205, -0.0994, -0.0007, -0.0593, -0.0690,  0.0050,  0.0183,  0.0007,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0409,  0.0112,  0.0043,  0.0030, -0.0178,  0.0061, -0.0417, -0.0730,
        -0.0008,  0.0014,  0.0037, -0.0071, -0.0097, -0.0794,  0.0065,  0.0074,
        -0.0605, -0.0532, -0.0575, -0.0719,  0.0070,  0.0037, -0.0487, -0.0849,
        -0.0048, -0.0386, -0.0723,  0.0007, -0.0434, -0.0972, -0.0072, -0.0020,
        -0.0325,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.4327e-03, -4.7292e-01,  1.1094e-03, -7.6624e-02,  2.8271e-03,
         1.2725e-02,  2.2460e-04, -2.4783e-02, -1.4878e-01,  9.5398e-03,
         1.6800e-02,  4.8087e-03,  1.9102e-02, -1.1790e-01, -1.9702e-02,
        -1.9779e-02, -2.2106e-02,  6.9072e-03,  1.7923e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0581, -0.0146,  0.0287, -0.0238, -0.0946,  0.0112, -0.1128, -0.0765,
        -0.1417, -0.0077, -0.0034,  0.0111,  0.0088,  0.0203,  0.0053,  0.0082,
        -0.0036, -0.0831, -0.0619, -0.0250, -0.0461, -0.0955, -0.0216, -0.0012,
        -0.0354,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0373, -0.2533, -0.1014,  0.0070, -0.0006, -0.1365, -0.2105,  0.0551,
        -0.0250, -0.0512, -0.0063, -0.0124, -0.0597, -0.0062, -0.0131,  0.0030,
        -0.0213,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0046,  0.0020, -0.0035,  0.0025,  0.0045, -0.0092, -0.0093, -0.0047,
         0.0050, -0.0764, -0.0262,  0.0163,  0.0228,  0.0099, -0.0063, -0.0214,
        -0.1584, -0.2596, -0.0231, -0.0341, -0.0051, -0.0650, -0.1245, -0.0133,
         0.0089, -0.0166,  0.0084,  0.0008,  0.0073,  0.0089, -0.0328,  0.0087,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.4270e-02, -1.8238e-01, -1.2873e-01, -6.9404e-03, -2.3708e-03,
        -1.0513e-02, -6.6799e-02, -1.3719e-01, -2.8581e-03,  1.2194e-02,
         1.3625e-02, -1.3538e-03, -4.5130e-03, -7.8489e-03, -6.9446e-03,
         1.5264e-02, -1.8429e-02, -1.3240e-02, -6.8791e-02, -3.2864e-02,
        -4.8353e-02,  1.0549e-04, -4.7213e-02, -8.6232e-02,  9.1402e-03,
        -2.2579e-02, -7.4566e-03, -2.1794e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0456, -0.0212, -0.0091,  0.0010,  0.0088, -0.1352, -0.1651, -0.0510,
        -0.0228, -0.0194, -0.0100, -0.0941,  0.0043,  0.0221, -0.0036, -0.0936,
        -0.0106, -0.0852, -0.0044, -0.0161, -0.0051,  0.0022, -0.0295,  0.0152,
        -0.0699,  0.0064,  0.0055, -0.0060, -0.0370,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0168, -0.1243, -0.0065, -0.0561, -0.1385, -0.0060,  0.0023,  0.0053,
        -0.0071,  0.0304, -0.0025, -0.0164, -0.0475, -0.1039,  0.0346, -0.0497,
         0.0422, -0.0645, -0.0105, -0.0305,  0.0101,  0.0104, -0.0338, -0.0090,
        -0.0291, -0.0400, -0.0356, -0.0020, -0.0016, -0.0062, -0.0011, -0.0240,
         0.0016,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.2187e-01, -3.1327e-01, -2.3207e-02, -8.3393e-02, -1.0502e-02,
        -6.6036e-02, -2.1049e-03, -2.5658e-02, -1.4583e-02, -6.1532e-02,
        -7.5265e-02, -5.6157e-03, -3.9873e-03,  8.3257e-03, -2.8519e-03,
         1.1097e-02, -1.6525e-02,  9.2061e-03, -5.3770e-02,  2.6553e-04,
        -3.5958e-02,  2.1435e-02, -1.0449e-02,  2.3095e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0154, -0.1896, -0.1396,  0.0007, -0.0879, -0.0055, -0.0209, -0.0547,
         0.0121, -0.0445,  0.0048, -0.0680, -0.0735, -0.0156,  0.0046,  0.0069,
         0.0214, -0.0362,  0.0004, -0.1049, -0.0603, -0.0017,  0.0309,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.8760e-02, -2.4855e-01,  8.7708e-03, -1.3606e-02, -2.7649e-02,
        -4.1082e-02,  1.9008e-02,  6.3407e-03,  1.6376e-03, -9.6128e-04,
         9.9580e-04,  4.7188e-03, -1.3227e-02,  7.9031e-03,  1.3353e-04,
         2.1599e-03, -5.8976e-02, -4.6564e-02, -2.8140e-02,  3.1377e-03,
        -9.3776e-03, -2.5256e-02, -8.0901e-03, -1.2338e-02, -6.2390e-02,
        -1.9788e-02,  3.1500e-03, -1.1841e-03, -5.5633e-03, -6.0989e-02,
        -5.2329e-02,  1.8989e-03, -3.7855e-02, -9.4205e-03, -4.0881e-02,
        -5.3516e-02, -3.2793e-04, -1.5430e-02,  7.8925e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0605, -0.0223, -0.0191, -0.0400, -0.1384,  0.0050, -0.0035,  0.0147,
         0.0081, -0.0038, -0.0065, -0.0007,  0.0053, -0.0016, -0.0185, -0.0544,
         0.0016, -0.0116,  0.0053,  0.0026, -0.0470,  0.0025,  0.0033,  0.0046,
        -0.0687, -0.1018,  0.0066, -0.0447, -0.0106, -0.0388, -0.0117, -0.0060,
         0.0148, -0.0209, -0.0099, -0.0011, -0.0216, -0.1092,  0.0212,  0.0069,
        -0.0100,  0.0087, -0.0058,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.4103e-02, -2.5270e-01, -4.5532e-03, -9.4720e-03,  3.6174e-03,
        -6.2820e-03, -2.7490e-03, -4.0151e-03, -6.5696e-02,  6.7078e-03,
         1.1913e-02, -1.1726e-03, -4.0086e-02, -2.2600e-03, -5.6587e-03,
        -7.5697e-03, -8.2369e-02, -6.4614e-02, -5.9441e-03, -5.4857e-03,
        -1.2762e-03, -1.2120e-03, -7.4223e-03,  2.4191e-03,  4.3764e-03,
         9.3258e-04,  6.9551e-03, -1.9148e-03,  4.8015e-03, -1.3538e-03,
         9.3104e-03, -1.8869e-02, -4.8233e-02,  7.0747e-03, -1.0370e-02,
        -1.0954e-02, -4.8345e-03, -3.5601e-02, -6.2832e-02, -1.2787e-02,
        -3.6331e-02, -5.2119e-02, -1.3182e-03,  2.7522e-03, -6.5248e-03,
         1.5767e-03, -4.2141e-04,  1.2477e-04,  4.3731e-03, -2.0845e-03,
         1.2018e-02, -4.4751e-03, -1.1891e-03,  1.8867e-03,  3.4059e-03,
        -4.9832e-03, -3.9153e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0088, -0.4304, -0.0308, -0.0526, -0.0473,  0.0090, -0.0793, -0.1310,
        -0.0196, -0.0073, -0.0499, -0.0078,  0.0593,  0.0669,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.1789e-02, -1.9635e-01, -6.1799e-02, -7.4424e-05, -1.2280e-03,
        -5.0211e-02, -8.6544e-03,  5.4967e-03, -2.6667e-02,  4.4127e-03,
        -3.9087e-02, -6.8653e-02, -2.0364e-02, -1.9581e-02, -5.3829e-02,
         5.4130e-03, -2.9250e-02, -3.8009e-02,  3.4387e-02, -5.1982e-02,
        -1.0478e-01, -1.3495e-02, -2.2380e-02,  9.7547e-03,  6.8017e-03,
         1.0646e-02, -1.7700e-03,  1.2337e-03,  1.4090e-02,  7.8101e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1357, -0.3723, -0.0048, -0.2298, -0.1927, -0.0284, -0.0316,  0.0049,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 5.9267e-02, -1.8909e-01, -2.4326e-02, -2.7992e-02, -2.9639e-02,
        -1.1467e-01,  1.2720e-03,  1.9956e-02, -7.8123e-03, -2.9746e-03,
        -5.9835e-03,  8.4758e-03, -6.0275e-02, -9.8746e-02, -2.9042e-03,
        -3.0280e-03,  1.1086e-02, -6.2648e-03,  4.8725e-03, -2.9665e-05,
        -6.5150e-03,  2.3627e-03, -1.5637e-03,  8.3281e-04, -5.9599e-02,
         1.8655e-03, -5.0362e-02, -9.3104e-03, -4.1880e-02, -1.0875e-01,
         4.1014e-03,  5.4349e-03, -2.8769e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.6110e-03, -1.9611e-01, -1.1214e-01, -2.5048e-02, -1.4842e-03,
        -3.5878e-02, -2.0297e-02, -7.9459e-03,  4.6810e-03, -3.0430e-02,
        -1.1081e-02, -1.4459e-02, -2.1766e-02, -6.7575e-04, -3.4798e-03,
         3.3487e-03, -1.6717e-04,  5.0329e-03, -1.5666e-02,  4.2071e-03,
        -1.3183e-03,  1.1548e-03,  8.8552e-03, -2.1290e-03,  6.0850e-03,
         5.1425e-03,  5.6552e-03, -4.0167e-02, -3.9629e-03,  1.4279e-03,
         5.6174e-03, -1.3491e-03, -2.5829e-02, -4.4152e-02, -1.1734e-03,
         2.1928e-03, -1.8027e-02,  9.5223e-03,  2.6812e-04,  2.4085e-03,
         1.7410e-03,  7.7583e-03, -1.0513e-03, -3.8848e-02,  5.5160e-03,
         6.2983e-03, -5.5118e-03,  6.1721e-04, -4.4487e-04,  3.8039e-03,
         5.6037e-03, -3.0511e-02, -1.6580e-02, -2.3045e-02, -8.6665e-03,
         2.8082e-03, -1.5777e-03,  7.3933e-03,  4.7992e-03, -2.6001e-02,
        -3.1907e-03, -1.5200e-02, -5.8775e-02,  2.4690e-03, -2.6232e-04,
        -1.1607e-03, -6.6899e-03,  5.3198e-04,  2.2151e-04, -4.5749e-03,
        -5.4034e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0760, -0.0126, -0.0274,  0.0050, -0.0614,  0.0018, -0.1129,  0.0033,
         0.0030,  0.0077,  0.0146,  0.0054,  0.0101, -0.0016,  0.0056,  0.0092,
         0.0006,  0.0183, -0.0148, -0.1058,  0.0024,  0.0059,  0.0145, -0.0615,
         0.0146, -0.0380, -0.0126,  0.0024,  0.0010,  0.0011, -0.0058,  0.0083,
        -0.1020, -0.0846, -0.0145, -0.0103, -0.0793, -0.0099,  0.0093,  0.0002,
        -0.0135, -0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.7873e-02,  2.5776e-01,  1.3056e-02,  2.3001e-02,  1.2166e-02,
        -9.5037e-03, -9.8767e-03, -8.6587e-04,  2.7754e-02,  4.3866e-02,
         1.5604e-02,  4.7463e-02,  5.3784e-02, -2.9661e-05,  1.4570e-03,
         6.9938e-03,  1.0451e-02,  4.0568e-02,  1.2126e-01, -7.6559e-03,
         1.0756e-02,  1.6965e-02, -5.4632e-03, -1.4721e-02,  2.9597e-02,
         3.6491e-02,  2.5788e-03,  4.4098e-04,  8.4912e-02, -3.3737e-03,
        -5.4042e-03,  1.0611e-02, -9.4212e-04, -3.1540e-03,  9.1145e-03,
         3.0534e-02,  1.3950e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0636, -0.5699,  0.0146, -0.0879, -0.0120,  0.0060, -0.0134,  0.0010,
        -0.0089,  0.0012, -0.0952,  0.0398, -0.0490, -0.0213,  0.0161,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0500, -0.0125,  0.0222,  0.0050, -0.1381, -0.2213, -0.0054, -0.1240,
        -0.1433, -0.0225,  0.0209,  0.0066, -0.0603,  0.0063,  0.0349, -0.0157,
         0.0077,  0.0417,  0.0066,  0.0550,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0261,  0.0013,  0.0009,  0.0109, -0.1307, -0.0057,  0.0048, -0.0229,
        -0.0060,  0.0025, -0.0779, -0.0797, -0.0094, -0.0175, -0.0389, -0.0263,
        -0.0073, -0.0248, -0.0067, -0.0202, -0.0057,  0.0077, -0.0120, -0.0030,
         0.0197, -0.0467, -0.0822, -0.0068,  0.0052, -0.0184, -0.0377,  0.0023,
         0.0030, -0.0048, -0.0364, -0.0487,  0.0056, -0.0225, -0.0076,  0.0023,
         0.0243,  0.0212,  0.0556,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0110, -0.0155, -0.0257,  0.0095, -0.0553, -0.0153, -0.0615, -0.1478,
        -0.0465,  0.0055, -0.0103,  0.0077, -0.0095, -0.0750, -0.1045, -0.0015,
        -0.0021, -0.0029,  0.0030,  0.0172, -0.0444, -0.0191,  0.0082, -0.0023,
        -0.0026, -0.0018, -0.0005, -0.0062, -0.0984, -0.0088, -0.0100, -0.0355,
        -0.0064, -0.0364, -0.0430, -0.0173,  0.0237,  0.0083,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0146, -0.2548, -0.1293, -0.1758,  0.0186, -0.0741,  0.0144,  0.0192,
        -0.0169, -0.0719, -0.0070, -0.0120, -0.0295,  0.0031,  0.0026, -0.0038,
         0.0210, -0.0665,  0.0024,  0.0162,  0.0205,  0.0258,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0163, -0.2126, -0.0253,  0.0088,  0.0044, -0.0233, -0.0263, -0.0009,
        -0.0217, -0.0881, -0.0191, -0.0011, -0.0027, -0.0014, -0.0465, -0.0557,
        -0.0703, -0.0188,  0.0006, -0.0427,  0.0004, -0.0052, -0.0424, -0.0249,
        -0.0081,  0.0024, -0.0120, -0.0013, -0.0059, -0.0567,  0.0157, -0.0073,
         0.0184, -0.0072, -0.0218, -0.0509,  0.0097, -0.0038,  0.0009, -0.0188,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.7201e-03, -1.0531e-02,  1.4021e-03,  5.6741e-03, -1.8935e-02,
        -3.7795e-02, -1.3983e-02, -1.8158e-03, -4.8585e-02,  7.2044e-03,
         5.8157e-03, -8.8641e-05, -6.9560e-03, -1.6937e-01, -1.3623e-01,
        -1.4454e-02,  2.0486e-02, -3.1142e-04, -8.6790e-02, -8.1651e-03,
         2.7028e-02,  4.5945e-03,  1.8957e-02, -2.2836e-02, -1.0046e-02,
        -7.4104e-02, -3.0500e-03, -4.2455e-02, -1.2214e-01, -1.9812e-02,
        -3.3728e-02,  1.9937e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0244, -0.2912, -0.0037, -0.0020, -0.0107,  0.0102,  0.0038, -0.0489,
        -0.1467, -0.0064, -0.0179, -0.0532,  0.0055,  0.0090,  0.0006, -0.0030,
        -0.0611, -0.0117,  0.0067,  0.0009,  0.0008,  0.0032, -0.0138, -0.1294,
        -0.0117, -0.0263, -0.0406, -0.0022, -0.0159, -0.0086, -0.0083,  0.0012,
        -0.0008, -0.0160, -0.0038,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.1076, -0.0336, -0.1021, -0.0727,  0.0182, -0.0230, -0.1047, -0.1089,
         0.0221, -0.1162,  0.0084, -0.0177, -0.0028, -0.0640, -0.0116, -0.0149,
        -0.0521, -0.0112,  0.0122, -0.0268,  0.0691,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0086, -0.3646,  0.0118, -0.0470, -0.1094, -0.0669, -0.0375, -0.0445,
        -0.1130,  0.0502,  0.0004, -0.0007,  0.0092, -0.0048, -0.0756,  0.0124,
        -0.0301, -0.0106, -0.0026,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0445, -0.0379, -0.1010,  0.0301, -0.0653, -0.0097, -0.0163,  0.0396,
        -0.0030, -0.0653, -0.1386, -0.0185, -0.0672, -0.1371,  0.0139, -0.0553,
        -0.0169, -0.0585, -0.0514, -0.0297,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0807, -0.3607,  0.0128, -0.0448,  0.0111,  0.0024,  0.0030, -0.0499,
        -0.0080, -0.0102,  0.0009, -0.0260,  0.0090, -0.0500, -0.0039, -0.0653,
        -0.0955, -0.0228, -0.0318, -0.0516,  0.0069,  0.0454,  0.0074,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0540, -0.1342, -0.0422, -0.0348, -0.0017, -0.0587, -0.0194, -0.0867,
        -0.0844, -0.1262, -0.0102, -0.0545, -0.0258, -0.0924,  0.0298, -0.0005,
         0.0099, -0.0014, -0.0303, -0.0364, -0.0393,  0.0021,  0.0170, -0.0034,
         0.0048,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.9417e-02, -3.3415e-01, -5.1051e-03, -6.7688e-03,  2.9131e-03,
        -6.8241e-04,  2.7530e-02, -2.0238e-03, -3.4320e-03, -4.4068e-04,
        -2.1047e-02,  1.8527e-03, -9.5857e-03,  1.5644e-02, -5.2519e-03,
         1.2027e-02, -8.0877e-05,  9.9737e-04,  2.1568e-02, -2.1772e-02,
        -5.9414e-02, -1.2283e-01, -7.6719e-02, -1.5714e-02, -3.7444e-03,
        -2.0479e-02, -2.0798e-02, -1.2702e-02, -1.8187e-02, -4.2573e-02,
        -1.9492e-02,  7.3223e-03, -1.6523e-02,  4.1141e-03,  7.0993e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0160, -0.0194,  0.0146, -0.1170, -0.0117,  0.0183, -0.0731, -0.0138,
        -0.1368,  0.0060, -0.0062, -0.0246, -0.0144, -0.0788, -0.0965, -0.1746,
         0.0002, -0.1003,  0.0062,  0.0139, -0.0461, -0.0115,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0481, -0.0510,  0.0129, -0.0200, -0.0157, -0.0424, -0.0295, -0.0155,
         0.0014, -0.3577, -0.0274, -0.0299, -0.0346, -0.0241,  0.0178, -0.0676,
         0.0521, -0.0254,  0.0221,  0.0265, -0.0785,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.1290e-02,  4.7568e-01, -1.0984e-04, -4.0923e-03,  4.2602e-03,
        -4.4568e-04,  1.0148e-01,  1.1204e-01, -3.1129e-02, -1.2602e-02,
         1.8023e-03, -2.1749e-02,  5.3231e-04,  6.8906e-02, -3.8015e-03,
         5.3340e-02, -6.7119e-03, -2.9397e-02, -2.0633e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0653,  0.0117, -0.0381,  0.0086,  0.0159, -0.0493,  0.0049,  0.0034,
         0.0103,  0.0009,  0.0104, -0.0171, -0.0081, -0.0282,  0.0037, -0.0106,
        -0.0427, -0.0568,  0.0155, -0.0437, -0.0007,  0.0004, -0.0456, -0.0008,
        -0.0914, -0.0109, -0.0199, -0.0286, -0.0003, -0.1365, -0.0007,  0.0035,
        -0.0653, -0.0762,  0.0096, -0.0044, -0.0004, -0.0035,  0.0010, -0.0240,
        -0.0023,  0.0031,  0.0012, -0.0067, -0.0054,  0.0071, -0.0050],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0077, -0.2938, -0.1628, -0.0049,  0.0138, -0.0053,  0.0027, -0.0095,
        -0.0504, -0.0007,  0.0056, -0.0006,  0.0164,  0.0225,  0.0040,  0.0046,
        -0.0631, -0.0066,  0.0029, -0.0130,  0.0072,  0.0246, -0.0971, -0.0911,
         0.0201,  0.0130, -0.0081,  0.0046,  0.0035,  0.0068,  0.0003, -0.0010,
         0.0287,  0.0028,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0713, -0.3605, -0.0010,  0.0201,  0.0204, -0.0458,  0.0100,  0.0045,
         0.0066, -0.0323,  0.0234,  0.0097, -0.0376, -0.1171, -0.0005,  0.0007,
        -0.0053, -0.0125, -0.0133, -0.0675, -0.0115,  0.0060, -0.0162,  0.0157,
        -0.0064,  0.0086,  0.0027, -0.0035, -0.0044, -0.0336,  0.0036,  0.0172,
         0.0036,  0.0069,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0404,  0.4608,  0.0120,  0.0186,  0.0188,  0.0713, -0.0391, -0.0010,
        -0.0176,  0.0839,  0.0186,  0.0034,  0.0285,  0.1007,  0.0075,  0.0258,
        -0.0363,  0.0158,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.0196e-02,  4.4607e-03,  1.0175e-02, -3.3360e-02, -3.9489e-03,
        -4.9399e-03,  2.4090e-02, -1.2132e-02, -8.0368e-02, -2.5442e-01,
        -4.0231e-02, -1.1520e-02, -1.6879e-02, -2.0773e-02,  6.1405e-03,
        -1.2094e-02, -1.8239e-02, -2.7191e-02, -2.1068e-03, -1.7898e-01,
         1.8494e-04,  4.5695e-02,  1.3468e-02,  1.6392e-02,  3.1322e-03,
        -2.3155e-02,  1.0572e-01,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.2289e-03,  1.4429e-02,  8.3146e-04, -7.7967e-03,  9.9003e-05,
        -5.6389e-03, -1.1035e-02, -2.2635e-02, -1.5955e-01,  4.0285e-02,
         1.2843e-02, -4.0501e-02, -1.8597e-01, -1.9068e-02,  1.3312e-03,
        -3.8917e-03, -1.5247e-02, -2.7512e-02,  1.6687e-02, -2.3795e-01,
         5.8153e-03,  3.0045e-02, -1.0929e-01, -7.7630e-03, -6.9897e-03,
        -1.3565e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0468, -0.2222, -0.0029, -0.0782, -0.1299, -0.0176, -0.0105,  0.0285,
        -0.0627, -0.0356, -0.0751, -0.0251,  0.0194, -0.0143,  0.0093, -0.0654,
        -0.0313, -0.0291, -0.0172,  0.0402,  0.0389,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1400, -0.1873, -0.1741, -0.1800, -0.0063, -0.0329, -0.1024,  0.0101,
        -0.0457, -0.0031, -0.0181, -0.0431,  0.0343, -0.0078,  0.0147,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0511,  0.0049, -0.0021, -0.0357,  0.0509,  0.0151, -0.0346,  0.0426,
        -0.0312, -0.0092, -0.0374, -0.0638,  0.3945,  0.0802,  0.0205,  0.0439,
         0.0105,  0.0719,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0402, -0.2268, -0.0967, -0.1206, -0.0010,  0.0262, -0.0599, -0.0850,
        -0.0212, -0.0701, -0.0566, -0.0381, -0.0069, -0.0681, -0.0022, -0.0026,
         0.0167, -0.0276,  0.0333,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.5538e-02, -1.5582e-01, -1.0880e-01, -4.3197e-02, -8.7000e-03,
        -2.9651e-02,  7.4450e-03, -6.1713e-03, -1.6754e-02, -6.0799e-03,
        -8.0331e-05, -8.6720e-03, -2.3666e-02,  7.6000e-03,  8.1508e-04,
        -7.3873e-03, -3.7881e-02,  4.5742e-03,  8.9240e-03,  1.3016e-02,
        -4.5951e-02,  4.0190e-03, -5.1038e-02, -3.3925e-03, -4.9192e-02,
        -5.8077e-04, -4.1690e-02,  1.6505e-02, -4.3753e-03, -2.3786e-02,
         4.8692e-03,  9.0678e-03, -4.8006e-02,  1.1340e-04, -2.3558e-02,
        -1.2491e-02, -5.6509e-02,  1.5869e-02, -1.2426e-02, -9.7911e-03,
        -3.6664e-03, -2.4411e-03, -1.6786e-02, -2.3107e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.3299e-02, -2.5283e-01, -1.9128e-02, -8.9239e-02, -9.5283e-02,
         1.7823e-02, -3.1393e-02, -1.9094e-02, -8.6648e-02, -4.0282e-03,
        -3.7837e-02, -5.9870e-03, -3.5060e-03, -1.9998e-03,  4.3590e-03,
        -2.4246e-04, -7.3837e-03, -8.2646e-02, -1.5821e-03, -2.1340e-02,
        -4.5553e-02, -5.5897e-03, -4.0757e-03,  3.2907e-03,  3.4970e-03,
         9.8634e-04,  3.7361e-03, -5.7945e-02,  5.2898e-04, -7.4168e-04,
        -3.5361e-03, -2.5511e-02,  8.2954e-03, -2.1927e-02,  9.1394e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0295,  0.3867,  0.0148,  0.0788,  0.0023, -0.0213,  0.1191,  0.0507,
         0.0378, -0.0008, -0.0073,  0.0414,  0.0787,  0.0028, -0.0146,  0.0429,
         0.0023, -0.0019, -0.0136, -0.0020, -0.0033, -0.0026,  0.0335,  0.0111,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0141, -0.1816, -0.0019, -0.0729,  0.0099, -0.0192,  0.0173, -0.0317,
        -0.1044, -0.0003, -0.0286,  0.0155,  0.0041, -0.0353, -0.0917, -0.0114,
        -0.0018, -0.0061, -0.0481, -0.0017, -0.0057, -0.0403,  0.0058, -0.0424,
        -0.0081, -0.0409, -0.0429, -0.0297, -0.0178, -0.0193, -0.0222, -0.0066,
         0.0105, -0.0102,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0116, -0.0025, -0.0186, -0.0420,  0.0014,  0.0244, -0.1572, -0.0240,
         0.0172,  0.0019,  0.0013, -0.0032, -0.0827, -0.0179, -0.1307, -0.1915,
        -0.0055, -0.0090, -0.0015,  0.0242, -0.0034, -0.0027, -0.0088, -0.0609,
        -0.0605, -0.0179, -0.0367,  0.0086,  0.0094,  0.0229,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0468,  0.0111, -0.0330, -0.1653, -0.0367,  0.0086,  0.0005,  0.0056,
        -0.1319, -0.3695, -0.0896,  0.0368,  0.0645,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0338,  0.0056,  0.0050,  0.0442, -0.0104, -0.0484, -0.0403, -0.1521,
        -0.2210,  0.0373,  0.0061, -0.1177,  0.0120, -0.0611, -0.1756, -0.0009,
        -0.0011, -0.0273,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1958,  0.3385,  0.2135, -0.0654,  0.1240, -0.0092,  0.0128,  0.0409,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0035, -0.2270, -0.0195, -0.0368, -0.0198,  0.0073, -0.0147,  0.0013,
        -0.0064, -0.0113, -0.0041,  0.0109, -0.0133,  0.0100,  0.0026, -0.0129,
         0.0105,  0.0052, -0.0007,  0.0031,  0.0052, -0.0163, -0.0477, -0.0228,
         0.0047, -0.0325,  0.0020, -0.0073, -0.0647, -0.0195,  0.0056, -0.0286,
        -0.0084, -0.0972,  0.0023, -0.0312, -0.0507, -0.0082,  0.0119,  0.0038,
        -0.0189,  0.0017, -0.0189,  0.0109,  0.0140, -0.0238,  0.0203],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0155, -0.2098,  0.0745, -0.4358, -0.0164, -0.0231, -0.0973,  0.0125,
         0.0282,  0.0208,  0.0661,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.2791e-02, -2.9156e-01, -6.1991e-02, -1.7718e-01, -6.0130e-02,
         2.2494e-02, -5.1905e-02, -1.5801e-01, -6.4805e-03, -3.8939e-02,
        -6.0763e-03,  1.3157e-02, -4.4749e-02, -1.0868e-02,  7.0548e-03,
        -2.2672e-05,  7.9100e-03,  8.6839e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1338, -0.0944, -0.3028, -0.0368,  0.0312,  0.0179, -0.1626, -0.0124,
        -0.0881, -0.0347,  0.0041, -0.0218,  0.0151, -0.0063, -0.0381,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0391, -0.3272, -0.0155, -0.0167,  0.0255, -0.1208, -0.1376, -0.0321,
        -0.0271, -0.0675, -0.1072,  0.0061,  0.0069, -0.0523, -0.0182,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0467,  0.0291, -0.0091,  0.0498, -0.0019, -0.0086, -0.0084, -0.0144,
        -0.0418, -0.0177, -0.0110, -0.0728, -0.1704,  0.0018,  0.0195, -0.0272,
        -0.0446, -0.0210,  0.0167,  0.0021, -0.1104, -0.1832,  0.0336,  0.0204,
        -0.0209, -0.0168,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0970, -0.0099,  0.0214,  0.0536, -0.0051,  0.0435,  0.0224, -0.2574,
        -0.3068, -0.0354, -0.0069,  0.0166,  0.0373,  0.0475, -0.0392,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.3356e-02, -3.6878e-01, -5.7168e-03, -4.9748e-02, -7.9901e-02,
        -1.9191e-02, -3.6815e-02, -1.1764e-02, -4.7492e-03, -1.3621e-01,
        -3.6171e-03,  1.5792e-02, -2.8791e-02, -1.1163e-02,  7.3021e-03,
         2.3415e-03, -2.6213e-02, -3.5153e-04, -2.3614e-02,  5.1611e-03,
        -2.4941e-02,  4.6295e-03,  9.3378e-04, -1.5223e-02,  4.9165e-02,
         4.4529e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0343,  0.0242,  0.0164,  0.0186,  0.0074, -0.0546, -0.0107,  0.0103,
        -0.0492, -0.0603, -0.1552, -0.0162, -0.0191,  0.0081, -0.1493, -0.1280,
         0.0087, -0.0508,  0.0116, -0.0662, -0.0116, -0.0414, -0.0052, -0.0260,
         0.0097, -0.0069,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-0.0555, -0.3666, -0.0039, -0.0011, -0.0211, -0.0013, -0.0312, -0.0774,
        -0.0046, -0.0276,  0.0007, -0.0286, -0.0690,  0.0012, -0.0213,  0.0031,
         0.0017,  0.0102, -0.0161, -0.0004,  0.0088, -0.0189, -0.0079, -0.0402,
        -0.0612,  0.0131, -0.0310, -0.0038,  0.0150,  0.0006,  0.0040, -0.0280,
        -0.0005,  0.0056,  0.0096, -0.0093,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0075, -0.0588, -0.0528,  0.0169, -0.0785, -0.0114, -0.0644, -0.0901,
        -0.0029,  0.0045, -0.0407, -0.0020, -0.0011,  0.0063, -0.0025, -0.0210,
         0.0087, -0.0074,  0.0016, -0.0007, -0.0074, -0.0587, -0.0079, -0.0112,
        -0.0506, -0.0777,  0.0030, -0.0408, -0.0407, -0.0067, -0.0107, -0.0451,
        -0.0647,  0.0180, -0.0517,  0.0253,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0054,  0.0192, -0.0201, -0.0328,  0.0125, -0.1815, -0.0522,  0.0129,
         0.0077,  0.0156, -0.0036, -0.1954, -0.2367, -0.0369,  0.0120, -0.0249,
         0.0029,  0.0076, -0.0210,  0.0990,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.3339e-02, -1.6057e-01, -2.7559e-02, -3.6318e-02,  2.3259e-02,
        -4.5485e-03,  2.2529e-02, -4.1013e-02, -4.0364e-03, -2.1184e-02,
         3.0535e-03, -9.3836e-03,  4.0491e-03,  1.9931e-02, -3.4064e-02,
        -8.6588e-02, -1.6245e-02, -5.9631e-02,  4.5656e-03, -8.5262e-04,
         1.3333e-02,  7.0416e-03,  6.7807e-03,  3.0517e-02,  7.8102e-03,
         1.2468e-02,  8.8804e-03, -4.5159e-03,  1.9092e-02, -4.5325e-02,
        -1.0095e-01, -1.5508e-02, -1.5841e-02, -5.8657e-02,  1.1690e-02,
         7.9687e-03,  1.0788e-02, -1.1741e-04,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0197, -0.2511, -0.1192, -0.0594, -0.0090,  0.0037, -0.0105,  0.0060,
         0.0050, -0.0056, -0.0437, -0.0611,  0.0006, -0.0031, -0.0089, -0.0396,
         0.0039, -0.0838, -0.0004,  0.0019,  0.0069,  0.0036,  0.0136,  0.0135,
         0.0079, -0.0051, -0.0080, -0.0194, -0.0556, -0.0043, -0.0084, -0.0110,
         0.0047,  0.0019, -0.0524,  0.0035, -0.0197, -0.0077, -0.0078, -0.0055,
         0.0035], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0284, -0.4239, -0.0096,  0.0365, -0.0017, -0.1076, -0.0371,  0.0008,
        -0.0123,  0.0025, -0.0014, -0.0590, -0.0906,  0.0062,  0.0078, -0.0378,
        -0.0005, -0.0433,  0.0070,  0.0157,  0.0036, -0.0667,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.2417e-02,  1.0755e-02,  4.9648e-03, -1.5689e-03, -2.1107e-03,
        -1.1542e-02,  9.6968e-03,  2.8545e-03, -2.0696e-02, -9.9423e-04,
        -1.3319e-02, -9.0442e-02, -5.3208e-03, -4.8320e-02, -1.6086e-02,
        -9.3895e-02,  2.1991e-02, -3.6993e-02,  9.0761e-03, -9.0982e-02,
        -1.0277e-01,  1.2091e-02,  1.2716e-02, -6.5430e-02, -5.0208e-04,
        -6.9592e-03, -6.1458e-03, -2.7995e-02, -1.1060e-01,  1.0804e-02,
        -3.1681e-02, -5.0950e-02, -1.8101e-02,  8.7032e-05, -2.9141e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.7958e-02, -1.5765e-02, -2.2810e-01, -1.8473e-02, -1.2295e-01,
        -1.3811e-02,  1.7326e-02,  4.4720e-03,  3.2094e-02, -4.7968e-03,
        -4.9226e-02, -3.9880e-03,  3.6884e-03,  8.4771e-03, -3.8854e-03,
         2.5493e-02,  2.6937e-02, -1.3218e-03, -5.8314e-02,  1.7565e-04,
        -3.3956e-02,  4.6472e-03, -4.1031e-02, -1.0356e-01, -1.2551e-02,
        -3.5856e-02, -7.6459e-02,  4.7531e-04, -2.4211e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.2691e-03, -2.2684e-01, -1.1232e-01, -4.4638e-02, -1.1789e-02,
        -9.2033e-02, -9.0962e-05, -3.3326e-03, -2.3244e-02,  1.4156e-02,
        -2.2527e-02, -4.1913e-02,  1.3837e-02, -9.9528e-03,  1.1623e-02,
        -1.1484e-01, -5.3773e-02,  3.3085e-03, -5.0986e-02,  3.0093e-03,
        -3.5102e-02, -1.1299e-02, -4.7538e-02,  3.0720e-02,  1.8854e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0066,  0.0162,  0.0048, -0.0041, -0.2057, -0.0134, -0.0244, -0.1267,
        -0.1439, -0.0175, -0.0064, -0.0003,  0.0239,  0.0210, -0.1036, -0.1614,
         0.0323, -0.0310, -0.0036,  0.0083,  0.0003,  0.0253,  0.0193,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0096, -0.0269,  0.0009,  0.0043,  0.0003, -0.0062,  0.0081, -0.0069,
         0.0045,  0.0003, -0.0019, -0.0045, -0.0838, -0.1560,  0.0009, -0.0104,
        -0.0101, -0.1052, -0.0078,  0.0072, -0.0149,  0.0033,  0.0004,  0.0085,
         0.0123, -0.0069, -0.1461, -0.0142, -0.0988, -0.0166, -0.0657, -0.0022,
        -0.0257, -0.0321, -0.0417, -0.0170, -0.0078, -0.0283, -0.0017,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1417, -0.0113,  0.0195, -0.1246, -0.2487,  0.0052, -0.0485, -0.1777,
        -0.1207,  0.0628, -0.0327, -0.0067,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.0202, -0.1391, -0.1902, -0.0110, -0.0114, -0.0117, -0.0626, -0.0102,
         0.0036, -0.0514, -0.0033, -0.0075,  0.0048, -0.0275, -0.1203,  0.0004,
        -0.0086, -0.0293, -0.0035,  0.0013, -0.0148, -0.0218,  0.0076,  0.0091,
        -0.0053,  0.0099,  0.0073, -0.0028, -0.0154,  0.0011,  0.0088,  0.0084,
        -0.0283,  0.0015,  0.0131,  0.0159, -0.0513, -0.0069,  0.0011,  0.0022,
         0.0096,  0.0125,  0.0104, -0.0076, -0.0094,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1071, -0.5434, -0.0556, -0.0868, -0.0115,  0.0022,  0.0214, -0.0102,
        -0.1133, -0.0195, -0.0191,  0.0099,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0228, -0.6671, -0.0019, -0.1923, -0.0454,  0.0080,  0.0069, -0.0555,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.7893e-03, -2.1699e-01,  6.1950e-04, -8.6448e-03, -7.1555e-02,
        -4.7721e-03, -3.8845e-03, -1.5697e-02, -2.6147e-02, -7.4424e-02,
        -7.7182e-03, -1.0241e-04, -4.7578e-03,  1.3226e-03,  5.2605e-03,
        -5.3757e-04, -1.0785e-02,  1.9185e-03,  4.9509e-03,  1.6354e-03,
         2.1847e-03,  2.9072e-03, -5.6315e-03, -6.3969e-03, -8.3864e-03,
         2.1297e-03,  1.0465e-03, -3.2718e-03, -2.8469e-02,  7.6208e-03,
         2.5668e-05, -1.0167e-02,  1.9499e-03,  3.1596e-03, -3.4898e-03,
         1.5498e-03, -9.1927e-04, -1.6823e-03,  6.2010e-04,  2.9882e-03,
        -3.1522e-03, -1.1901e-03, -5.0453e-03, -5.3678e-03,  5.1120e-03,
         8.9120e-03, -7.7196e-02, -2.0161e-03, -5.4308e-02, -1.0216e-01,
        -2.8440e-02,  6.4002e-03, -4.8676e-02,  3.6517e-03, -9.0927e-03,
        -2.2550e-03, -1.3214e-02, -5.9106e-03,  4.0137e-03,  2.2496e-02,
         2.9275e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0220, -0.0029,  0.0096, -0.0317, -0.1078,  0.0218, -0.0456,  0.0067,
        -0.0144,  0.0146, -0.0128, -0.0879, -0.0152, -0.0699, -0.1400, -0.0010,
        -0.0208,  0.0123, -0.0360, -0.0088, -0.0581, -0.0803, -0.0145,  0.0115,
        -0.0368, -0.0215, -0.0030, -0.0186, -0.0093, -0.0376,  0.0069,  0.0154,
        -0.0048,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0214, -0.0259, -0.1343, -0.2047, -0.0167,  0.0045, -0.0585,  0.0129,
         0.0057, -0.0022, -0.0014, -0.0025, -0.0452, -0.0076, -0.0233,  0.0846,
        -0.0024, -0.0499, -0.1476,  0.0154, -0.0064, -0.0244,  0.0676,  0.0352,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0144, -0.1484, -0.0841, -0.0573, -0.0224, -0.1155,  0.0213, -0.0466,
        -0.0020, -0.0836, -0.0819, -0.0664, -0.0997, -0.0067, -0.0033,  0.0049,
         0.0148, -0.0065,  0.0168, -0.0005, -0.0327, -0.0087,  0.0091, -0.0523,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0108, -0.0128, -0.0625, -0.1106, -0.0157,  0.0110,  0.0067, -0.0711,
         0.0035,  0.0162,  0.0022, -0.0026,  0.0150,  0.0042, -0.0486, -0.0526,
        -0.0626,  0.0126,  0.0182, -0.0200, -0.1054, -0.0066, -0.0658, -0.1093,
         0.0089,  0.0101, -0.0501,  0.0098, -0.0050, -0.0391, -0.0186,  0.0003,
        -0.0078,  0.0038,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0340,  0.0054, -0.0083,  0.0120, -0.0942,  0.0012,  0.0110,  0.0030,
         0.0075,  0.0209,  0.0005,  0.0116,  0.0032, -0.1988, -0.0112, -0.0076,
        -0.1862,  0.0104, -0.0951,  0.0092, -0.0752, -0.1416,  0.0293,  0.0091,
        -0.0135,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0371,  0.0438, -0.0336, -0.0433, -0.0255, -0.0037, -0.0945, -0.0801,
        -0.1272, -0.3059,  0.0130, -0.0281, -0.0658, -0.0123, -0.0749, -0.0111,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0190, -0.2207, -0.0279, -0.0940, -0.0496, -0.0635, -0.1262, -0.0103,
         0.0135, -0.0070, -0.0066, -0.0148, -0.0421,  0.0065, -0.0034,  0.0004,
        -0.0455, -0.1272, -0.0554, -0.0135,  0.0009, -0.0135, -0.0388,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0811,  0.0523,  0.0313, -0.1675, -0.4022,  0.0488,  0.0457, -0.0066,
        -0.0255,  0.0375,  0.0189,  0.0126, -0.0079, -0.0620,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([-0.0292, -0.1592, -0.0677, -0.1927, -0.0083, -0.0099, -0.0568, -0.0093,
        -0.0037,  0.0043, -0.0949, -0.0947, -0.0114, -0.0203, -0.0038, -0.0013,
        -0.0036, -0.0878, -0.0088, -0.0680, -0.0153,  0.0058, -0.0147,  0.0285,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0067, -0.0052, -0.1352,  0.0013, -0.3965, -0.0618, -0.3361, -0.0094,
        -0.0327, -0.0037,  0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0066,  0.0384,  0.0187,  0.0212,  0.0182, -0.1160, -0.1708, -0.0027,
        -0.1089,  0.0042, -0.0602,  0.0344, -0.0346,  0.0381, -0.0714, -0.0213,
        -0.0405, -0.1057,  0.0055,  0.0299, -0.0526,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0204,  0.0188,  0.0353,  0.0152,  0.0614, -0.0007, -0.0407, -0.3361,
        -0.1553, -0.2646, -0.0043, -0.0102,  0.0101, -0.0269,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0314,  0.0156, -0.0014, -0.0038, -0.0143,  0.0119,  0.0029, -0.0645,
        -0.0365, -0.0171,  0.0270,  0.0645,  0.4715,  0.0413,  0.0198, -0.0024,
         0.0608,  0.0224, -0.0026,  0.0103,  0.0779,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0184, -0.2794,  0.0091, -0.1305, -0.0119, -0.1655, -0.0164, -0.2017,
        -0.0222, -0.0676, -0.0206, -0.0005, -0.0563,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.6261e-02, -1.3869e-01, -9.1811e-03, -5.4684e-03, -6.8167e-03,
        -2.2426e-03, -5.8770e-02, -7.6622e-02, -1.1880e-02,  9.9627e-03,
        -2.9950e-03,  4.2350e-03, -4.9162e-04, -6.3441e-03, -5.0269e-03,
        -2.3608e-02, -2.6755e-03, -2.7165e-03,  3.5370e-03,  1.3695e-02,
        -6.7220e-03,  1.3615e-02,  4.0546e-03, -6.2347e-03, -1.5073e-01,
         4.9696e-03, -5.5573e-02, -7.0109e-02, -7.3183e-02, -1.1564e-02,
        -1.0608e-04,  1.8559e-03, -3.8674e-03, -4.9421e-03,  6.1660e-03,
        -5.1760e-03, -4.0000e-02,  2.9891e-04, -4.2110e-04, -2.9186e-02,
        -2.9988e-02, -6.6197e-03,  8.6414e-03, -1.2559e-02, -3.2203e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0363, -0.0341, -0.1148, -0.0116, -0.0010,  0.0186, -0.0008, -0.0244,
        -0.0029, -0.0059, -0.0825,  0.0087,  0.0143,  0.0079, -0.0503, -0.0048,
        -0.1319,  0.0024,  0.0035,  0.0044, -0.0232, -0.0813, -0.0056, -0.0704,
         0.0088, -0.0124, -0.0207, -0.0208, -0.1256, -0.0078,  0.0047,  0.0146,
        -0.0131,  0.0301,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0510, -0.0076,  0.0357,  0.0208, -0.0033, -0.0126, -0.0049,  0.0304,
         0.0166,  0.0044, -0.0045,  0.0057, -0.0035,  0.0052,  0.0119,  0.0262,
        -0.0013, -0.0164,  0.0211, -0.2848, -0.3119, -0.0046, -0.0322, -0.0321,
         0.0228, -0.0285,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.8559e-03, -3.6585e-01,  1.2730e-02, -7.4800e-02,  2.1485e-02,
         1.1484e-02, -2.1861e-02, -6.4690e-02,  3.2303e-02,  2.4203e-03,
         2.0976e-05,  3.4127e-03,  7.0767e-03, -3.5160e-04, -4.7835e-04,
        -1.0510e-01, -1.7833e-01, -2.1269e-03, -2.9086e-02, -6.5314e-04,
        -2.5407e-02, -1.3946e-02, -2.2540e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0268, -0.1938, -0.0300, -0.0549,  0.0003, -0.0012, -0.0180, -0.0418,
        -0.0680, -0.0147, -0.0050, -0.0048, -0.0012,  0.0131, -0.1259,  0.0084,
        -0.0047, -0.0310, -0.0514, -0.0104, -0.0066, -0.0083,  0.0024,  0.0032,
         0.0015,  0.0018, -0.0563, -0.0851,  0.0050,  0.0020, -0.0037,  0.0185,
        -0.0024,  0.0078, -0.0101,  0.0047, -0.0339, -0.0020,  0.0231, -0.0162,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0149,  0.0007,  0.0027, -0.0460, -0.0853, -0.0235, -0.0090, -0.0559,
         0.0064, -0.0162,  0.0051, -0.0095, -0.0568, -0.0056, -0.0423, -0.0818,
        -0.0893, -0.0174, -0.0165, -0.0418, -0.0105, -0.0797, -0.0800, -0.0052,
        -0.0165, -0.0057, -0.0067, -0.0368,  0.0012, -0.0202, -0.0767, -0.0023,
        -0.0223, -0.0094,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0329, -0.0214, -0.0143, -0.0145, -0.1101,  0.0031,  0.0216, -0.0109,
        -0.0018,  0.0037,  0.0050, -0.0243,  0.0353, -0.0032, -0.1266, -0.0104,
         0.0072, -0.0021, -0.0522, -0.0003,  0.0079,  0.0011, -0.0164,  0.0188,
        -0.0451, -0.0727,  0.0125, -0.0027,  0.0019,  0.0109, -0.0151,  0.0014,
        -0.0012,  0.0124, -0.0297, -0.0011, -0.0043, -0.0353, -0.0019, -0.0482,
        -0.0491, -0.0097, -0.0298, -0.0035,  0.0006,  0.0072,  0.0052, -0.0013,
        -0.0131, -0.0015, -0.0340,  0.0033], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0050, -0.0010, -0.0030, -0.0113, -0.1312,  0.0144, -0.0155,  0.0230,
        -0.0025, -0.0158,  0.0035, -0.0056, -0.0088, -0.1197, -0.0187, -0.0966,
        -0.1530, -0.0676, -0.0959,  0.0046, -0.0247, -0.0185, -0.0382, -0.0732,
         0.0076,  0.0077,  0.0018,  0.0094,  0.0129,  0.0030, -0.0064,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0298, -0.2337,  0.0079, -0.0137,  0.0286,  0.0139, -0.0210,  0.0124,
        -0.0057,  0.0447, -0.0005, -0.0522,  0.0420,  0.0012, -0.0601, -0.0789,
         0.0370, -0.0651, -0.1050, -0.0760, -0.0038,  0.0035, -0.0243,  0.0224,
         0.0164,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1402,  0.0029, -0.0020, -0.0371,  0.0119, -0.2201, -0.4322, -0.0674,
        -0.0251, -0.0222,  0.0087, -0.0302,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.2039e-02, -3.4870e-01, -2.5644e-02, -1.8610e-02, -1.0779e-01,
        -3.2439e-02, -3.3737e-03, -6.7232e-03,  1.3871e-03, -9.9530e-03,
         8.2017e-03, -1.5776e-02, -7.2773e-02, -8.6912e-02,  4.2232e-03,
        -2.8718e-02, -1.7764e-02, -1.9916e-02, -2.5398e-04,  6.9893e-04,
        -4.2534e-02,  1.6284e-03, -1.8573e-02,  1.1714e-02, -2.7550e-05,
        -4.3476e-02,  1.5675e-02,  6.9957e-04,  3.3773e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0518,  0.3204,  0.0142, -0.0463,  0.0200,  0.2126,  0.2259, -0.0393,
        -0.0129,  0.0072,  0.0057,  0.0436,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0511, -0.3995, -0.0239, -0.0638,  0.0348, -0.1216, -0.1564,  0.0032,
        -0.0872, -0.0274,  0.0085, -0.0226,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1089,  0.0520, -0.1967,  0.0153, -0.0709, -0.0027, -0.0051,  0.0031,
        -0.1789, -0.0181,  0.0105, -0.0901, -0.0093,  0.0067, -0.0018, -0.1186,
         0.0065,  0.0399, -0.0274, -0.0376,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.6360e-02, -2.5843e-01, -1.3031e-02, -9.2158e-02, -1.1205e-01,
        -1.3520e-02, -3.4394e-02, -1.6517e-02, -3.5259e-02, -2.5778e-03,
        -6.9003e-03, -3.2547e-02, -3.7510e-02, -1.7578e-02, -6.7470e-03,
        -8.3118e-03, -3.2002e-02,  1.4490e-02, -8.8929e-03,  3.0035e-03,
        -4.2815e-02,  1.2235e-03,  7.0678e-05, -4.1707e-03,  8.2533e-03,
         3.6379e-03,  7.8020e-03, -1.9980e-02, -4.9735e-02,  1.3419e-03,
         4.4199e-03,  2.0232e-03, -1.3738e-02, -4.5884e-02, -6.2529e-03,
        -1.6065e-02,  3.0278e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0676, -0.7384, -0.0688,  0.0119, -0.0288, -0.0040,  0.0185, -0.0593,
        -0.0028,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0052,  0.0024,  0.0090, -0.1106,  0.0078, -0.0117, -0.0554, -0.0662,
         0.0211,  0.0071, -0.0223, -0.0358, -0.1222, -0.1241,  0.0310, -0.0135,
         0.0066,  0.0142, -0.0913, -0.1113,  0.0282,  0.0122,  0.0076,  0.0238,
        -0.0015, -0.0412, -0.0165,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0034, -0.2633,  0.0175, -0.0586, -0.0107, -0.0317, -0.0035,  0.0047,
        -0.0004, -0.0122,  0.0237, -0.0314, -0.0135,  0.0027,  0.0030,  0.0131,
        -0.0804,  0.0033,  0.0109,  0.0118, -0.0008, -0.0818, -0.1174, -0.0156,
        -0.0085, -0.0289, -0.0056, -0.0278, -0.0713, -0.0004, -0.0038,  0.0010,
         0.0370, -0.0004,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0055, -0.0005, -0.0019,  0.0071, -0.0688, -0.0234,  0.0029, -0.0059,
         0.0057,  0.0077,  0.0008, -0.0989,  0.0242,  0.0012,  0.0099, -0.0386,
        -0.0863, -0.0035, -0.0292,  0.0123,  0.0224,  0.0128, -0.0196,  0.0025,
        -0.0126, -0.0021,  0.0042,  0.0145, -0.0043,  0.0025,  0.0175, -0.0621,
        -0.1413,  0.0178, -0.0217, -0.0346, -0.0705, -0.0090,  0.0025, -0.0037,
        -0.0193, -0.0169, -0.0085,  0.0374,  0.0057,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0203, -0.0118, -0.0092, -0.0089, -0.0609, -0.1759, -0.3096, -0.0029,
         0.0051, -0.1884,  0.0106, -0.0224, -0.0558,  0.0421, -0.0294, -0.0466,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0120,  0.0054,  0.0033, -0.0026,  0.0043, -0.0273, -0.0122, -0.0104,
        -0.1200, -0.1686, -0.0476, -0.2330,  0.0363, -0.0589, -0.1105,  0.0043,
        -0.0068,  0.0045,  0.0101, -0.0021, -0.0080,  0.0095,  0.0065,  0.0112,
         0.0143,  0.0138,  0.0564,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.7809e-02,  1.2207e-02, -5.7654e-02, -6.1021e-03, -3.1015e-02,
        -7.6507e-03, -2.1335e-02, -8.2660e-03, -2.5666e-02,  7.7078e-04,
         3.0357e-05, -4.2154e-03, -2.6915e-02,  8.6079e-04, -1.0419e-02,
         1.0811e-03, -2.7069e-02,  4.1045e-03, -1.3877e-02, -4.5011e-02,
         1.7909e-03, -3.1950e-02, -1.2708e-02, -6.3183e-02, -8.3367e-03,
        -2.5432e-04, -2.3647e-02, -6.3118e-02,  6.5239e-03, -3.6813e-02,
        -3.2846e-03, -2.7994e-02, -4.3147e-02, -8.6839e-03, -1.0703e-02,
        -4.4531e-02, -2.8599e-02, -7.1783e-02,  1.4391e-02,  1.0493e-02,
        -2.6687e-02, -4.3563e-02, -1.4953e-03,  1.6857e-02, -3.0825e-02,
        -3.6581e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0080,  0.0197, -0.0068, -0.0050, -0.0236, -0.0160,  0.0022, -0.1797,
         0.0037, -0.0191, -0.0260, -0.0320,  0.0105, -0.0165, -0.0241,  0.0015,
         0.0083,  0.0063, -0.1506, -0.2702,  0.0096, -0.0024,  0.0035, -0.0160,
         0.0023,  0.0157, -0.0906,  0.0037, -0.0134, -0.0130,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0742, -0.0331,  0.0693,  0.0275, -0.0132, -0.0220, -0.0065,  0.4750,
        -0.0395, -0.0312,  0.2084,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0834, -0.3401,  0.0065, -0.0123, -0.1048, -0.0007, -0.0815, -0.1267,
        -0.0154,  0.0219, -0.0028,  0.0096,  0.0112, -0.0114,  0.0192, -0.0817,
        -0.0137, -0.0279, -0.0213, -0.0077,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0386, -0.3501, -0.2402, -0.1048, -0.0259, -0.0621, -0.1134,  0.0111,
         0.0223, -0.0138,  0.0179,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0246,  0.2541,  0.1371,  0.1616,  0.0245,  0.0272,  0.0262,  0.0621,
         0.1108, -0.0119,  0.0098,  0.0162,  0.0129,  0.0226, -0.0119,  0.0310,
        -0.0140,  0.0049, -0.0074, -0.0105, -0.0187,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0739, -0.2596, -0.1548,  0.0120, -0.0189, -0.1991, -0.0028, -0.0318,
        -0.0174, -0.0101, -0.0048, -0.1087, -0.0053, -0.0067, -0.0075, -0.0008,
        -0.0311, -0.0547,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0237, -0.1362, -0.1654,  0.0019, -0.0149,  0.0037, -0.0026,  0.0016,
        -0.0036, -0.0137, -0.0013,  0.0124,  0.0053, -0.0428, -0.0466, -0.0012,
        -0.0161, -0.0605, -0.0358,  0.0110, -0.0219, -0.0500, -0.0837, -0.0068,
        -0.0443, -0.0054, -0.0408, -0.0163, -0.0315, -0.0487,  0.0031,  0.0029,
         0.0115,  0.0070, -0.0089,  0.0013,  0.0044, -0.0083, -0.0026,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0553, -0.4730, -0.0207, -0.0693, -0.0040, -0.0068, -0.0096, -0.0026,
        -0.0043, -0.0184, -0.0169, -0.0660, -0.0111,  0.0031,  0.0020, -0.0015,
        -0.0455, -0.0040, -0.0078, -0.0062, -0.0097,  0.0069,  0.0021,  0.0049,
         0.0014,  0.0008, -0.0144, -0.0101, -0.0382, -0.0671, -0.0018, -0.0058,
        -0.0075,  0.0011,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #200: [tensor([ 0.0361, -0.0447, -0.2890, -0.2841,  0.0448, -0.1623, -0.0587, -0.0756,
        -0.0046,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0080, -0.2149, -0.0084, -0.1027, -0.0236, -0.0898, -0.0007, -0.0227,
        -0.0414, -0.0018, -0.0323, -0.0214, -0.0002, -0.0089, -0.0118, -0.0587,
         0.0011, -0.1093, -0.0085,  0.0017, -0.0056,  0.0035,  0.0118, -0.0058,
        -0.0375, -0.0003, -0.0041, -0.0096,  0.0046, -0.0005, -0.0089, -0.0421,
         0.0036, -0.0425,  0.0094, -0.0028, -0.0070,  0.0201,  0.0058,  0.0065,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.4649e-02, -1.1362e-01, -1.8025e-01,  3.1635e-03, -5.0937e-02,
        -7.3838e-03, -2.7161e-05, -6.9969e-03, -4.1118e-02, -5.8760e-02,
         3.7146e-03, -2.8447e-03,  1.0265e-03, -2.4249e-03, -3.1653e-03,
        -4.0495e-03,  7.4201e-03,  1.3329e-02, -4.7050e-02,  1.1206e-02,
        -3.3756e-02, -4.1272e-03,  4.9055e-03, -4.0744e-03,  7.8814e-03,
        -1.3752e-02, -3.6989e-02, -6.3709e-03, -4.5715e-02, -6.6948e-02,
        -3.2812e-03, -4.0887e-03, -3.2561e-03,  2.1082e-03,  5.4544e-03,
         1.0803e-03, -3.5192e-02, -4.5064e-02, -7.1281e-04, -2.6668e-03,
         1.4869e-03, -8.4260e-03,  2.5001e-03,  4.3593e-03,  2.1718e-03,
         2.6892e-03,  6.0872e-04, -2.1538e-03, -8.5052e-04,  1.5801e-02,
         3.2894e-03,  2.5105e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0565, -0.2758, -0.0113, -0.0689, -0.0236, -0.0068, -0.0926, -0.0322,
        -0.0365,  0.0133,  0.0045, -0.0089, -0.0719, -0.0132, -0.0284, -0.0777,
        -0.0538, -0.0266, -0.0579,  0.0022,  0.0066, -0.0307,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0821, -0.4027,  0.0240, -0.0361,  0.0056, -0.0319,  0.0013, -0.0462,
         0.0145, -0.0396,  0.0035, -0.0440, -0.0161, -0.0496, -0.0012, -0.0981,
        -0.0080, -0.0315,  0.0238, -0.0260, -0.0141,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.5095e-03, -2.1715e-01, -4.4388e-02, -1.1653e-02, -3.4107e-03,
        -7.5522e-03, -4.1718e-02,  5.4375e-03, -2.0757e-02,  3.9660e-03,
         4.0835e-03, -6.8355e-03,  3.1206e-03, -1.2452e-03, -3.4782e-04,
         8.8945e-04,  1.0184e-02,  1.4809e-02, -4.1509e-02,  4.8186e-03,
         6.5754e-03,  7.6558e-03, -5.2761e-03, -1.5058e-04,  2.6523e-03,
        -2.8581e-03, -1.2034e-02, -3.6449e-02,  5.8517e-03, -3.4869e-02,
        -5.1490e-02, -4.6961e-02, -2.4615e-03, -3.7660e-02, -2.6009e-02,
        -7.8706e-03, -5.5285e-03, -1.3272e-02, -4.1147e-04, -6.9026e-03,
        -5.7519e-02, -6.5132e-03,  6.4307e-05,  8.2272e-03, -3.1077e-02,
         2.8172e-03, -3.1459e-02, -2.8129e-02, -4.7397e-03, -2.7449e-03,
        -2.5735e-02, -6.5336e-03, -1.4792e-02, -6.9515e-03,  1.2375e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.6184e-03,  8.6472e-03,  2.6647e-02, -1.4850e-02,  1.7009e-03,
        -3.1468e-02, -9.8552e-02,  6.1750e-03, -4.9906e-03,  3.4159e-04,
        -5.0193e-03,  8.0476e-03,  1.1824e-05, -1.9713e-02, -5.1249e-02,
        -3.0016e-03, -5.3017e-03, -4.1243e-02, -1.0434e-02, -7.8928e-02,
        -1.5733e-02, -2.1307e-02, -4.9851e-02,  1.0231e-02,  7.1363e-03,
         7.3611e-03, -4.9095e-03, -5.1313e-02,  1.2797e-02,  6.6945e-03,
         8.7594e-03,  1.5184e-02, -1.7474e-02, -7.5238e-02, -4.1929e-04,
        -2.1305e-02, -4.0947e-03, -7.7717e-02, -5.6071e-02, -4.5879e-02,
        -5.9071e-03, -5.0815e-03,  8.1500e-03, -1.8847e-02, -2.6601e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0078, -0.2296, -0.1138, -0.1044, -0.0075, -0.0165, -0.0226,  0.0177,
        -0.0594,  0.0177,  0.0218, -0.0425,  0.0102, -0.0680, -0.0563,  0.0105,
        -0.0057, -0.0605, -0.0091, -0.0444, -0.0609,  0.0065, -0.0055,  0.0011,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0208,  0.0087,  0.0079, -0.0031, -0.0277, -0.0139, -0.0430, -0.0536,
         0.0051, -0.0157,  0.0042, -0.0009, -0.0315, -0.0279,  0.0043,  0.0321,
        -0.0562, -0.0548, -0.0409, -0.0815, -0.0119, -0.0178, -0.0684, -0.0759,
        -0.0217, -0.0358, -0.0505, -0.0130, -0.0210, -0.0884, -0.0146, -0.0355,
         0.0120,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0269,  0.4381,  0.0047,  0.0950, -0.0098,  0.0065, -0.0181,  0.0111,
         0.1103,  0.0063, -0.0163, -0.0027, -0.0102,  0.0803,  0.0161,  0.0573,
         0.0355,  0.0330,  0.0218,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0027, -0.0012, -0.0069, -0.0169, -0.1118,  0.0189, -0.0974, -0.0910,
        -0.2061, -0.0107, -0.0025,  0.0099, -0.0083,  0.0265,  0.0055,  0.0015,
        -0.0275, -0.0720, -0.0491, -0.0096, -0.0496, -0.1199, -0.0072,  0.0179,
         0.0294,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0126, -0.2568, -0.1211,  0.0030, -0.0372, -0.0868, -0.2837,  0.0294,
        -0.0081, -0.0384, -0.0152,  0.0083, -0.0661,  0.0191, -0.0015, -0.0080,
         0.0047,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0617,  0.0082, -0.0180,  0.0037,  0.0061, -0.0115,  0.0059,  0.0073,
         0.0047, -0.0862, -0.0123,  0.0107,  0.0202,  0.0047,  0.0013, -0.0269,
        -0.1175, -0.1413,  0.0116, -0.0803,  0.0031, -0.1119, -0.1281,  0.0107,
        -0.0093,  0.0060,  0.0005, -0.0007,  0.0076, -0.0278, -0.0381,  0.0162,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.1132e-02, -9.8957e-02, -1.2686e-01,  1.1122e-02,  9.4244e-03,
        -3.3124e-03, -6.2070e-02, -1.1912e-01, -5.0155e-03,  1.3527e-02,
         7.2250e-04,  2.2173e-04, -7.6401e-03,  9.3706e-05, -1.9798e-02,
         1.5527e-02, -1.5543e-02,  8.9774e-03, -5.9732e-02, -2.2323e-02,
        -5.9874e-02, -1.1258e-03, -8.6542e-02, -1.0156e-01, -2.3770e-02,
        -2.7635e-02, -3.0803e-02,  1.7568e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0068, -0.0334,  0.0005,  0.0195, -0.0322, -0.1617, -0.2026, -0.0516,
         0.0032, -0.0266, -0.0152, -0.0936, -0.0092,  0.0008, -0.0079, -0.0809,
        -0.0163, -0.0971,  0.0012,  0.0148, -0.0104,  0.0029,  0.0004, -0.0025,
        -0.0523,  0.0057,  0.0189,  0.0090, -0.0227,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0288, -0.1630, -0.0035, -0.0648, -0.1597, -0.0092, -0.0103,  0.0037,
        -0.0039,  0.0133, -0.0043, -0.0115, -0.0702, -0.0868,  0.0230, -0.0255,
        -0.0130, -0.0977, -0.0027, -0.0302,  0.0118,  0.0042, -0.0398,  0.0056,
        -0.0052, -0.0573, -0.0358,  0.0003,  0.0007, -0.0018, -0.0063,  0.0033,
        -0.0028,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0111, -0.2324, -0.0150, -0.1183, -0.0125, -0.0471,  0.0082, -0.0601,
        -0.0111, -0.0795, -0.0728,  0.0066,  0.0082,  0.0049, -0.0119,  0.0065,
        -0.0463,  0.0035, -0.1073,  0.0104, -0.0562, -0.0105,  0.0453,  0.0144,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0743, -0.1175, -0.1365, -0.0051, -0.0407,  0.0044, -0.0126, -0.0448,
         0.0140, -0.0483, -0.0074, -0.0626, -0.0991, -0.0014, -0.0070, -0.0060,
         0.0020, -0.0409, -0.0052, -0.1261, -0.0551,  0.0623,  0.0266,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.6797e-02, -1.8847e-01, -1.0298e-02, -3.3912e-03, -4.2813e-02,
        -4.7990e-02, -2.1432e-03, -4.3376e-03,  5.3711e-03,  4.6724e-03,
         3.2490e-03,  1.0145e-03, -1.5554e-02,  3.9995e-04,  6.9989e-04,
        -1.7723e-02, -7.3480e-02, -5.6337e-02, -3.6672e-02,  2.0911e-02,
         4.2624e-05, -4.9152e-02, -6.8269e-03, -9.1485e-03, -5.2545e-02,
        -1.6214e-02, -5.2205e-03, -2.0776e-03, -1.0181e-02, -5.5803e-02,
        -4.5818e-02, -1.5003e-03, -1.6238e-02, -1.5701e-02, -4.2374e-02,
        -6.0835e-02, -2.7668e-03, -1.4088e-02, -1.1148e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0242, -0.0492, -0.0046, -0.0233, -0.1097, -0.0102,  0.0080,  0.0074,
        -0.0092,  0.0251,  0.0060,  0.0015,  0.0026,  0.0049,  0.0054, -0.0598,
        -0.0068, -0.0105,  0.0035, -0.0061, -0.0698,  0.0007, -0.0040, -0.0216,
        -0.1071, -0.0759, -0.0171, -0.0506, -0.0128, -0.0820, -0.0164, -0.0017,
        -0.0081, -0.0046, -0.0071,  0.0046, -0.0272, -0.0709,  0.0084,  0.0023,
        -0.0045, -0.0196,  0.0051,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.1203e-02, -2.4209e-01, -1.7060e-02,  2.4781e-03, -2.5975e-02,
         1.2704e-03,  1.1141e-04,  1.1394e-03, -7.0943e-02,  2.1947e-02,
         4.8433e-03, -1.6900e-03, -2.8380e-02, -2.8437e-03,  5.7233e-04,
         3.8000e-03, -3.3729e-02, -5.1903e-02,  3.0250e-03, -4.3762e-03,
         1.4814e-02,  7.1056e-04,  4.7446e-03,  1.8979e-03,  6.7997e-03,
         6.3407e-04,  1.6251e-02, -6.2976e-04,  1.4725e-03,  4.1891e-03,
        -7.3088e-03, -5.4513e-02, -7.6304e-02, -4.8238e-03, -5.3791e-03,
        -7.7484e-03,  3.6173e-03, -5.1073e-02, -4.2806e-02, -7.3589e-03,
        -5.4277e-02, -4.1304e-02, -4.7903e-03, -1.0851e-03,  3.7406e-03,
         1.1652e-03, -1.9632e-03,  1.3690e-03,  1.5557e-03, -1.3603e-03,
         4.4081e-03,  1.2778e-03, -1.5533e-03,  3.6262e-03, -2.3099e-03,
        -8.1092e-03, -1.3646e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0225, -0.4715,  0.0206, -0.0163, -0.0599,  0.0036, -0.0957, -0.0853,
         0.0023, -0.0220, -0.0639, -0.0086, -0.0625,  0.0654,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0036, -0.1904, -0.0346,  0.0114, -0.0094, -0.0616,  0.0003,  0.0009,
        -0.0401,  0.0078, -0.0497, -0.0757, -0.0053, -0.0359, -0.0636, -0.0029,
        -0.0267, -0.0351,  0.0178, -0.0657, -0.1327, -0.0261, -0.0452,  0.0060,
         0.0099,  0.0037, -0.0041,  0.0062, -0.0242, -0.0033,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0525, -0.1754,  0.1031, -0.3365, -0.2570, -0.0515,  0.0077,  0.0162,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.0186, -0.1668, -0.0182, -0.0360, -0.0727, -0.1466,  0.0015,  0.0022,
         0.0016, -0.0019, -0.0039, -0.0047, -0.0855, -0.0857, -0.0096, -0.0155,
        -0.0078,  0.0064, -0.0098, -0.0011,  0.0059,  0.0088,  0.0033,  0.0034,
        -0.0381,  0.0079, -0.0391, -0.0019, -0.0504, -0.0988, -0.0146,  0.0100,
         0.0221,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0012, -0.1937, -0.0762, -0.0298, -0.0146, -0.0370, -0.0166,  0.0031,
         0.0156, -0.0268, -0.0076, -0.0185, -0.0146,  0.0049,  0.0085,  0.0066,
         0.0003,  0.0055, -0.0229,  0.0044,  0.0012, -0.0003,  0.0027,  0.0030,
         0.0010,  0.0120,  0.0045, -0.0637,  0.0059,  0.0007, -0.0027,  0.0106,
        -0.0196, -0.0185, -0.0060, -0.0110, -0.0257, -0.0015, -0.0015, -0.0012,
        -0.0014,  0.0037,  0.0008, -0.0601, -0.0075,  0.0043, -0.0018,  0.0005,
        -0.0029, -0.0058, -0.0086, -0.0150, -0.0070, -0.0136, -0.0034,  0.0050,
        -0.0008, -0.0109, -0.0032, -0.0392, -0.0068, -0.0127, -0.0396,  0.0016,
         0.0051, -0.0013, -0.0104,  0.0069,  0.0012, -0.0068, -0.0103],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0373, -0.0113, -0.0453,  0.0009, -0.0549, -0.0180, -0.1208, -0.0046,
         0.0071,  0.0127,  0.0096, -0.0035, -0.0048,  0.0052,  0.0067, -0.0011,
         0.0013,  0.0041,  0.0119, -0.1012, -0.0134,  0.0071,  0.0045, -0.0551,
         0.0004, -0.0701, -0.0068,  0.0061, -0.0049,  0.0007, -0.0008, -0.0049,
        -0.0914, -0.0697, -0.0304, -0.0088, -0.0636, -0.0586,  0.0042,  0.0018,
         0.0255,  0.0092,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.0726e-02, -3.2077e-01,  4.7704e-03,  6.2196e-04, -5.8235e-03,
        -5.4831e-03,  9.6091e-03,  2.3489e-02, -1.7006e-02, -4.1153e-02,
        -1.7402e-02, -5.2085e-02, -7.3313e-02, -2.9464e-03,  6.2612e-04,
         6.9852e-04, -3.7171e-03, -6.8927e-02, -5.5130e-02,  3.2038e-03,
        -1.8035e-02, -2.3972e-02, -2.4690e-03, -5.3842e-03, -8.4340e-03,
        -4.2858e-02,  7.0942e-05,  6.2864e-03, -6.9907e-02, -4.0360e-03,
         5.0881e-03, -7.1084e-03,  6.4655e-03,  1.1583e-02,  4.0368e-03,
         9.7634e-03,  3.6998e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0338, -0.3878, -0.0580, -0.2110,  0.0101,  0.0047,  0.0040,  0.0107,
         0.0461,  0.0079, -0.0888,  0.0416, -0.0757, -0.0120, -0.0077,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0102, -0.0466, -0.0153, -0.0033, -0.1314, -0.1871, -0.0073, -0.1521,
        -0.2092, -0.0271, -0.0260,  0.0076, -0.0318,  0.0065,  0.0479,  0.0008,
         0.0080,  0.0098, -0.0030,  0.0691,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0078, -0.0067, -0.0151,  0.0037, -0.0985,  0.0006, -0.0008, -0.0163,
         0.0136,  0.0038, -0.0573, -0.0639,  0.0034, -0.0040, -0.0307, -0.0304,
         0.0006, -0.0227,  0.0096, -0.0281,  0.0030,  0.0012,  0.0028, -0.0019,
         0.0068, -0.0772, -0.1275,  0.0180,  0.0014, -0.0555, -0.0865, -0.0056,
         0.0123,  0.0083, -0.0434, -0.0806,  0.0013, -0.0272,  0.0046, -0.0017,
         0.0044,  0.0106, -0.0006,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0497, -0.0012, -0.0059, -0.0154, -0.0532, -0.0163, -0.0431, -0.0891,
        -0.0491, -0.0141,  0.0110, -0.0160, -0.0127, -0.0590, -0.1030, -0.0042,
         0.0128,  0.0076,  0.0019, -0.0042, -0.0792, -0.0097,  0.0050,  0.0080,
        -0.0031, -0.0118, -0.0235, -0.0078, -0.0733, -0.0019,  0.0104, -0.0538,
        -0.0078, -0.0384, -0.0563, -0.0140, -0.0013, -0.0253,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0054, -0.2670, -0.1346, -0.0970, -0.0016, -0.0457,  0.0042,  0.0184,
        -0.0472, -0.0896, -0.0117, -0.0148, -0.0663,  0.0022,  0.0055, -0.0024,
         0.0062, -0.0535, -0.0206,  0.0237,  0.0541,  0.0282,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0204, -0.1558, -0.0201, -0.0028,  0.0109, -0.0125, -0.0108,  0.0208,
         0.0197, -0.0940,  0.0077,  0.0041, -0.0023, -0.0061, -0.0398, -0.0523,
        -0.0721, -0.0081, -0.0089, -0.0583, -0.0092,  0.0017, -0.0370, -0.0082,
        -0.0149,  0.0090,  0.0030,  0.0080, -0.0088, -0.0546,  0.0006, -0.0266,
         0.0033,  0.0125, -0.0323, -0.0603,  0.0074, -0.0098,  0.0355,  0.0297,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0747, -0.0011, -0.0041, -0.0166, -0.0017, -0.0445, -0.0452,  0.0145,
        -0.0671, -0.0039,  0.0037, -0.0113, -0.0356, -0.1896, -0.0992, -0.0096,
         0.0139, -0.0047, -0.0794,  0.0058,  0.0018,  0.0098,  0.0139,  0.0041,
        -0.0082, -0.0527,  0.0207, -0.0655, -0.0632, -0.0058, -0.0192, -0.0089,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.6320e-03, -3.7123e-01, -1.8694e-02, -8.6235e-05, -6.3875e-03,
         5.4373e-03, -1.5110e-02, -9.2936e-02, -1.2328e-01,  2.4030e-03,
        -4.5681e-03, -3.0672e-02,  1.0384e-02,  3.3120e-03,  2.0653e-02,
         8.5809e-03, -5.5307e-02,  2.5136e-03,  7.9323e-03,  1.1535e-02,
        -4.0344e-03, -3.0171e-03, -3.6961e-03, -6.1846e-02, -3.2087e-03,
        -4.8528e-03, -4.0761e-02, -3.2728e-03, -2.9629e-03, -1.3848e-02,
         9.1824e-03,  1.4722e-02, -4.9161e-03,  2.5056e-03, -3.2525e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 0.0051, -0.0077, -0.1235, -0.1334, -0.0266,  0.0009, -0.1476, -0.1389,
        -0.0090, -0.1984,  0.0044, -0.0399, -0.0055, -0.0635, -0.0109, -0.0080,
        -0.0336, -0.0244,  0.0043, -0.0113,  0.0032,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.4378e-03, -3.7033e-01, -2.6726e-04, -6.9639e-02, -1.2348e-01,
        -6.7510e-02, -4.0411e-02, -6.4641e-02, -9.5873e-02,  2.6292e-02,
         2.2988e-03, -1.7887e-03,  1.2701e-02,  4.5596e-03, -7.9106e-02,
        -7.9593e-03, -2.3722e-02, -2.9349e-04, -1.6881e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0220,  0.0023, -0.1304, -0.0039, -0.0587, -0.0145,  0.0039,  0.0224,
        -0.0043, -0.0495, -0.1336,  0.0026, -0.1376, -0.2207, -0.0025, -0.0598,
        -0.0308, -0.0416,  0.0419, -0.0171,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.1507e-03, -3.5452e-01,  8.0506e-03, -5.9273e-02, -1.1158e-02,
        -3.0395e-04,  2.4991e-02, -5.0747e-02, -9.3310e-03,  3.0176e-03,
        -7.0145e-03, -2.3634e-02, -1.1541e-03, -1.2018e-01, -2.1612e-02,
        -9.9092e-02, -8.7400e-02, -2.5799e-03, -4.0015e-02, -5.3753e-02,
         6.3837e-04, -8.7510e-03, -7.6307e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0144, -0.2332, -0.0430,  0.0003, -0.0250, -0.0777, -0.0169, -0.0498,
        -0.1247, -0.1029, -0.0060, -0.0716,  0.0250, -0.0664, -0.0061,  0.0056,
        -0.0005, -0.0094, -0.0212, -0.0126, -0.0431, -0.0096,  0.0184,  0.0011,
         0.0155,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0702, -0.3435, -0.0068, -0.0159,  0.0118, -0.0080, -0.0022,  0.0114,
         0.0127, -0.0107, -0.0570,  0.0036,  0.0041, -0.0033,  0.0076, -0.0014,
         0.0027,  0.0010,  0.0038, -0.0142, -0.0975, -0.0954, -0.0244, -0.0232,
        -0.0192, -0.0225,  0.0020,  0.0017, -0.0115, -0.0121, -0.0315,  0.0137,
        -0.0261,  0.0008, -0.0266,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0563, -0.0099,  0.0217, -0.1543, -0.0138, -0.0204, -0.1022, -0.0255,
        -0.0970, -0.0081, -0.0059, -0.0283, -0.0139, -0.0814, -0.0637, -0.1181,
        -0.0058, -0.0901,  0.0420,  0.0108, -0.0272, -0.0036,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1395, -0.0089, -0.0447,  0.0175, -0.0320, -0.0447, -0.0178, -0.0033,
        -0.0054, -0.3173,  0.0151,  0.0254,  0.0399,  0.0244, -0.0607, -0.0871,
        -0.0345, -0.0068, -0.0239, -0.0250, -0.0260,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.9721e-02, -3.7254e-01, -8.2375e-03,  7.8632e-03,  1.5984e-04,
         1.1718e-02, -7.9197e-02, -7.0653e-02,  1.3211e-02, -1.6020e-03,
         2.3003e-02,  3.3209e-02, -2.7844e-03, -1.3986e-01,  6.0999e-02,
        -5.7116e-02,  5.4077e-03, -4.3679e-02,  1.9042e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.5290e-03, -5.3299e-03, -2.6874e-02,  5.0711e-03, -1.0187e-02,
        -8.0665e-02,  5.7764e-03,  1.5726e-03,  4.6532e-04, -1.7661e-03,
        -2.4856e-03, -1.0110e-02,  5.6970e-04, -3.9475e-02,  1.4967e-03,
        -1.7559e-02, -5.8762e-02, -6.5163e-02,  7.0529e-03, -3.0552e-02,
        -9.2583e-03, -2.3587e-02, -2.8663e-02,  9.9485e-05, -6.7243e-02,
        -6.6986e-03, -4.5475e-03, -1.7560e-02,  9.1215e-03, -1.2343e-01,
        -1.2594e-02,  4.0393e-03, -5.3970e-02, -1.2815e-01,  6.2867e-04,
        -5.0879e-03,  1.3525e-04,  2.3113e-03, -1.2558e-02, -3.8713e-02,
        -3.1848e-03,  5.6744e-03, -1.2792e-03,  1.1344e-03, -2.6841e-03,
        -1.4661e-02, -4.6523e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0085, -0.2922, -0.1456, -0.0146,  0.0017, -0.0010,  0.0008,  0.0005,
        -0.0483,  0.0117,  0.0092,  0.0071,  0.0015,  0.0117,  0.0110, -0.0005,
        -0.0883, -0.0027,  0.0056, -0.0031,  0.0013,  0.0116, -0.0653, -0.1329,
         0.0110,  0.0160, -0.0124,  0.0014,  0.0025,  0.0100,  0.0081,  0.0171,
        -0.0288,  0.0158,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.3276e-02, -2.6649e-01,  1.9705e-02,  6.4342e-04, -3.6581e-02,
        -5.4012e-02,  1.3582e-02,  1.0400e-02,  5.5341e-03, -3.6215e-02,
         1.1712e-02,  7.9873e-03, -5.8023e-02, -9.5770e-02,  7.3756e-03,
        -1.2524e-02,  7.7291e-03,  1.9227e-04, -3.4372e-03, -9.6779e-02,
        -1.7649e-02, -1.7873e-02, -5.7468e-03, -1.1286e-02, -4.2267e-03,
         1.1565e-02,  1.7120e-03, -1.4991e-02,  5.0367e-03, -4.1137e-02,
         1.1544e-02,  2.1431e-02, -5.0712e-02, -1.7127e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0191, -0.4888, -0.0172,  0.0012, -0.0254, -0.0890,  0.0056, -0.0060,
        -0.0156, -0.0997,  0.0057, -0.0062, -0.0458, -0.1217, -0.0103, -0.0200,
         0.0006, -0.0222,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0660, -0.0019, -0.0091, -0.0025, -0.0064,  0.0109, -0.0056, -0.0094,
        -0.1566, -0.2718,  0.0232, -0.0109, -0.0354, -0.0137, -0.0013, -0.0004,
        -0.0277,  0.0428, -0.0140, -0.1648, -0.0184, -0.0006, -0.0247,  0.0127,
         0.0065, -0.0021,  0.0607,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.6797e-02,  1.3726e-02,  1.4600e-02, -1.4667e-03,  5.8733e-02,
        -8.0009e-04,  2.5708e-03, -3.0634e-02, -1.5218e-01, -1.6018e-02,
        -2.3600e-03, -5.5235e-03, -2.1891e-01, -2.2378e-02,  2.7714e-03,
        -1.0626e-03, -1.9599e-02, -2.4551e-02, -1.9440e-02, -1.9584e-01,
         6.1834e-03, -1.6911e-02, -5.0493e-02, -1.6055e-04, -3.4329e-02,
        -4.1958e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0029, -0.3093, -0.0052, -0.0930, -0.0943,  0.0075,  0.0028,  0.0170,
        -0.0667, -0.0394, -0.1144, -0.0815,  0.0118,  0.0039,  0.0062, -0.0537,
        -0.0127, -0.0251, -0.0158, -0.0025,  0.0346,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0667,  0.2416,  0.1607,  0.1657,  0.0577,  0.0384,  0.1399,  0.0101,
         0.0106, -0.0005,  0.0019,  0.0614,  0.0240,  0.0086, -0.0122,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0520, -0.0111, -0.0150, -0.0169,  0.0357, -0.0117, -0.0065, -0.0207,
         0.0027,  0.0135, -0.0319, -0.0325,  0.4795,  0.1981,  0.0101, -0.0071,
        -0.0054, -0.0496,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0390, -0.2720, -0.0868, -0.0814, -0.0204, -0.0203, -0.0288, -0.0899,
        -0.0102, -0.0723, -0.1022, -0.0487,  0.0365, -0.0553, -0.0094,  0.0148,
        -0.0044, -0.0029,  0.0048,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0031, -0.1209, -0.1647, -0.0028,  0.0002, -0.0215,  0.0197, -0.0012,
        -0.0258,  0.0012,  0.0049,  0.0008, -0.0260,  0.0053,  0.0055,  0.0150,
        -0.0360,  0.0078,  0.0105, -0.0060, -0.0665,  0.0076, -0.0422, -0.0046,
        -0.0551, -0.0062, -0.0358,  0.0148,  0.0058, -0.0250,  0.0044,  0.0096,
        -0.0384, -0.0072, -0.0420, -0.0270, -0.0619,  0.0125, -0.0268, -0.0007,
         0.0034, -0.0002,  0.0083,  0.0151], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0181, -0.3188, -0.0265, -0.0956, -0.0558, -0.0133, -0.0181, -0.0107,
        -0.0643,  0.0008, -0.0365,  0.0107,  0.0151,  0.0028,  0.0102, -0.0010,
        -0.0057, -0.0837, -0.0039, -0.0250, -0.0374,  0.0103,  0.0051,  0.0055,
         0.0139,  0.0042, -0.0059, -0.0369, -0.0007, -0.0054,  0.0068, -0.0173,
         0.0165, -0.0118, -0.0056,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0355, -0.4214, -0.0031, -0.0514, -0.0022,  0.0256, -0.1222, -0.0658,
        -0.0301,  0.0092,  0.0122, -0.0265, -0.0668, -0.0019,  0.0104, -0.0402,
         0.0025, -0.0059,  0.0073,  0.0018,  0.0013, -0.0201,  0.0237,  0.0129,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0177, -0.2438, -0.0167, -0.0440, -0.0224, -0.0321,  0.0097, -0.0096,
        -0.0714, -0.0012, -0.0265,  0.0086,  0.0223, -0.0204, -0.0427,  0.0055,
         0.0238, -0.0019, -0.0502, -0.0230, -0.0016, -0.0469,  0.0033, -0.0203,
         0.0092, -0.0342, -0.0609, -0.0279,  0.0030, -0.0157, -0.0242,  0.0059,
        -0.0424,  0.0111,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0262,  0.0250,  0.0146, -0.0026,  0.0163,  0.0188, -0.0848,  0.0016,
         0.0342,  0.0003,  0.0152, -0.0096, -0.1207,  0.0085, -0.1750, -0.1532,
        -0.0119, -0.0146, -0.0123,  0.0040, -0.0033,  0.0109,  0.0063, -0.0575,
        -0.0390, -0.0122, -0.0929, -0.0047, -0.0024, -0.0217,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 3.4555e-02,  2.8217e-03, -6.4043e-02, -2.8479e-01, -1.2480e-02,
        -2.5726e-04, -2.2997e-03, -1.2172e-02, -1.7293e-01, -3.6975e-01,
        -6.6554e-03,  1.7062e-02, -2.0181e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0387,  0.0010,  0.0705, -0.0502,  0.0167, -0.0391, -0.0400, -0.1633,
        -0.2063, -0.0073, -0.0201, -0.0893,  0.0080, -0.0462, -0.1013,  0.0262,
         0.0596,  0.0160,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0322, -0.2046, -0.3220, -0.0129, -0.2626,  0.0786, -0.0461, -0.0409,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.2702e-05, -1.7870e-01, -1.2025e-02, -4.0778e-02, -3.1480e-02,
        -2.7238e-03, -8.2431e-03, -1.5145e-03,  2.5009e-03, -1.9535e-02,
         1.3240e-02,  1.4869e-02, -2.2550e-02,  5.0805e-03,  1.4730e-03,
        -7.8861e-03,  4.7713e-04, -9.6161e-04,  2.8638e-03,  8.1454e-06,
         1.0669e-02,  1.0436e-02, -3.0894e-02, -3.0279e-02,  9.4576e-03,
        -3.1022e-02, -9.3309e-03, -3.8306e-03, -1.0704e-01, -3.5925e-02,
        -1.6813e-03, -3.4596e-02, -1.2982e-02, -9.9400e-02, -1.1525e-02,
        -2.3430e-02, -5.1665e-02, -1.6765e-02,  2.9572e-03,  7.4438e-04,
        -1.4327e-02, -1.2090e-02, -1.3139e-02,  8.5516e-03,  6.1999e-03,
        -1.0286e-02,  3.3801e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0779, -0.0108,  0.0037, -0.6379, -0.0509, -0.0093,  0.0656,  0.0197,
        -0.0650, -0.0283,  0.0310,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0244, -0.3719, -0.0630, -0.0571, -0.0151,  0.0181, -0.0615, -0.1721,
        -0.0189, -0.0631, -0.0147,  0.0181, -0.0517, -0.0005,  0.0123,  0.0132,
        -0.0015, -0.0228,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1046, -0.2016, -0.3318,  0.0098,  0.0058,  0.0187, -0.1687,  0.0042,
        -0.0830, -0.0234,  0.0039,  0.0050,  0.0059,  0.0281, -0.0056,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0334, -0.3985,  0.0526, -0.0271, -0.0245, -0.0839, -0.0913,  0.0097,
        -0.0256, -0.0838, -0.1157, -0.0057,  0.0027, -0.0223, -0.0232,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0236,  0.0028,  0.0133,  0.0044, -0.0023, -0.0133, -0.0146, -0.0183,
        -0.1086, -0.0279, -0.0039, -0.1148, -0.1544,  0.0007, -0.0011,  0.0014,
        -0.0564, -0.0551,  0.0272,  0.0020, -0.1207, -0.1554,  0.0284, -0.0279,
        -0.0151, -0.0062,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.4218e-02, -2.6027e-02,  1.4587e-02,  4.4379e-04,  1.2054e-02,
        -1.0637e-03, -4.9090e-03, -2.0884e-01, -5.0956e-01, -1.6447e-02,
         6.9196e-03, -2.7888e-03,  3.0999e-02,  2.3414e-02, -5.7723e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0468, -0.3157, -0.0038, -0.0545, -0.0863, -0.0138, -0.0614, -0.0141,
        -0.0105, -0.1459,  0.0106,  0.0053, -0.0428,  0.0046, -0.0003, -0.0100,
        -0.0275,  0.0051, -0.0266,  0.0113, -0.0272,  0.0170,  0.0097,  0.0083,
         0.0159,  0.0251,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0387,  0.0086,  0.0074,  0.0171,  0.0127, -0.0345,  0.0041,  0.0096,
        -0.1011, -0.0630, -0.1706, -0.0320, -0.0253, -0.0039, -0.1269, -0.0907,
         0.0162, -0.0575, -0.0008, -0.0803, -0.0252, -0.0208, -0.0026, -0.0227,
         0.0223, -0.0055,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([-8.5144e-04, -3.2423e-01,  5.5683e-03, -1.2751e-03, -9.8038e-03,
        -9.4054e-03, -3.0692e-02, -9.3039e-02, -6.4200e-03, -3.7996e-02,
        -2.8530e-02, -5.6675e-02, -7.3501e-02, -1.0953e-03, -3.7151e-02,
        -1.1512e-02, -4.3798e-03, -2.1353e-03, -1.5670e-02, -1.3253e-02,
         1.3575e-02, -1.9942e-02,  1.2592e-02, -3.5616e-02, -5.8359e-02,
         1.2298e-02, -9.9184e-03,  7.8239e-03,  1.9616e-02, -2.5299e-03,
        -6.2177e-03, -5.6341e-03,  9.5898e-05, -5.7142e-03,  1.9001e-02,
         7.8841e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.1991e-02, -1.2910e-01, -5.0011e-02, -5.1980e-03, -3.7715e-02,
        -7.9717e-03, -8.0636e-02, -7.3676e-02, -1.8126e-03, -8.9279e-03,
        -4.9959e-02, -1.3762e-03,  2.2343e-04,  8.9402e-03,  4.4732e-03,
        -3.7316e-02, -3.8897e-03, -1.4449e-03,  3.5392e-03, -1.4925e-02,
         1.8166e-02, -5.8665e-02,  5.2201e-05, -5.7380e-05, -2.3302e-02,
        -6.0346e-02, -1.3350e-03, -4.3014e-02, -6.4029e-02, -5.2926e-03,
        -8.2923e-03, -2.9719e-02, -5.5349e-02,  5.8967e-03,  4.0794e-02,
        -2.5654e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0021, -0.0379, -0.0067, -0.0167,  0.0231, -0.2027, -0.0544, -0.0033,
        -0.0176,  0.0202, -0.0007, -0.1281, -0.3004,  0.0803,  0.0160, -0.0251,
         0.0088,  0.0217, -0.0112, -0.0229,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0203, -0.1976,  0.0163, -0.0055,  0.0139,  0.0123, -0.0019, -0.0515,
        -0.0104, -0.0311,  0.0044, -0.0059, -0.0051, -0.0059, -0.0876, -0.0579,
        -0.0184, -0.0711, -0.0053, -0.0037,  0.0171, -0.0065,  0.0050,  0.0098,
         0.0348,  0.0176, -0.0005, -0.0067,  0.0046, -0.0489, -0.0650, -0.0231,
        -0.0161, -0.0573, -0.0032,  0.0287,  0.0218, -0.0073,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4563e-02, -2.5408e-01, -9.5105e-02, -6.5634e-02, -6.0104e-03,
         1.4272e-02, -8.1038e-04,  2.7232e-02, -8.2649e-03,  3.1133e-03,
        -1.9218e-02, -7.1796e-02, -1.6497e-03,  1.8011e-03, -5.6372e-03,
        -4.8649e-02, -4.8662e-04, -6.1676e-02, -1.2753e-02, -2.2473e-04,
         5.9358e-03,  1.3265e-02,  3.5246e-03,  6.4327e-03,  2.2818e-03,
         9.7427e-03, -3.1472e-02, -4.7545e-02, -7.0244e-02, -3.8048e-03,
        -7.7955e-03, -5.0818e-03,  6.3172e-03,  1.2534e-03, -2.4706e-02,
         4.9788e-03, -6.8751e-03,  6.7621e-03,  1.2822e-02, -8.7742e-03,
         7.4071e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0560, -0.3740,  0.0123,  0.0067,  0.0036, -0.0714, -0.0075, -0.0060,
        -0.0269, -0.0019,  0.0134, -0.0618, -0.1254, -0.0287,  0.0100, -0.0888,
         0.0173, -0.0677, -0.0058,  0.0060,  0.0019, -0.0070,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0112,  0.0034, -0.0006,  0.0013,  0.0161, -0.0534,  0.0062,  0.0095,
        -0.0143, -0.0040, -0.0018, -0.0925, -0.0043, -0.0596, -0.0017, -0.0691,
         0.0041, -0.0584,  0.0041, -0.0402, -0.1158,  0.0196,  0.0053, -0.0527,
        -0.0045,  0.0075, -0.0059, -0.0431, -0.1130, -0.0237, -0.0466, -0.0396,
         0.0151,  0.0096, -0.0425,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0235, -0.0503, -0.2066, -0.0262, -0.1305, -0.0071,  0.0108,  0.0336,
         0.0460,  0.0020, -0.0932,  0.0154,  0.0041,  0.0068,  0.0182,  0.0191,
         0.0137, -0.0193, -0.0766,  0.0149, -0.0592,  0.0099,  0.0095, -0.0081,
        -0.0062,  0.0020, -0.0702,  0.0053, -0.0117,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0464, -0.2498, -0.1139, -0.0649, -0.0285, -0.0552, -0.0136, -0.0136,
        -0.0297,  0.0023, -0.0382, -0.0525, -0.0286, -0.0080,  0.0023, -0.0566,
        -0.0595, -0.0079, -0.0280,  0.0140, -0.0140, -0.0006, -0.0417,  0.0152,
         0.0148,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0041, -0.0042, -0.0028,  0.0208, -0.1744,  0.0029,  0.0039, -0.1290,
        -0.1859, -0.0151, -0.0121,  0.0376,  0.0042,  0.0618, -0.1056, -0.1385,
         0.0070, -0.0482, -0.0006, -0.0027, -0.0054, -0.0171, -0.0161,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0683, -0.0100,  0.0068,  0.0293,  0.0059,  0.0152,  0.0111,  0.0038,
        -0.0020, -0.0042, -0.0066, -0.0069, -0.1099, -0.1861, -0.0025, -0.0160,
         0.0005, -0.0435,  0.0107,  0.0068,  0.0047,  0.0098,  0.0146, -0.0054,
        -0.0191,  0.0032, -0.0968, -0.0045, -0.1392,  0.0021, -0.0315,  0.0132,
        -0.0088, -0.0237, -0.0528, -0.0010, -0.0095,  0.0091,  0.0046,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0902, -0.0236, -0.0663, -0.1616, -0.2424,  0.0513, -0.0175, -0.1424,
        -0.1522,  0.0254,  0.0152, -0.0121,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0191, -0.1242, -0.2291, -0.0142, -0.0038, -0.0094, -0.0512,  0.0141,
        -0.0042, -0.0347,  0.0037,  0.0156, -0.0169, -0.0022, -0.0483,  0.0022,
        -0.0122, -0.0210,  0.0013, -0.0095, -0.0023, -0.0116,  0.0125, -0.0017,
        -0.0021,  0.0165,  0.0011,  0.0162,  0.0053, -0.0483, -0.0144,  0.0281,
        -0.0730, -0.0005,  0.0130, -0.0006, -0.0366, -0.0050,  0.0055, -0.0009,
         0.0014,  0.0039, -0.0019,  0.0377,  0.0233,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0142, -0.3926, -0.0415, -0.1272, -0.0687,  0.0014,  0.0055, -0.0482,
        -0.1621, -0.0362, -0.0716,  0.0309,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0714, -0.6492, -0.0320, -0.1948, -0.0099,  0.0298, -0.0074,  0.0056,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.3695e-02, -1.3057e-01, -4.4805e-03, -2.8196e-02, -3.8932e-02,
         4.1435e-03,  4.7891e-03, -5.7772e-03, -2.8980e-02, -9.4285e-02,
        -5.6475e-03, -8.9276e-03, -3.0962e-03, -2.4011e-03,  1.7780e-03,
         6.6642e-03, -2.1408e-03, -1.1080e-03,  3.3380e-03,  1.9275e-03,
         1.2234e-03,  5.1304e-03, -2.5755e-03, -5.1534e-03, -3.4084e-03,
        -7.3728e-03, -7.2038e-04, -9.4584e-03, -2.1285e-02, -2.1196e-03,
        -2.9279e-03, -4.2811e-03, -2.9151e-05, -1.6546e-04, -3.2325e-03,
         2.4680e-03,  3.3749e-03,  5.6370e-03,  4.3762e-03,  2.4059e-03,
        -2.7146e-03, -1.7063e-03, -7.2377e-03,  6.9799e-04, -3.7698e-03,
         6.3010e-03, -8.8210e-02, -7.5642e-03, -8.6678e-02, -1.6663e-01,
        -2.5681e-02, -1.2620e-02, -3.3841e-02, -5.5641e-03, -1.4889e-02,
        -3.3377e-03, -7.6121e-03, -1.0549e-02, -8.3326e-04, -4.3127e-03,
         2.1028e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0077,  0.0084,  0.0042, -0.0206, -0.0761, -0.0022, -0.0567,  0.0025,
        -0.0022, -0.0046, -0.0234, -0.0998,  0.0107, -0.0947, -0.0781,  0.0076,
        -0.0092,  0.0064, -0.0305, -0.0272, -0.0722, -0.1183, -0.0157, -0.0194,
        -0.0346, -0.0325, -0.0235, -0.0207, -0.0329, -0.0244, -0.0027, -0.0063,
        -0.0242,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0248,  0.0052, -0.1276, -0.0853, -0.0182,  0.0214, -0.0574,  0.0046,
         0.0053,  0.0042, -0.0067,  0.0006, -0.0498, -0.0144, -0.0223,  0.0281,
         0.0179, -0.1151, -0.2287,  0.0284, -0.0155, -0.0100,  0.0448, -0.0638,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0369, -0.1015, -0.0790, -0.0423,  0.0134, -0.0886,  0.0268, -0.0694,
        -0.0156, -0.0804, -0.0778, -0.0837, -0.1483,  0.0030,  0.0104, -0.0033,
         0.0028, -0.0227,  0.0255,  0.0025, -0.0378, -0.0091, -0.0037,  0.0157,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0269,  0.0016, -0.1048, -0.0768,  0.0105,  0.0044,  0.0023, -0.0671,
         0.0052,  0.0227,  0.0083,  0.0057,  0.0080,  0.0038, -0.0624, -0.0781,
        -0.0884,  0.0224, -0.0013,  0.0043, -0.0749,  0.0129, -0.0416, -0.0788,
         0.0030,  0.0037, -0.0320, -0.0007,  0.0061, -0.0283, -0.0214, -0.0066,
         0.0586, -0.0264,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0043,  0.0195,  0.0116, -0.0193, -0.0995, -0.0022, -0.0134, -0.0006,
        -0.0086, -0.0057,  0.0189,  0.0479, -0.0068, -0.2599, -0.0064,  0.0005,
        -0.0819,  0.0255, -0.1006, -0.0192, -0.0628, -0.0826,  0.0321, -0.0188,
        -0.0514,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0305,  0.0266, -0.0241, -0.0199, -0.0247, -0.0044, -0.1702, -0.1082,
        -0.1051, -0.2854, -0.0841, -0.0215, -0.0504, -0.0215,  0.0157,  0.0078,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1132, -0.2153,  0.0008, -0.0880, -0.0014, -0.0595, -0.1227, -0.0070,
         0.0202,  0.0021,  0.0036,  0.0201, -0.0387, -0.0064, -0.0028, -0.0162,
        -0.0366, -0.1194, -0.0484, -0.0108, -0.0077,  0.0522, -0.0070,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0763, -0.0728,  0.0120, -0.2258, -0.3810, -0.0441,  0.0295, -0.0047,
        -0.0408,  0.0085, -0.0559, -0.0010, -0.0116,  0.0361,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #50: [tensor([ 0.0772, -0.0767, -0.0941, -0.1597, -0.0258, -0.0139, -0.0723,  0.0070,
        -0.0294, -0.0026, -0.0706, -0.0911, -0.0177,  0.0042, -0.0058,  0.0167,
         0.0136, -0.0848, -0.0029, -0.0785, -0.0140, -0.0037,  0.0065,  0.0311,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0060,  0.0306, -0.1747,  0.0096, -0.3143,  0.0078, -0.3437,  0.0140,
         0.0448, -0.0541,  0.0004,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0050,  0.0256, -0.0009, -0.0074,  0.0188, -0.1338, -0.1982, -0.0246,
        -0.1144, -0.0089, -0.0860,  0.0115, -0.0918, -0.0075, -0.0521, -0.0103,
        -0.0541, -0.0826, -0.0324,  0.0140,  0.0202,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0599,  0.0075,  0.0161,  0.0470,  0.0416,  0.0066, -0.0095, -0.3856,
        -0.0868, -0.2436, -0.0235, -0.0307,  0.0254, -0.0161,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0697, -0.0380, -0.0219, -0.0013,  0.0034, -0.0030, -0.0094, -0.0429,
         0.0025,  0.0028, -0.0585,  0.1829, -0.3746, -0.0412, -0.0139, -0.0043,
        -0.0507,  0.0216, -0.0079,  0.0115, -0.0379,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0697, -0.3862,  0.0043, -0.1601,  0.0080, -0.0688,  0.0134, -0.1223,
         0.0122, -0.0552, -0.0226, -0.0455,  0.0318,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0040, -0.1432, -0.0040, -0.0014, -0.0016, -0.0063, -0.0298, -0.0523,
        -0.0005, -0.0013,  0.0072,  0.0042, -0.0155,  0.0004, -0.0062, -0.0279,
        -0.0047, -0.0022,  0.0042, -0.0087, -0.0065,  0.0039,  0.0037,  0.0052,
        -0.1339,  0.0385, -0.0893, -0.0874, -0.1300, -0.0143,  0.0006, -0.0070,
         0.0062, -0.0072,  0.0075, -0.0027, -0.0483, -0.0032,  0.0037, -0.0074,
        -0.0282,  0.0010,  0.0030,  0.0006, -0.0351], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0597, -0.0322, -0.0838, -0.0053,  0.0137,  0.0193, -0.0064, -0.0331,
        -0.0065, -0.0203, -0.0925, -0.0147,  0.0118, -0.0022, -0.0588, -0.0154,
        -0.0738,  0.0097,  0.0055,  0.0017, -0.0107, -0.0893, -0.0014, -0.1078,
         0.0091,  0.0040, -0.0090, -0.0175, -0.0945, -0.0368,  0.0137, -0.0007,
        -0.0277, -0.0114,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0059,  0.0749,  0.0041, -0.0375, -0.0269, -0.0189,  0.0031,  0.0021,
         0.0365,  0.0200, -0.0288, -0.0079, -0.0086,  0.0004,  0.0165,  0.0139,
         0.0180,  0.0195, -0.0478, -0.2058, -0.3334,  0.0282,  0.0026, -0.0171,
         0.0042, -0.0173,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.1805e-03, -4.0026e-01,  8.6801e-03, -8.1911e-02,  1.7500e-03,
        -2.0325e-02,  1.9625e-03, -3.4412e-02,  3.4622e-04,  9.9993e-03,
         3.7201e-03,  2.8612e-03,  7.4486e-03,  1.2029e-02,  1.5463e-02,
        -8.2511e-02, -1.7617e-01,  1.3566e-02, -4.5818e-02, -5.5099e-03,
        -4.2328e-02,  1.9221e-02, -9.5252e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0163, -0.1656, -0.0102, -0.0782, -0.0048,  0.0009, -0.0058, -0.0379,
        -0.0690, -0.0109, -0.0066, -0.0089,  0.0028,  0.0031, -0.1600,  0.0036,
        -0.0066, -0.0447, -0.0476, -0.0088, -0.0037,  0.0021, -0.0006, -0.0031,
         0.0007,  0.0063, -0.0658, -0.0980, -0.0093, -0.0153,  0.0004,  0.0228,
        -0.0032,  0.0206, -0.0025,  0.0125, -0.0139,  0.0035,  0.0144, -0.0090,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0405,  0.0332,  0.0011, -0.0392, -0.0360,  0.0125, -0.0234, -0.0774,
         0.0034, -0.0107, -0.0005, -0.0002, -0.0286, -0.0052, -0.0327, -0.0903,
        -0.0652,  0.0069,  0.0004, -0.0382, -0.0235, -0.0795, -0.0478, -0.0094,
        -0.0304, -0.0101, -0.0108, -0.0493, -0.0152, -0.0672, -0.0667,  0.0087,
        -0.0026,  0.0333,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 2.6758e-02, -2.4259e-02, -6.0083e-02, -3.6797e-02, -5.1688e-02,
        -1.4347e-02,  1.0643e-02,  5.8978e-04,  1.3324e-03,  1.8693e-03,
         2.6383e-03, -3.8213e-02,  3.2606e-03, -6.6023e-03, -1.5991e-01,
        -1.1548e-02,  5.5281e-03, -4.7064e-03, -6.1780e-02, -6.0339e-05,
        -2.7253e-03,  2.8792e-03,  6.5710e-03,  8.4467e-05, -5.8099e-02,
        -1.2473e-01,  3.3514e-03, -4.1835e-03,  9.2322e-03,  6.0708e-03,
         1.6386e-03,  2.4752e-04,  5.1963e-03, -6.9729e-05, -3.7084e-02,
        -3.6682e-04, -2.3935e-02, -4.4141e-02, -8.9476e-03, -2.3992e-02,
        -2.5311e-02, -8.6032e-03, -2.2744e-02, -1.1209e-02,  2.8122e-03,
        -1.5322e-03,  1.1346e-03, -2.5309e-03, -1.5302e-02, -5.2693e-03,
        -7.9871e-05,  1.7319e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0041,  0.0038, -0.0057, -0.0033, -0.1266, -0.0163, -0.0229,  0.0467,
         0.0054,  0.0075,  0.0121,  0.0137, -0.0075, -0.1527, -0.0407, -0.0467,
        -0.1032, -0.0210, -0.0455, -0.0068, -0.0211,  0.0038, -0.0649, -0.1480,
        -0.0028, -0.0056, -0.0013,  0.0035, -0.0022, -0.0026,  0.0520,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0225, -0.1822,  0.0227, -0.0378,  0.0085,  0.0055, -0.0363,  0.0103,
        -0.0077,  0.0336,  0.0099, -0.0760,  0.0267, -0.0155, -0.1083, -0.0621,
         0.0061, -0.0787, -0.0785, -0.0921, -0.0295,  0.0066,  0.0042, -0.0046,
         0.0341,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0399, -0.0135,  0.0235,  0.0140,  0.0209, -0.2705, -0.4177, -0.0068,
        -0.0695,  0.0374, -0.0368, -0.0496,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.1179e-02, -3.4431e-01,  2.0214e-02, -7.0334e-02, -6.8288e-02,
        -3.2463e-02,  1.6464e-03, -3.0062e-03, -2.1127e-04, -1.0061e-02,
        -1.2424e-02, -1.3200e-02, -7.3173e-02, -1.1931e-01, -1.6245e-02,
        -4.7990e-02,  3.4147e-03,  2.0959e-02,  2.0522e-03, -1.0320e-02,
        -1.7569e-02, -3.4251e-04, -1.0310e-02,  3.7384e-03, -5.8924e-03,
        -1.2265e-02, -3.4014e-04, -5.3459e-02,  5.2838e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0860, -0.4312, -0.0071,  0.0268,  0.0197, -0.1423, -0.1701, -0.0069,
         0.0171,  0.0026,  0.0597, -0.0304,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0100, -0.3850, -0.0120, -0.1024, -0.0313, -0.1293, -0.1702,  0.0258,
        -0.0684,  0.0106, -0.0102,  0.0451,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0210,  0.0569, -0.1500, -0.0041, -0.1126,  0.0078, -0.0089,  0.0064,
        -0.1293,  0.0176,  0.0203, -0.2170, -0.0164,  0.0186, -0.0010, -0.1014,
        -0.0261,  0.0263, -0.0076,  0.0508,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0047, -0.2284, -0.0018, -0.0682, -0.0988, -0.0147, -0.0362, -0.0048,
        -0.0333, -0.0082, -0.0076, -0.0460, -0.0286, -0.0329, -0.0040, -0.0095,
        -0.0340, -0.0020, -0.0069, -0.0024, -0.0383,  0.0117,  0.0067,  0.0076,
         0.0110,  0.0031,  0.0082, -0.0498, -0.0313, -0.0092, -0.0020, -0.0028,
        -0.0238, -0.0574, -0.0197,  0.0384, -0.0060,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1359,  0.6464, -0.0503,  0.0562, -0.0201, -0.0011,  0.0096,  0.0200,
         0.0604,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0122, -0.0119,  0.0322, -0.1721,  0.0025,  0.0176, -0.0747, -0.0886,
         0.0155,  0.0077, -0.0061, -0.0219, -0.1980, -0.1368,  0.0031, -0.0068,
         0.0143,  0.0022, -0.0354, -0.0602,  0.0032, -0.0016, -0.0018,  0.0071,
        -0.0067,  0.0116, -0.0482,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0312, -0.2469,  0.0083, -0.0807, -0.0133, -0.0355, -0.0024,  0.0051,
        -0.0052, -0.0055,  0.0036, -0.0915, -0.0202,  0.0073,  0.0013, -0.0003,
        -0.0452,  0.0070, -0.0065, -0.0012, -0.0040, -0.0342, -0.0881, -0.0070,
        -0.0078,  0.0052,  0.0039, -0.0428, -0.1263, -0.0164,  0.0056,  0.0016,
         0.0117, -0.0274,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #150: [tensor([-0.0149, -0.0004,  0.0096,  0.0072, -0.0570, -0.0108,  0.0003,  0.0002,
         0.0114,  0.0108,  0.0043, -0.0543,  0.0128,  0.0039,  0.0172, -0.0552,
        -0.0973, -0.0118, -0.0233,  0.0035,  0.0156,  0.0029, -0.0388, -0.0011,
         0.0052,  0.0047,  0.0107, -0.0182,  0.0017,  0.0061,  0.0226, -0.0736,
        -0.1250, -0.0035, -0.0122, -0.0492, -0.0661, -0.0190, -0.0011, -0.0015,
        -0.0250, -0.0253, -0.0072,  0.0205, -0.0366,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0282, -0.0111, -0.0054, -0.0004, -0.0253, -0.2073, -0.2991, -0.0302,
         0.0040, -0.2104, -0.0330, -0.0268, -0.0650, -0.0084,  0.0096, -0.0357,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0121, -0.0045,  0.0165,  0.0141, -0.0101, -0.0072,  0.0050, -0.0139,
        -0.1595, -0.1510, -0.0327, -0.1452,  0.0124, -0.0651, -0.1255, -0.0008,
        -0.0065, -0.0048,  0.0076, -0.0155,  0.0063,  0.0554,  0.0068, -0.0062,
        -0.0008,  0.0336,  0.0808,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0038,  0.0091, -0.0500,  0.0033, -0.0509, -0.0081, -0.0224, -0.0009,
        -0.0115, -0.0027, -0.0031, -0.0144, -0.0494,  0.0034, -0.0383,  0.0057,
        -0.0228,  0.0055,  0.0014, -0.0332,  0.0098, -0.0275, -0.0175, -0.0638,
        -0.0022,  0.0035, -0.0234, -0.0556,  0.0133, -0.0502, -0.0043, -0.0217,
        -0.0509, -0.0108,  0.0006, -0.0363, -0.0271, -0.0813,  0.0012,  0.0048,
        -0.0296, -0.0563, -0.0108, -0.0014,  0.0257, -0.0303], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0706,  0.0092,  0.0170,  0.0111,  0.0014, -0.0025,  0.0028, -0.2208,
         0.0360, -0.0166,  0.0138,  0.0145, -0.0021, -0.0024,  0.0138,  0.0102,
         0.0093,  0.0143, -0.1266, -0.2235, -0.0203, -0.0093,  0.0052,  0.0126,
         0.0100,  0.0011, -0.0799,  0.0089,  0.0225, -0.0115,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0690, -0.4839,  0.0476,  0.0224, -0.0012, -0.0014, -0.0197,  0.2255,
         0.0293, -0.0173,  0.0827,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0687, -0.2860,  0.0089, -0.0456, -0.0669, -0.0320, -0.0823, -0.1578,
        -0.0096, -0.0135, -0.0184, -0.0065, -0.0057, -0.0340,  0.0014, -0.0626,
        -0.0397, -0.0212,  0.0030,  0.0363,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0712, -0.3788, -0.1764, -0.1238,  0.0041, -0.0678, -0.0788, -0.0333,
        -0.0159,  0.0441, -0.0058,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0039, -0.3287, -0.1317, -0.1193,  0.0122, -0.0233, -0.0202, -0.0441,
        -0.0928, -0.0161,  0.0039, -0.0235, -0.0028, -0.0475,  0.0053, -0.0422,
        -0.0091,  0.0086, -0.0214,  0.0040,  0.0396,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0149, -0.3663, -0.1749, -0.0052,  0.0235, -0.2386, -0.0053,  0.0041,
        -0.0206,  0.0126,  0.0004, -0.0602, -0.0064, -0.0074, -0.0194,  0.0167,
        -0.0021, -0.0215,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0146, -0.1347, -0.1424, -0.0097, -0.0223,  0.0005,  0.0050, -0.0005,
        -0.0047, -0.0148, -0.0015, -0.0029, -0.0078, -0.0749, -0.0609,  0.0066,
        -0.0540, -0.0454, -0.0439, -0.0292, -0.0038, -0.0455, -0.0791, -0.0042,
        -0.0418,  0.0010, -0.0307, -0.0146, -0.0086, -0.0455, -0.0064, -0.0028,
         0.0037,  0.0061, -0.0198, -0.0031,  0.0036,  0.0022, -0.0009,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-5.3595e-02, -4.7402e-01,  2.3842e-03, -6.7264e-02, -1.8167e-03,
        -4.5022e-03,  1.1704e-02,  1.0753e-02, -7.3566e-03, -1.1316e-02,
        -1.6603e-02, -7.2931e-02, -1.4938e-02,  3.5277e-03,  3.8456e-04,
         2.9086e-03, -6.0085e-02,  4.1869e-03, -4.5785e-03, -2.3478e-03,
        -1.7357e-03,  3.8262e-03,  1.6814e-03,  2.3698e-03, -3.3171e-03,
         1.2486e-03, -2.9940e-02,  9.2637e-04, -4.0484e-02, -5.4830e-02,
        -2.5815e-03, -4.7908e-03, -3.1559e-03,  2.1906e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.0640,  0.0455, -0.2252, -0.2339,  0.0952, -0.1451,  0.0516, -0.0759,
        -0.0637,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0095, -0.1457,  0.0329, -0.0722, -0.0318, -0.1003,  0.0008, -0.0292,
        -0.0348, -0.0139, -0.0432, -0.0174, -0.0053, -0.0043,  0.0008, -0.0724,
        -0.0195, -0.1182,  0.0109,  0.0065, -0.0022,  0.0103,  0.0058,  0.0020,
        -0.0181,  0.0073,  0.0066, -0.0032, -0.0085,  0.0002, -0.0023, -0.0463,
         0.0018, -0.0524,  0.0003, -0.0060, -0.0079,  0.0117,  0.0165, -0.0209,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0149, -0.1038, -0.1464, -0.0079, -0.0531, -0.0076,  0.0006, -0.0097,
        -0.0298, -0.0732, -0.0024, -0.0012,  0.0004,  0.0002, -0.0166, -0.0056,
         0.0090,  0.0054, -0.0346, -0.0044, -0.0486,  0.0021, -0.0071, -0.0176,
        -0.0030, -0.0313, -0.0325,  0.0020, -0.0718, -0.0883, -0.0024,  0.0006,
         0.0034,  0.0002,  0.0083,  0.0038, -0.0291, -0.0475, -0.0035,  0.0010,
         0.0029,  0.0113, -0.0042,  0.0059,  0.0067,  0.0017, -0.0031,  0.0013,
        -0.0020, -0.0041, -0.0033, -0.0224,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0149, -0.2855,  0.0035, -0.0967, -0.0179, -0.0178, -0.1089, -0.0126,
        -0.0605,  0.0054,  0.0102,  0.0006, -0.0909,  0.0065, -0.0447, -0.0597,
        -0.0404,  0.0025, -0.0733,  0.0032, -0.0034,  0.0410,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0184, -0.5322,  0.0345,  0.0232,  0.0178, -0.0057,  0.0060, -0.0697,
        -0.0028, -0.0068, -0.0121, -0.0505, -0.0236, -0.0143, -0.0283, -0.0752,
        -0.0189, -0.0133,  0.0117, -0.0155, -0.0194,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.9115e-02, -1.7282e-01, -2.6428e-02, -3.8365e-02, -1.0742e-02,
         1.1348e-02, -2.0602e-02,  4.1470e-03, -3.3958e-02, -7.0211e-03,
         4.2057e-03, -7.9510e-03,  7.9710e-04,  3.8196e-03,  1.2295e-02,
         3.7487e-03,  1.1204e-02,  1.9536e-03, -1.8085e-02,  2.1686e-03,
         5.1975e-03, -9.8902e-04,  3.6433e-03,  8.3438e-03,  9.8522e-04,
         6.0683e-03, -5.4248e-03, -4.1392e-02, -8.8189e-04, -3.0422e-02,
        -6.5455e-02, -5.6188e-02,  6.3582e-03, -3.0589e-02, -1.5122e-02,
         1.8715e-05, -5.5690e-03, -2.2266e-03,  1.4803e-03, -2.8680e-03,
        -5.9070e-02,  2.1134e-03,  2.2883e-02,  1.7083e-02, -3.5652e-02,
         6.3928e-03, -3.7824e-02, -2.0926e-02,  1.4748e-03,  1.5163e-02,
        -1.2513e-02, -6.2411e-03, -8.7116e-03,  1.5281e-02, -8.6695e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0038,  0.0056,  0.0086, -0.0300, -0.0103, -0.0150, -0.0683,  0.0059,
        -0.0005,  0.0042, -0.0054,  0.0171, -0.0027, -0.0299, -0.0701, -0.0027,
        -0.0037, -0.0298,  0.0051, -0.1020, -0.0164, -0.0306, -0.0485,  0.0012,
         0.0079,  0.0161,  0.0188, -0.0748,  0.0142,  0.0134,  0.0101, -0.0002,
        -0.0130, -0.0643, -0.0107, -0.0440, -0.0083, -0.0698, -0.0585, -0.0391,
        -0.0057, -0.0038,  0.0008, -0.0065, -0.0023,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0019, -0.1868, -0.0577, -0.0864,  0.0051, -0.0267, -0.0353,  0.0018,
        -0.0695, -0.0258,  0.0061, -0.0403,  0.0037, -0.0765, -0.0680, -0.0114,
        -0.0131, -0.0856, -0.0063, -0.0545, -0.0897,  0.0078, -0.0188,  0.0214,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0225,  0.0200, -0.0052,  0.0068, -0.0020, -0.0100, -0.0595, -0.0528,
         0.0268,  0.0071,  0.0103,  0.0090, -0.0253, -0.0128,  0.0046,  0.0124,
        -0.0369, -0.0495, -0.0627, -0.0962, -0.0121,  0.0084, -0.0597, -0.0844,
        -0.0040, -0.0478, -0.0601,  0.0034, -0.0326, -0.0783, -0.0294,  0.0206,
        -0.0264,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0435, -0.3226, -0.0226, -0.1008,  0.0181,  0.0011, -0.0034, -0.0255,
        -0.1557,  0.0072, -0.0061, -0.0067,  0.0277, -0.0972,  0.0032, -0.0687,
        -0.0279, -0.0371,  0.0250,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.3940e-02,  1.0822e-02,  2.3229e-02, -1.2705e-02, -8.6965e-02,
         2.1461e-02, -9.9438e-02, -1.1818e-01, -1.4722e-01, -6.3650e-03,
         7.9284e-05, -1.0213e-02, -6.8799e-03,  1.0772e-02,  2.7899e-03,
         9.6095e-04, -1.3033e-02, -5.8671e-02, -9.7949e-02,  2.2244e-02,
        -6.7980e-02, -5.2047e-02, -3.2613e-03, -2.0772e-02, -2.2029e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0517, -0.1757, -0.1445,  0.0437,  0.0031, -0.1580, -0.2203, -0.0289,
        -0.0209, -0.0388,  0.0058, -0.0079, -0.0480, -0.0011,  0.0074,  0.0036,
        -0.0406,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-3.1491e-02, -7.0695e-03, -1.0592e-02,  1.6408e-04, -1.0886e-02,
         1.5056e-02, -7.2797e-04, -2.3307e-02, -3.3101e-03, -6.6622e-02,
        -1.4307e-02,  3.2525e-03,  7.7690e-03,  4.0313e-03,  9.8789e-03,
        -1.9034e-03, -1.3943e-01, -1.8638e-01, -1.6472e-02, -8.2869e-02,
        -3.2035e-05, -7.7179e-02, -1.8390e-01, -9.7843e-03, -1.2023e-02,
        -9.9951e-03,  1.4519e-02,  1.5101e-02,  4.5287e-03, -1.3443e-02,
         2.0788e-02,  3.1872e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0305, -0.1511, -0.1322, -0.0002,  0.0158, -0.0095, -0.0587, -0.1224,
         0.0058,  0.0106,  0.0137, -0.0138, -0.0190,  0.0083,  0.0027, -0.0052,
        -0.0087, -0.0046, -0.0509, -0.0069, -0.0739,  0.0023, -0.0552, -0.1115,
        -0.0334, -0.0140, -0.0248, -0.0144,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.6172e-02, -6.1704e-02, -1.9415e-02, -1.4870e-02, -3.8225e-02,
        -1.2872e-01, -2.3113e-01, -5.4616e-02, -2.3777e-02, -4.9160e-03,
        -3.5021e-03, -1.1378e-01, -5.0747e-03, -1.9746e-03,  8.0768e-03,
        -5.9743e-02, -1.2991e-02, -3.2368e-02, -2.1515e-02,  9.9389e-04,
        -1.3658e-02,  1.4493e-03, -1.4836e-02, -8.6296e-03, -4.8124e-02,
        -3.2092e-05,  3.8992e-02,  1.0099e-02,  1.0617e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0049, -0.1166, -0.0032, -0.0730, -0.1195, -0.0025,  0.0015,  0.0038,
        -0.0020,  0.0071,  0.0089,  0.0052, -0.0448, -0.1502,  0.0384, -0.0384,
         0.0024, -0.0817, -0.0073, -0.0149,  0.0154,  0.0056, -0.0391, -0.0021,
        -0.0243, -0.0815, -0.0638,  0.0053, -0.0062,  0.0011,  0.0032, -0.0108,
         0.0153,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0100, -0.3491, -0.0093, -0.0639,  0.0073, -0.0411,  0.0144, -0.0271,
        -0.0038, -0.0741, -0.0887, -0.0171,  0.0012, -0.0037, -0.0005,  0.0102,
        -0.0307, -0.0056, -0.1105,  0.0162, -0.0704,  0.0083,  0.0193, -0.0175,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0600, -0.2186, -0.1424,  0.0026, -0.0629, -0.0048,  0.0004, -0.0480,
         0.0175, -0.0535, -0.0068, -0.0407, -0.0560, -0.0114, -0.0098, -0.0098,
         0.0032, -0.0346, -0.0100, -0.0735, -0.0365,  0.0618,  0.0351,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.3035e-02, -2.4659e-01, -8.5396e-03,  1.4012e-03, -3.2379e-02,
        -3.9333e-02,  4.3063e-03,  7.8611e-03, -6.0872e-03,  3.9053e-03,
        -9.8107e-03, -2.1331e-03, -1.5654e-02,  1.7060e-04, -8.6787e-03,
         1.5684e-02, -1.0108e-01, -4.4531e-02, -5.2845e-02,  2.3147e-03,
         1.3948e-02, -2.3989e-02,  1.0688e-03,  5.6699e-03, -4.2565e-02,
        -1.6345e-02, -2.2702e-03,  1.0344e-03, -1.4500e-02, -3.1406e-02,
        -6.4741e-02, -5.1730e-03, -2.7673e-02, -9.9115e-04, -3.0662e-02,
        -5.2879e-02, -2.0039e-03,  2.8131e-02,  1.8605e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0022, -0.0844, -0.0156, -0.0385, -0.1539, -0.0159,  0.0082, -0.0039,
        -0.0125,  0.0075, -0.0061, -0.0055,  0.0066, -0.0079,  0.0035, -0.0714,
        -0.0193, -0.0075,  0.0140, -0.0159, -0.0665,  0.0041,  0.0188, -0.0021,
        -0.0820, -0.0706,  0.0055, -0.0203, -0.0051, -0.0445, -0.0115, -0.0039,
         0.0141, -0.0039, -0.0081, -0.0077, -0.0158, -0.0684,  0.0068,  0.0054,
         0.0030,  0.0274,  0.0043,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0338, -0.2583, -0.0162,  0.0004,  0.0068,  0.0038, -0.0206,  0.0097,
        -0.0495,  0.0023,  0.0065, -0.0011, -0.0188,  0.0025, -0.0066, -0.0081,
        -0.0546, -0.0563,  0.0039, -0.0015,  0.0056,  0.0041, -0.0059,  0.0062,
         0.0069,  0.0030,  0.0073,  0.0064,  0.0114,  0.0021,  0.0116, -0.0400,
        -0.0801,  0.0017,  0.0004,  0.0012, -0.0042, -0.0658, -0.0384, -0.0037,
        -0.0262, -0.0428,  0.0003,  0.0071, -0.0032,  0.0005,  0.0025,  0.0042,
         0.0024, -0.0004,  0.0178,  0.0005,  0.0019,  0.0023,  0.0011, -0.0120,
         0.0073], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0556,  0.5816, -0.0230,  0.0095,  0.0335, -0.0106,  0.0976,  0.0714,
         0.0146,  0.0048,  0.0527,  0.0034, -0.0381, -0.0036,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0236, -0.1845, -0.0388,  0.0019, -0.0162, -0.0400,  0.0069,  0.0149,
        -0.0163,  0.0118, -0.0656, -0.1446, -0.0301,  0.0158, -0.0602, -0.0035,
        -0.0278, -0.0382,  0.0058, -0.0669, -0.1075, -0.0183, -0.0195, -0.0006,
         0.0124,  0.0029, -0.0008,  0.0055,  0.0106, -0.0088,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([0.0070, 0.3180, 0.0040, 0.2417, 0.2745, 0.0188, 0.0560, 0.0800, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
        0.0000, 0.0000, 0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 1.5708e-02, -2.1248e-01, -1.9439e-02, -2.8445e-02, -4.6361e-02,
        -9.8340e-02,  1.7288e-03,  8.7438e-03, -5.9739e-03, -1.8630e-02,
         1.5383e-02,  7.4179e-03, -5.9424e-02, -1.1770e-01,  6.8349e-03,
        -1.1426e-02, -8.2027e-03, -2.5124e-03, -1.6984e-03, -5.2498e-05,
         3.3972e-03,  2.4203e-03, -1.1680e-03, -3.4103e-03, -7.6320e-02,
        -2.1522e-02, -3.5519e-02, -1.1628e-03, -4.7001e-02, -8.8632e-02,
        -1.0584e-02,  1.0962e-02, -1.1406e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1370e-02, -2.0324e-01, -1.3214e-01, -3.2921e-02, -7.6452e-03,
        -2.8086e-02, -2.7118e-02, -1.1518e-02,  9.7006e-03, -4.0961e-02,
        -4.7015e-03, -8.0264e-03, -1.1506e-02, -4.9640e-03,  1.7327e-02,
         1.1545e-03, -5.4128e-04,  1.1492e-02, -2.3066e-02,  2.9146e-03,
        -4.1223e-04, -4.4365e-03, -5.1684e-03,  1.8588e-03,  4.4843e-03,
        -1.8750e-03, -3.1771e-03, -2.4479e-02,  3.2168e-03, -2.9185e-04,
        -1.1929e-03, -9.6821e-05, -2.0169e-02, -2.9810e-02,  8.6649e-03,
        -9.6727e-03, -1.5856e-02,  4.7574e-03, -7.7437e-03,  2.4525e-03,
        -1.9638e-03, -4.9392e-03, -7.3878e-03, -4.5886e-02, -1.2227e-02,
        -5.5423e-03, -3.2856e-03, -2.5249e-03, -4.6681e-03,  5.5962e-04,
         2.4365e-03, -1.4880e-02, -4.4198e-03, -9.7452e-03,  4.3957e-05,
         3.2700e-03, -4.3923e-04, -1.7149e-03,  3.7300e-04, -3.4280e-02,
        -2.9720e-03, -1.0585e-02, -2.2695e-02,  6.0665e-04,  4.3092e-04,
        -3.5833e-03, -1.0910e-02,  2.6750e-02,  8.7921e-03, -7.8325e-03,
         4.0442e-03], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0209,  0.0115, -0.0204, -0.0019, -0.0745, -0.0194, -0.1150, -0.0265,
        -0.0060, -0.0012, -0.0089, -0.0064, -0.0015,  0.0127,  0.0033,  0.0120,
         0.0007,  0.0104, -0.0128, -0.0877, -0.0155,  0.0151,  0.0235, -0.0489,
        -0.0023, -0.0717, -0.0082,  0.0124, -0.0076, -0.0058,  0.0183,  0.0010,
        -0.0708, -0.0778, -0.0193,  0.0048, -0.0707, -0.0419,  0.0042,  0.0020,
        -0.0069, -0.0177,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.2403e-02, -3.0961e-01,  1.1863e-02, -2.6320e-03, -3.5556e-03,
         7.8326e-03,  4.3381e-03,  3.7226e-03, -1.1617e-02, -4.6294e-02,
         2.4231e-04, -3.6922e-02, -8.1577e-02, -1.4151e-02, -1.0233e-03,
        -3.0594e-03, -9.1810e-03, -7.1490e-02, -6.2395e-02, -1.4487e-02,
        -5.7076e-03, -2.6497e-02, -1.5465e-03,  8.1272e-06, -2.0552e-02,
        -6.7424e-02, -6.6862e-03, -2.9057e-03, -7.4952e-02,  1.3516e-02,
         1.9165e-02, -1.7166e-02,  3.3630e-03, -9.0673e-03, -5.9494e-03,
        -5.2678e-03,  1.8270e-03,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0376, -0.4846, -0.0487, -0.0754,  0.0106,  0.0073, -0.0252, -0.0010,
        -0.0086, -0.0136, -0.0662,  0.0237, -0.0923, -0.0506,  0.0545,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0568, -0.0536,  0.0062, -0.0277, -0.1823, -0.1710, -0.0177, -0.0832,
        -0.1806, -0.0005, -0.0068, -0.0097, -0.0437, -0.0287,  0.0388,  0.0010,
        -0.0032, -0.0467,  0.0115,  0.0304,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0303,  0.0053,  0.0077, -0.0024, -0.1349, -0.0098,  0.0062, -0.0178,
         0.0017,  0.0042, -0.0396, -0.0572, -0.0046, -0.0029, -0.0517, -0.0404,
         0.0072, -0.0418,  0.0183, -0.0216, -0.0024, -0.0009, -0.0052,  0.0016,
         0.0024, -0.0372, -0.0598,  0.0067,  0.0087, -0.0405, -0.0615,  0.0080,
         0.0008, -0.0118, -0.0540, -0.0867,  0.0079, -0.0526,  0.0043,  0.0066,
         0.0013,  0.0082,  0.0254,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0174, -0.0059, -0.0302, -0.0089, -0.0864, -0.0053, -0.0313, -0.1016,
        -0.0355, -0.0128,  0.0058,  0.0039,  0.0104, -0.0743, -0.0739,  0.0015,
         0.0003, -0.0035,  0.0021, -0.0193, -0.0843, -0.0229,  0.0132, -0.0101,
        -0.0265, -0.0117, -0.0083, -0.0243, -0.0758,  0.0076,  0.0047, -0.0510,
        -0.0097, -0.0245, -0.0454, -0.0017, -0.0128, -0.0353,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0068, -0.2778, -0.1245, -0.0858,  0.0053, -0.0845, -0.0028,  0.0415,
        -0.0447, -0.1128,  0.0059, -0.0149, -0.0487, -0.0127,  0.0098, -0.0056,
         0.0013, -0.0687, -0.0068,  0.0097,  0.0044, -0.0249,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.4898e-02, -1.1444e-01,  1.8538e-02,  2.6473e-05,  1.2268e-03,
        -3.8324e-03,  1.2030e-02,  2.2668e-02,  4.3454e-04, -8.8539e-02,
         1.2740e-04, -7.6835e-03, -4.7282e-03,  4.3437e-03, -5.7409e-02,
        -6.8567e-02, -7.9638e-02, -7.0700e-03, -1.1625e-02, -6.6189e-02,
         8.8746e-03, -1.1866e-02, -3.5295e-02,  1.2978e-02, -6.9713e-05,
        -5.0899e-03, -1.6397e-02, -1.0922e-02, -5.8044e-03, -1.5264e-02,
         1.1758e-02,  4.1050e-02,  1.8672e-02, -2.1043e-02, -4.6051e-02,
        -4.1215e-02, -3.0021e-03, -3.5840e-02,  1.2042e-02, -3.2756e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0312, -0.0257, -0.0112,  0.0078, -0.0185, -0.0529,  0.0043, -0.0184,
        -0.0455, -0.0173, -0.0116, -0.0176, -0.0059, -0.1601, -0.0954, -0.0389,
         0.0186, -0.0041, -0.1114, -0.0150,  0.0020,  0.0010, -0.0051,  0.0061,
        -0.0011, -0.0751, -0.0005, -0.0781, -0.0977, -0.0117,  0.0025,  0.0077,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0111, -0.2314, -0.0060, -0.0018, -0.0104, -0.0065, -0.0271, -0.0589,
        -0.1283,  0.0052, -0.0080, -0.0638,  0.0074, -0.0117,  0.0137,  0.0028,
        -0.0962, -0.0165,  0.0004,  0.0008, -0.0005, -0.0085,  0.0039, -0.0720,
        -0.0073,  0.0103, -0.0318,  0.0018, -0.0087, -0.0135, -0.0118,  0.0075,
         0.0090, -0.0473,  0.0583,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #350: [tensor([ 4.7785e-02, -2.8018e-02, -1.7292e-01, -1.1919e-01, -1.3773e-02,
        -4.8988e-03, -1.4591e-01, -1.0667e-01, -3.6155e-02, -1.3602e-01,
        -1.6151e-02, -4.6651e-02,  1.3277e-02, -2.8657e-02,  9.9034e-07,
        -6.6447e-03, -1.5458e-02, -4.7887e-03, -3.3157e-03,  2.0424e-02,
         3.3289e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.3905e-02, -4.0447e-01,  4.3282e-02, -6.2542e-02, -1.2712e-01,
        -4.9056e-02,  2.5330e-03, -4.9705e-02, -5.2042e-02, -6.3891e-03,
        -2.8408e-03,  5.0696e-03,  6.0347e-03,  5.6312e-03, -4.3267e-02,
        -2.9739e-04, -2.9985e-02, -3.3363e-02,  3.2470e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0347,  0.0023, -0.0883, -0.0011, -0.0711, -0.0130,  0.0260,  0.0394,
        -0.0063, -0.0948, -0.1975,  0.0233, -0.0786, -0.1524,  0.0027, -0.0645,
         0.0070, -0.0576,  0.0067, -0.0325,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0101, -0.3930, -0.0184, -0.1214, -0.0012,  0.0014, -0.0093, -0.0353,
        -0.0040, -0.0043,  0.0008, -0.0213, -0.0035, -0.0540, -0.0311, -0.0804,
        -0.0700, -0.0118, -0.0196, -0.0350, -0.0038,  0.0089,  0.0616,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.8306e-02, -2.6888e-01, -6.5604e-02, -3.8533e-02, -1.1372e-03,
        -4.9695e-02, -1.1024e-02, -7.5293e-02, -9.0339e-02, -8.2419e-02,
        -2.9783e-02, -4.7910e-02, -2.3515e-03, -4.2182e-02,  5.1497e-06,
        -6.2126e-03,  1.5911e-03,  1.2384e-03, -2.4612e-02, -2.7895e-02,
        -5.0363e-02, -1.5622e-02, -9.5102e-03, -9.7302e-03, -9.7682e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.2323e-02, -3.2214e-01, -1.6071e-02, -9.4447e-03,  4.0256e-03,
        -1.6656e-02,  1.1980e-02, -1.0498e-02,  1.2358e-02, -4.2872e-03,
        -1.0767e-01,  9.1891e-03,  7.4074e-03, -3.3230e-03,  6.2752e-03,
        -6.9962e-03,  8.8966e-03,  4.9708e-03,  9.3454e-03, -1.9105e-03,
        -6.7664e-02, -6.0469e-02, -3.8339e-02, -1.4778e-02, -5.2931e-02,
        -1.9377e-02,  2.7929e-03, -9.9838e-05, -1.0272e-02, -2.9613e-02,
        -3.0669e-02,  1.6768e-02, -2.9756e-02, -4.1769e-03, -1.6523e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0242, -0.0155,  0.0135, -0.1127, -0.0302,  0.0187, -0.0859, -0.0248,
        -0.1603,  0.0054,  0.0334,  0.0051, -0.0045, -0.0604, -0.0632, -0.1591,
        -0.0155, -0.1394,  0.0049,  0.0026,  0.0003, -0.0205,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1041, -0.0581,  0.0460, -0.0184,  0.0073, -0.0493, -0.0104, -0.0315,
        -0.0479, -0.3129,  0.0090, -0.0474,  0.0159,  0.0123,  0.0282, -0.0924,
         0.0016,  0.0372,  0.0491,  0.0063,  0.0148,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0167, -0.4164, -0.0407, -0.0162, -0.0065,  0.0082, -0.1509, -0.1163,
        -0.0196, -0.0105,  0.0270,  0.0177, -0.0070, -0.0547, -0.0086, -0.0456,
        -0.0037,  0.0040,  0.0297,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4407e-02,  4.7229e-03, -1.5838e-02,  1.8403e-02,  5.5583e-03,
        -5.6567e-02,  4.5483e-03,  3.6357e-03,  2.1565e-03, -3.4504e-03,
        -9.1456e-03, -2.0037e-02, -1.4042e-02, -4.0042e-02,  2.7267e-03,
        -1.3172e-02, -7.3851e-02, -7.6299e-02, -1.0142e-02, -5.8091e-02,
        -2.0908e-02,  4.3163e-03, -7.5066e-02, -4.1626e-03, -4.6766e-02,
        -1.0832e-02, -5.3135e-03, -3.1336e-02, -1.7805e-02, -1.1097e-01,
         2.2703e-03, -7.2347e-03, -3.4919e-02, -5.4837e-02,  4.2796e-03,
        -5.1582e-03, -1.7929e-02,  8.4295e-03, -1.4658e-03, -3.7005e-02,
        -3.4736e-03,  2.4238e-03,  4.3344e-03, -2.0220e-02, -9.8469e-03,
         1.1762e-02, -1.0707e-04], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0244, -0.2360, -0.1666, -0.0020,  0.0112, -0.0110,  0.0026, -0.0091,
        -0.0683, -0.0038,  0.0107, -0.0008,  0.0093,  0.0079,  0.0086, -0.0133,
        -0.0866, -0.0106,  0.0064, -0.0061, -0.0023, -0.0106, -0.1253, -0.0924,
         0.0066, -0.0050, -0.0200, -0.0017,  0.0033,  0.0035,  0.0012,  0.0240,
         0.0068, -0.0018,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0209, -0.4441,  0.0067,  0.0241,  0.0158, -0.0358,  0.0032,  0.0129,
        -0.0007, -0.0203,  0.0045, -0.0045, -0.0393, -0.0745, -0.0059,  0.0021,
        -0.0183,  0.0108,  0.0202, -0.0617, -0.0067,  0.0181, -0.0131, -0.0155,
        -0.0036, -0.0032,  0.0007, -0.0059, -0.0121, -0.0405, -0.0163,  0.0048,
        -0.0237, -0.0100,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 1.4781e-02,  4.2736e-01, -2.0350e-02,  2.4786e-03,  3.5830e-02,
         1.0191e-01, -1.8759e-02,  1.3390e-02, -3.4501e-04,  1.2279e-01,
         9.1926e-03, -9.1849e-03,  5.9271e-02,  1.0803e-01,  1.0083e-02,
        -7.7633e-03, -1.9747e-02, -1.8738e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0120, -0.0236, -0.0091, -0.0429,  0.0041, -0.0135, -0.0050, -0.0043,
        -0.1742, -0.2712, -0.0148,  0.0272, -0.0066, -0.0230,  0.0127, -0.0084,
        -0.0377,  0.0053, -0.0031, -0.2168, -0.0032,  0.0260, -0.0032,  0.0122,
         0.0016,  0.0038, -0.0346,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0538,  0.0135,  0.0172, -0.0042,  0.0048, -0.0109,  0.0220,  0.0118,
        -0.1523, -0.0122,  0.0062, -0.0398, -0.2312, -0.0175, -0.0047,  0.0090,
        -0.0172, -0.0078,  0.0343, -0.2104,  0.0057,  0.0261, -0.0432, -0.0134,
        -0.0037, -0.0271,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0352, -0.2259, -0.0024, -0.1255, -0.0700, -0.0212, -0.0252, -0.0117,
        -0.0893, -0.0186, -0.0683, -0.0501, -0.0105, -0.0075,  0.0028, -0.0756,
        -0.0473, -0.0463, -0.0102, -0.0111, -0.0453,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0542, -0.2334, -0.2307, -0.1956,  0.0005, -0.0016, -0.0970, -0.0020,
        -0.0448, -0.0052, -0.0111, -0.0674,  0.0056,  0.0186, -0.0323,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.2533e-04, -2.1605e-02, -3.8977e-03, -1.3889e-02, -5.4245e-02,
        -1.1673e-02,  5.0342e-02, -4.7193e-02, -9.9298e-03, -1.8750e-02,
         5.0601e-02,  4.0643e-02, -4.5650e-01, -6.3088e-02,  2.7008e-02,
        -4.1973e-02,  4.6827e-02, -4.1712e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0306, -0.2820, -0.1103, -0.1151,  0.0058,  0.0116,  0.0351, -0.0498,
         0.0039, -0.0580, -0.0935, -0.0404,  0.0414, -0.0742, -0.0038, -0.0010,
         0.0100,  0.0237,  0.0098,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0175, -0.2148, -0.1278, -0.0030, -0.0092, -0.0140,  0.0142,  0.0024,
        -0.0183, -0.0045,  0.0009,  0.0033, -0.0224,  0.0007,  0.0052, -0.0209,
        -0.0585, -0.0025, -0.0006,  0.0009, -0.0736,  0.0034, -0.0358,  0.0030,
        -0.0330,  0.0074, -0.0553,  0.0140, -0.0036, -0.0182, -0.0082, -0.0059,
        -0.0348, -0.0048, -0.0158, -0.0172, -0.0503,  0.0048, -0.0265, -0.0100,
        -0.0081, -0.0104,  0.0040,  0.0105], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.3166e-03, -2.1784e-01, -1.0391e-02, -1.1784e-01, -9.4641e-02,
        -3.1425e-02, -4.4989e-02, -4.0047e-03, -7.3089e-02, -9.2046e-03,
        -2.7966e-02,  1.4428e-03,  1.2584e-02,  2.0515e-03,  1.1469e-02,
         2.6641e-03,  3.7545e-02, -8.3498e-02,  7.5946e-03, -2.0004e-02,
        -3.1829e-02, -1.2964e-03,  1.5795e-03,  2.3433e-03,  7.9354e-03,
        -6.5073e-04,  9.3321e-05, -4.2331e-02, -4.9694e-03,  5.1475e-03,
         7.8890e-05, -2.1531e-02,  4.4358e-03,  2.7278e-02,  3.3940e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0296, -0.4505, -0.0022, -0.0561,  0.0032,  0.0053, -0.0536, -0.0649,
         0.0067,  0.0068, -0.0068, -0.0553, -0.0827, -0.0021,  0.0080, -0.0545,
         0.0070, -0.0015,  0.0027, -0.0054,  0.0070, -0.0075,  0.0077,  0.0729,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.6481e-02, -3.0162e-01, -5.3608e-03, -8.2937e-02, -7.3985e-03,
        -2.2436e-02,  1.1122e-02, -3.1403e-02, -8.2610e-02, -1.2595e-02,
        -3.8873e-02,  7.9553e-03, -3.2570e-03, -3.1481e-02, -4.5607e-02,
         3.1086e-03, -2.7115e-03, -5.3366e-03, -4.4412e-02, -5.8271e-03,
        -8.4407e-03, -3.5845e-02, -1.4062e-04, -2.3843e-02,  1.1004e-02,
        -3.9441e-02, -2.9041e-02, -2.9325e-02,  2.0033e-03,  3.1211e-03,
        -2.4750e-02,  3.6041e-03, -1.7001e-02, -9.9112e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 4.9738e-02,  2.7140e-02,  4.6584e-03,  9.8872e-03,  9.7903e-03,
         3.3044e-03, -9.0142e-02, -1.1618e-02,  2.7942e-02,  1.3836e-02,
         3.7699e-03, -4.0929e-03, -5.4601e-02, -1.0178e-02, -1.1570e-01,
        -1.8909e-01, -1.3374e-02,  6.3274e-03,  6.0419e-03,  2.0829e-02,
         7.6933e-05,  8.1747e-03,  2.0379e-02, -7.7093e-02, -9.4986e-02,
         2.0037e-02, -5.9523e-02,  3.3658e-02,  4.0576e-04, -1.3605e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #450: [tensor([-0.0572, -0.0232,  0.0035, -0.3117,  0.0182, -0.0078, -0.0262, -0.0147,
        -0.0721, -0.4015, -0.0027,  0.0087, -0.0526,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1122,  0.0177, -0.0485, -0.0012,  0.0170, -0.0477, -0.0213, -0.1740,
        -0.2108,  0.0023, -0.0288, -0.0851,  0.0024, -0.0476, -0.1046, -0.0036,
         0.0408, -0.0344,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1064,  0.2122,  0.3617,  0.0100,  0.1920, -0.0268,  0.0874,  0.0035,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0107, -0.2598, -0.0182, -0.0385, -0.0429,  0.0004, -0.0134,  0.0007,
         0.0033, -0.0104,  0.0082,  0.0117, -0.0125,  0.0042,  0.0043, -0.0169,
        -0.0027, -0.0022, -0.0004, -0.0038,  0.0054,  0.0057, -0.0039, -0.0221,
         0.0051, -0.0289, -0.0056,  0.0048, -0.0546, -0.0305,  0.0090, -0.0326,
        -0.0258, -0.0881, -0.0228, -0.0401, -0.0350, -0.0015,  0.0045,  0.0024,
        -0.0210, -0.0074, -0.0119, -0.0056,  0.0079,  0.0221,  0.0307],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0410, -0.1217,  0.2374, -0.4284, -0.0194,  0.0192, -0.0036, -0.0206,
        -0.0406,  0.0525,  0.0156,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0324, -0.2256, -0.0527, -0.1771, -0.0646,  0.0136, -0.0244, -0.1452,
         0.0011, -0.0712, -0.0060,  0.0042, -0.0683,  0.0162, -0.0133, -0.0287,
        -0.0229,  0.0325,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0821, -0.1108, -0.3190,  0.0029, -0.0247, -0.0316, -0.1427,  0.0371,
        -0.1085, -0.0280,  0.0082,  0.0222, -0.0154, -0.0508,  0.0160,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0853, -0.4266, -0.0013,  0.0058, -0.0041, -0.1004, -0.1102, -0.0256,
        -0.0062, -0.1273, -0.0800, -0.0039, -0.0096, -0.0007,  0.0131,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0534, -0.0008,  0.0081,  0.0184,  0.0019,  0.0160,  0.0076,  0.0006,
        -0.0757, -0.0021,  0.0024, -0.0263, -0.0906, -0.0016, -0.0010, -0.0184,
        -0.0903,  0.1922, -0.0757,  0.0301, -0.1006, -0.1213,  0.0109,  0.0033,
         0.0122, -0.0388,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0351, -0.0456,  0.0593,  0.0024, -0.0140,  0.0081,  0.0088, -0.2545,
        -0.3400, -0.0287,  0.0603, -0.0118, -0.0076,  0.0558, -0.0679,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0121, -0.2862,  0.0070, -0.1035, -0.1010, -0.0197, -0.0835, -0.0178,
        -0.0035, -0.1261,  0.0218,  0.0238, -0.0244, -0.0060,  0.0061, -0.0051,
        -0.0159,  0.0170, -0.0421,  0.0094, -0.0221,  0.0118,  0.0184,  0.0069,
         0.0073, -0.0015,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0042, -0.0066,  0.0027,  0.0093,  0.0003, -0.0265,  0.0048, -0.0075,
        -0.0633, -0.0820, -0.1697, -0.0222, -0.0124,  0.0078, -0.2228, -0.0658,
         0.0163, -0.0334, -0.0016, -0.1002, -0.0290, -0.0408, -0.0077, -0.0422,
         0.0034,  0.0176,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 0.0137,  0.3122,  0.0354, -0.0110,  0.0260,  0.0126,  0.0578,  0.0571,
         0.0031,  0.0355,  0.0100,  0.0416,  0.0757,  0.0038,  0.0241,  0.0349,
         0.0035,  0.0044,  0.0346,  0.0011, -0.0048,  0.0188, -0.0061,  0.0245,
         0.0511, -0.0141,  0.0260,  0.0056, -0.0037, -0.0007, -0.0010,  0.0190,
         0.0045, -0.0046, -0.0103, -0.0075,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0130, -0.1322, -0.0849,  0.0019, -0.0721,  0.0004, -0.0479, -0.0467,
        -0.0033,  0.0161, -0.0431, -0.0108, -0.0021,  0.0164,  0.0009, -0.0188,
        -0.0046, -0.0042, -0.0011, -0.0026, -0.0024, -0.0818, -0.0198, -0.0051,
        -0.0501, -0.0678,  0.0081, -0.0597, -0.0395, -0.0058, -0.0064, -0.0485,
        -0.0502, -0.0053, -0.0177,  0.0088,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0308,  0.0378, -0.0220,  0.0063, -0.0249, -0.1855, -0.0624, -0.0094,
         0.0310,  0.0170, -0.0110, -0.1424, -0.2771,  0.0376,  0.0319, -0.0036,
         0.0290,  0.0169, -0.0216,  0.0019,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.1578e-02, -1.9181e-01, -6.0374e-02, -5.7343e-02,  9.6724e-03,
         3.0303e-02, -9.3542e-04, -3.0941e-02, -1.2750e-02, -3.9965e-02,
         1.6993e-03, -2.0329e-03, -1.5625e-04,  1.3569e-02, -5.9689e-02,
        -5.0006e-02, -5.4154e-03, -2.9988e-02, -4.0638e-03, -7.1863e-03,
        -1.8108e-03, -6.6937e-03,  1.6228e-03,  1.2643e-02,  3.9645e-03,
         6.9252e-03,  4.7764e-03,  4.0556e-03,  7.3925e-03, -3.0118e-02,
        -9.4304e-02, -2.2253e-02, -1.5893e-02, -5.8092e-02,  1.2391e-02,
         3.7689e-02, -2.0531e-02, -1.9370e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.1888e-02, -9.9985e-02, -1.0943e-01, -8.3085e-02, -1.0385e-02,
         1.0407e-03, -1.2566e-02,  1.3247e-04,  1.4850e-02, -2.0720e-03,
        -4.9665e-02, -6.2387e-02, -5.0021e-03,  9.2289e-03, -1.9953e-02,
        -5.7596e-02, -1.0714e-02, -5.2711e-02, -1.8971e-03,  1.8637e-03,
         2.1466e-03, -4.0896e-05,  1.0835e-02,  6.3755e-03,  2.9045e-02,
        -7.6261e-03, -3.9319e-02, -5.9308e-02, -5.1704e-02, -1.4611e-03,
        -7.7651e-03, -5.1018e-03,  4.2502e-04,  2.1033e-03, -2.6661e-02,
         8.9539e-04, -1.2477e-02, -1.0733e-02, -3.0166e-03,  2.3040e-02,
         4.3466e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0589, -0.3969, -0.0008,  0.0292,  0.0010, -0.0866,  0.0030, -0.0017,
        -0.0306, -0.0017,  0.0079, -0.0927, -0.1265, -0.0162, -0.0042, -0.0418,
        -0.0031, -0.0683,  0.0031,  0.0132, -0.0093, -0.0032,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0040, -0.0002, -0.0013,  0.0009, -0.0139, -0.0423, -0.0072,  0.0033,
        -0.0598, -0.0138,  0.0019, -0.1123,  0.0055, -0.0304,  0.0065, -0.1204,
        -0.0043, -0.0605, -0.0156, -0.0559, -0.1041,  0.0196,  0.0117, -0.0389,
         0.0035, -0.0035,  0.0045, -0.0425, -0.0754,  0.0131, -0.0337, -0.0605,
         0.0015, -0.0132, -0.0140,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0059, -0.0492, -0.1788, -0.0171, -0.1732, -0.0279,  0.0025,  0.0097,
        -0.0097,  0.0014, -0.0444, -0.0081,  0.0076,  0.0049, -0.0096, -0.0041,
         0.0353, -0.0275, -0.0868, -0.0012, -0.0892,  0.0061,  0.0134, -0.0623,
        -0.0276, -0.0104, -0.0691, -0.0025, -0.0143,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0244, -0.3338, -0.0896, -0.0569, -0.0223, -0.0883,  0.0108,  0.0080,
        -0.0039, -0.0120, -0.0353, -0.0189,  0.0063, -0.0072, -0.0135, -0.0496,
        -0.0215, -0.0106, -0.0252, -0.0094, -0.0227, -0.0374, -0.0238, -0.0304,
         0.0385,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0557,  0.0173, -0.0005,  0.0168, -0.1479,  0.0038, -0.0104, -0.1313,
        -0.1889, -0.0620, -0.0209,  0.0167, -0.0040, -0.0284, -0.0724, -0.1087,
         0.0325, -0.0168,  0.0074,  0.0161, -0.0134,  0.0149, -0.0132,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0709,  0.0032,  0.0170,  0.0130,  0.0072,  0.0124,  0.0086, -0.0004,
        -0.0007, -0.0037,  0.0042,  0.0024, -0.1076, -0.1211, -0.0087, -0.0254,
        -0.0184, -0.0628, -0.0023,  0.0120,  0.0140,  0.0025,  0.0095,  0.0066,
        -0.0251,  0.0040, -0.1012,  0.0046, -0.0961, -0.0258, -0.0883, -0.0027,
         0.0072, -0.0193, -0.0485,  0.0116,  0.0059, -0.0047,  0.0203,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0027, -0.0101,  0.0429, -0.1466, -0.3035, -0.0109, -0.0459, -0.2023,
        -0.1736,  0.0268, -0.0157, -0.0191,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #550: [tensor([-2.1222e-02, -1.7520e-01, -2.1858e-01,  3.9552e-03, -4.4249e-03,
         1.5023e-02, -7.1208e-02,  8.2054e-03, -3.0231e-03, -3.4041e-02,
         8.3491e-04,  8.2112e-03, -2.1103e-02, -1.0886e-02, -6.8622e-02,
         1.4468e-02, -4.0932e-03, -5.3067e-02, -7.5094e-03, -3.3140e-03,
        -2.4280e-02, -3.2374e-02,  4.8553e-03, -2.9010e-04, -1.8981e-02,
         4.8241e-03, -4.2079e-03, -1.2958e-02, -1.8477e-02, -1.3519e-02,
         2.1630e-03,  2.4856e-02, -1.6185e-02,  1.6883e-03, -2.6282e-03,
        -1.3289e-02, -2.8517e-02,  5.0709e-04,  6.0344e-03, -1.4474e-04,
         4.7086e-03,  3.7414e-03, -9.4188e-04, -9.1186e-03,  3.7224e-03,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0254, -0.5086, -0.0636, -0.0810, -0.0474, -0.0189,  0.0228, -0.0140,
        -0.1394,  0.0165, -0.0136, -0.0488,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0527, -0.6562,  0.1007, -0.1044, -0.0122, -0.0056, -0.0169,  0.0514,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.7983e-03, -2.3830e-01, -8.4164e-03, -1.0719e-02, -6.5903e-02,
        -1.3295e-02,  1.1290e-02,  6.5762e-03, -3.0660e-02, -5.5038e-02,
        -1.0917e-02, -2.2111e-03, -4.1090e-03,  1.4859e-03,  4.8856e-04,
         1.9967e-03, -9.7254e-04, -4.1447e-03,  2.1004e-03,  1.6231e-03,
        -1.8349e-04,  4.1517e-03, -1.8092e-03, -2.9248e-03, -2.2986e-03,
         2.3991e-03, -1.2950e-03, -2.7890e-03, -2.0112e-02,  2.5929e-03,
        -1.9317e-03, -6.9944e-03,  1.5030e-03,  6.9385e-04, -2.0753e-03,
        -9.3343e-04,  6.0304e-03,  1.0387e-03, -1.7594e-03,  5.0196e-04,
         4.3864e-03,  1.0294e-03, -1.9244e-03, -1.9316e-03,  3.6343e-03,
         1.6362e-02, -5.4190e-02,  5.2268e-04, -5.2210e-02, -1.2778e-01,
        -4.1246e-02, -6.0374e-04, -4.8100e-02, -1.6035e-02, -1.7306e-02,
         1.4529e-04, -2.7110e-02, -3.9610e-03, -1.9036e-03,  7.3423e-04,
         3.5826e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0646, -0.0290,  0.0066, -0.0251, -0.0765,  0.0051, -0.0555,  0.0049,
         0.0068, -0.0064, -0.0106, -0.0738, -0.0003, -0.0733, -0.1073, -0.0113,
        -0.0004, -0.0145, -0.0400, -0.0128, -0.0755, -0.0689, -0.0168,  0.0034,
        -0.0376, -0.0464, -0.0092, -0.0264, -0.0317, -0.0107,  0.0045,  0.0156,
        -0.0284,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.1942e-02, -3.7537e-02, -1.3955e-01, -8.2565e-02,  3.4932e-03,
         1.6252e-03, -7.2050e-02,  2.7448e-03,  6.3488e-03, -4.9234e-03,
         1.4376e-02,  3.8929e-03, -7.3876e-02,  3.9061e-03, -4.3459e-03,
         3.2399e-02, -1.8682e-03, -1.0705e-01, -2.5500e-01,  1.2535e-04,
         1.8878e-02,  1.1820e-02,  1.6279e-02, -4.3405e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.2921e-02, -1.3755e-01, -5.0739e-02, -6.2686e-02,  1.5634e-02,
        -6.3582e-02,  5.5603e-03, -4.8410e-02, -3.4027e-05, -1.0172e-01,
        -1.2029e-01, -1.0399e-01, -1.2309e-01, -1.9967e-02,  1.3576e-02,
        -1.5672e-03,  4.2395e-03,  7.3265e-03,  6.6796e-03, -1.4725e-02,
        -2.9354e-02, -2.2468e-02,  3.2056e-02,  1.8310e-03,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0325, -0.0117, -0.0462, -0.1133,  0.0007,  0.0003,  0.0027, -0.0528,
         0.0025,  0.0004,  0.0010,  0.0037, -0.0101,  0.0031, -0.0621, -0.0577,
        -0.0780, -0.0137, -0.0090, -0.0060, -0.1008, -0.0011, -0.0855, -0.0900,
        -0.0129,  0.0071, -0.0509, -0.0074, -0.0063, -0.0504, -0.0250, -0.0200,
        -0.0024, -0.0326,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0438, -0.0324, -0.0084,  0.0054, -0.0883,  0.0038, -0.0128,  0.0084,
        -0.0313, -0.0249, -0.0139,  0.0065, -0.0071, -0.1977,  0.0100, -0.0174,
        -0.1140, -0.0105, -0.0857,  0.0138, -0.0901, -0.1259,  0.0223, -0.0243,
         0.0015,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0554, -0.0108, -0.0315, -0.0243, -0.0089,  0.0077, -0.1099, -0.1047,
        -0.0890, -0.3228,  0.0120, -0.0351, -0.0657, -0.0238,  0.0297, -0.0685,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0517, -0.2507,  0.0110, -0.1056, -0.0280, -0.0766, -0.1510, -0.0020,
         0.0309, -0.0038,  0.0084,  0.0010, -0.0255,  0.0101,  0.0193, -0.0137,
        -0.0545, -0.0814, -0.0279, -0.0088, -0.0078,  0.0262,  0.0041,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0649,  0.0286,  0.0096, -0.2105, -0.4227, -0.0080,  0.0550,  0.0060,
        -0.0329,  0.0806,  0.0139,  0.0395, -0.0107,  0.0173,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
