Iter #50: [tensor([ 6.8721e-02,  7.1904e-02,  4.2431e-02,  2.0692e-02, -1.7658e-02,
        -2.9538e-02, -3.5847e-02,  4.7896e-02, -7.5105e-02,  4.8642e-03,
         2.8653e-02,  1.5821e-02,  2.5578e-02,  1.5290e-02, -1.0983e-02,
        -3.5684e-05,  1.0293e-02,  3.9181e-03, -1.6330e-02, -4.8865e-04,
         6.8778e-02,  8.6127e-03,  5.7414e-03,  4.2891e-02,  9.1948e-03,
        -6.6102e-03,  1.9964e-02, -1.1516e-02,  1.9125e-02,  1.9570e-03,
         8.7801e-03, -2.3414e-02, -1.3036e-02, -3.6636e-02,  6.3435e-03,
         3.0398e-02, -3.3528e-02,  1.3263e-02,  2.2473e-02,  3.5347e-02,
         1.1826e-02,  4.1877e-03,  8.7624e-03,  1.5571e-02,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2432,  0.0548, -0.0056,  0.0494,  0.0013,  0.0449,  0.1117,  0.0054,
         0.0472,  0.0141,  0.0266,  0.0365,  0.0326,  0.0405,  0.0465,  0.0054,
        -0.0013, -0.0195, -0.0216, -0.0269, -0.0176, -0.0798, -0.0210,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1882,  0.0242, -0.1561,  0.0895, -0.0531,  0.2471, -0.0369,  0.0331,
         0.0722,  0.0160, -0.0270, -0.0331,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0594, -0.0686, -0.0915, -0.0105,  0.0069,  0.0583,  0.0340,  0.0003,
         0.0178,  0.0147,  0.0631, -0.1106,  0.0032,  0.0710, -0.0296, -0.0807,
        -0.0188,  0.0311,  0.0449, -0.0624,  0.0337,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.2006,  0.1347, -0.0268, -0.0146,  0.0525,  0.0919,  0.0923,  0.2065,
         0.1797,  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,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.3847,  0.0186, -0.0266,  0.0316, -0.0079, -0.0071,  0.0345, -0.0114,
        -0.0668, -0.0108, -0.0087,  0.0020,  0.0146,  0.0065, -0.0233, -0.0251,
        -0.0148,  0.0358,  0.0120,  0.0284, -0.0198,  0.1640, -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([ 0.1066, -0.0308,  0.0595, -0.0357, -0.0044,  0.0058, -0.0064,  0.0489,
        -0.0072, -0.0080,  0.0009,  0.0033,  0.0234,  0.0187, -0.0315, -0.0357,
        -0.0046, -0.0344, -0.0151, -0.0079,  0.0402,  0.0278, -0.0073,  0.0014,
         0.0136,  0.0246, -0.0168,  0.0151, -0.0007,  0.0008, -0.0142,  0.0364,
        -0.0218, -0.0041, -0.0139, -0.0272, -0.0496, -0.0246,  0.0287,  0.0563,
         0.0292,  0.0088, -0.0148,  0.0230,  0.0101], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0522, -0.0500,  0.0314, -0.0827,  0.0414,  0.0109, -0.0445, -0.0233,
        -0.0618, -0.0010, -0.0272, -0.0227,  0.0341, -0.0478, -0.0148,  0.0663,
        -0.1100,  0.0530,  0.1567, -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,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0483,  0.0590,  0.0371, -0.0788,  0.0258, -0.0195, -0.0253, -0.0182,
        -0.0104,  0.0402,  0.0247,  0.0179,  0.0108, -0.0003, -0.0225,  0.0057,
        -0.0026,  0.0284, -0.0420, -0.0142, -0.0778, -0.0110, -0.0374, -0.0625,
         0.1272, -0.1527,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  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,  0.0404,  0.0596, -0.0153,  0.0636,  0.1004,  0.0099,  0.0329,
         0.0053,  0.0286,  0.0407,  0.0593,  0.0192,  0.0297,  0.0199,  0.0376,
         0.0276,  0.0221, -0.0320, -0.0011,  0.0100,  0.0204, -0.0173,  0.0448,
        -0.0069,  0.0204,  0.0991, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0005, -0.0659, -0.0012, -0.2823, -0.1223,  0.0690,  0.0521,  0.3678,
         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,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0511,  0.0008, -0.0095, -0.1276, -0.0098,  0.0956, -0.0145, -0.0247,
        -0.0011,  0.0262,  0.0596, -0.0138, -0.0284,  0.0347, -0.0333,  0.0274,
        -0.1045,  0.0991, -0.0196,  0.0142, -0.0159, -0.0203, -0.0003, -0.0024,
         0.0284,  0.0963,  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], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([-0.0256, -0.0602, -0.0072, -0.0205,  0.0045, -0.0149,  0.0167, -0.0224,
         0.0029,  0.0028,  0.0149,  0.0187,  0.0061,  0.0265,  0.0220,  0.0161,
         0.0313,  0.0063,  0.0369,  0.0295, -0.0408,  0.0598,  0.2064,  0.0792,
         0.0173,  0.0194, -0.0413, -0.0120,  0.0158, -0.0022,  0.0118,  0.0288,
        -0.0009,  0.0076, -0.0102, -0.0124,  0.0396,  0.0086,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0561,  0.0115,  0.0035, -0.0094, -0.0090,  0.0159, -0.0050, -0.0089,
         0.0124,  0.0129, -0.0275,  0.0024, -0.0043, -0.0117, -0.0121,  0.0128,
        -0.0073,  0.0035, -0.0589,  0.0158, -0.0342, -0.0132,  0.0574, -0.0489,
        -0.1070,  0.0491, -0.1025, -0.0683, -0.0214, -0.0057,  0.0105, -0.0007,
        -0.0218, -0.0284, -0.0025,  0.0099,  0.0046,  0.1128,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0093, -0.2077,  0.0354, -0.0897,  0.0166, -0.0067, -0.0262, -0.0081,
        -0.0182, -0.0190, -0.0111,  0.0069, -0.0102, -0.0311, -0.0050, -0.0082,
         0.0138, -0.0141, -0.0412,  0.0514, -0.0081, -0.0332,  0.0491, -0.0298,
         0.0090,  0.0449,  0.0169,  0.0990, -0.0558, -0.0240,  0.0000,  0.0000,
         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,  0.0070, -0.0331,  0.0508,  0.0169,  0.0423,  0.0084, -0.0014,
         0.0347,  0.0037,  0.0037, -0.0017, -0.0128,  0.0167,  0.0044,  0.0103,
         0.0108,  0.0076,  0.0138,  0.0256,  0.0122, -0.0155,  0.0122,  0.0882,
         0.1288, -0.0233,  0.0387,  0.0167, -0.0045, -0.0072, -0.0254,  0.0502,
         0.0144, -0.0013,  0.0194,  0.0044, -0.0606, -0.0534,  0.0390,  0.0018],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0847, -0.0066, -0.0545,  0.0244,  0.0107, -0.0036, -0.0225, -0.0073,
        -0.0309, -0.0068, -0.0084,  0.0059,  0.0378,  0.0067, -0.0241,  0.0216,
        -0.0190, -0.0124, -0.0564,  0.0033, -0.0349, -0.0105, -0.0268,  0.0137,
        -0.0351, -0.0014, -0.0165, -0.0042, -0.0133,  0.0014,  0.0063,  0.0425,
        -0.0039, -0.0438,  0.0630,  0.0408, -0.1657,  0.0285,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0274,  0.0073,  0.0613,  0.1493,  0.0480,  0.0367,  0.0651, -0.0509,
         0.0116,  0.0172,  0.0197,  0.0610,  0.0401,  0.0434,  0.0567,  0.0525,
        -0.0395, -0.0057,  0.0088,  0.0064, -0.0252,  0.0649, -0.0418,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0709,  0.0959,  0.0981,  0.0093,  0.0014,  0.0145, -0.0041, -0.0480,
         0.0145,  0.0698,  0.0218,  0.0471, -0.0178,  0.0098,  0.0022,  0.0428,
        -0.0275, -0.0763,  0.0204, -0.0382, -0.0124,  0.0285, -0.0005,  0.0167,
         0.0120,  0.0210,  0.0019,  0.0163,  0.0263, -0.0183,  0.0090, -0.1068,
         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, -0.0320,  0.1130,  0.1067, -0.0684, -0.0218,  0.1727, -0.0931,
         0.0185, -0.0071, -0.0914, -0.0396,  0.1508, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0231,  0.0616,  0.0288, -0.0568, -0.0890,  0.0296, -0.0040,  0.0201,
         0.0454, -0.0021,  0.0492,  0.0188, -0.0408, -0.0300, -0.0332, -0.0595,
        -0.0520, -0.0185, -0.0006,  0.0303, -0.0457,  0.0895,  0.1591,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1180,  0.1991,  0.0754,  0.0019, -0.0501, -0.0332,  0.0282,  0.0356,
        -0.0214, -0.0218, -0.0028,  0.0088, -0.0307, -0.0307,  0.0264,  0.0122,
         0.1380,  0.0436,  0.0289, -0.0469, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2932,  0.0878,  0.0616, -0.0069, -0.0124, -0.0018, -0.0489,  0.0436,
         0.0072,  0.0165, -0.0047,  0.0254,  0.0344,  0.0798, -0.0360,  0.0304,
         0.0021,  0.0008, -0.0028,  0.0238,  0.0178,  0.0096,  0.0068, -0.0103,
         0.0442,  0.0753, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0710, -0.0686,  0.0360,  0.0428,  0.0933,  0.0192,  0.0036,  0.0075,
         0.0689,  0.0620,  0.0591,  0.0389,  0.0252, -0.0010, -0.1017,  0.0214,
         0.1581,  0.1216,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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([ 0.0159, -0.0253,  0.0471,  0.0894,  0.0137,  0.0332, -0.0214, -0.0097,
         0.0038,  0.0110, -0.0460,  0.0316,  0.0007, -0.0177,  0.0047,  0.0062,
         0.0028,  0.0224, -0.0055,  0.0528,  0.0174,  0.0748,  0.0156, -0.0053,
         0.0014, -0.0104, -0.0153,  0.0081,  0.0080,  0.0039,  0.0136,  0.0086,
         0.0291,  0.0061, -0.0048, -0.0175,  0.0426, -0.0080, -0.0527,  0.0892,
        -0.0206,  0.0074,  0.0025, -0.0006,  0.0354,  0.0404,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1037,  0.0022,  0.0092,  0.0214, -0.0183,  0.0154,  0.0253, -0.0230,
         0.0463, -0.0149,  0.0092, -0.0405,  0.0258,  0.0051,  0.0122, -0.0063,
         0.0114,  0.0129,  0.0036,  0.0309,  0.0058, -0.0556,  0.0286, -0.0305,
         0.0791,  0.0093,  0.0117,  0.0346, -0.0310, -0.0026, -0.0090,  0.0075,
         0.0238,  0.0023,  0.0121,  0.0016,  0.0061,  0.0259,  0.0053,  0.0069,
        -0.0118,  0.0091,  0.0314,  0.0285,  0.0263,  0.0445,  0.0185,  0.0029],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0531, -0.0281,  0.0707, -0.0209,  0.0168, -0.1294,  0.0219,  0.0837,
         0.0010,  0.0368,  0.0357,  0.0162,  0.0245,  0.0216, -0.0015, -0.0177,
         0.0144, -0.0068,  0.0218,  0.0103,  0.0329, -0.0564, -0.0390,  0.0101,
        -0.0067,  0.0053, -0.0207,  0.0162,  0.0232,  0.0067, -0.0075,  0.0614,
         0.0183,  0.0629,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2223, -0.0965, -0.3512, -0.1626,  0.0816, -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,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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,  0.0506,  0.0133,  0.0286, -0.0263,  0.0530,  0.0059,  0.0148,
        -0.0043,  0.0014, -0.0040,  0.0007,  0.0028,  0.0463,  0.0479,  0.0260,
         0.0450, -0.0146,  0.0351,  0.0341,  0.0100, -0.0717, -0.0008, -0.0043,
        -0.0279,  0.0239,  0.0210,  0.0070,  0.0361,  0.0235, -0.0008,  0.0066,
         0.0046, -0.0142,  0.0023, -0.0050,  0.0059, -0.0209,  0.0148, -0.0036,
         0.0077, -0.0015,  0.0138,  0.0173,  0.0648, -0.0417,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.6865e-02,  4.9803e-02, -2.5296e-02,  1.5047e-02,  2.5715e-02,
         1.7657e-03, -2.5694e-02,  3.5970e-05, -6.0436e-03, -3.9237e-02,
        -1.5360e-02, -4.0396e-03,  5.5185e-02,  1.2763e-02, -1.0065e-02,
        -4.3794e-03,  7.5812e-02, -2.3149e-02,  1.0178e-01,  2.1892e-03,
        -2.1471e-01,  2.6507e-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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.4489,  0.0128,  0.0412,  0.0231,  0.1795,  0.0176, -0.0328,  0.0771,
        -0.0961, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1226,  0.0340,  0.0045,  0.0208,  0.1052, -0.0663,  0.0641, -0.0545,
        -0.0420,  0.0353,  0.0023, -0.0314, -0.1093, -0.1442,  0.0227, -0.1085,
         0.0263, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2177, -0.0243,  0.0525,  0.0561,  0.0487,  0.0211,  0.0754, -0.0006,
         0.0415, -0.0395,  0.1183, -0.0269, -0.0797,  0.0391,  0.0170, -0.0568,
        -0.0849,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1218, -0.1169, -0.0220, -0.0623,  0.0165, -0.0201, -0.0114,  0.0083,
        -0.0190,  0.1161, -0.0030,  0.0698, -0.0330,  0.2179, -0.0502, -0.1119,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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, -0.1690,  0.0305, -0.0606, -0.0747, -0.0266,  0.0647,  0.1549,
        -0.0937, -0.1147,  0.0198, -0.0120, -0.0245,  0.0068, -0.0014, -0.0060,
        -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2107,  0.0417, -0.1106, -0.0327, -0.1541,  0.0129, -0.0293,  0.0247,
         0.0291, -0.1166,  0.0173,  0.0708, -0.0183,  0.0125, -0.1187,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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([ 0.2042,  0.0168,  0.0499,  0.0344,  0.1041, -0.0467,  0.0168,  0.1103,
         0.0263,  0.0121, -0.0602, -0.0273,  0.0149, -0.0697,  0.2063,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  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, -0.0796,  0.0539,  0.0515,  0.0831, -0.0592, -0.0271,  0.0733,
         0.0259,  0.1064,  0.0370, -0.0394,  0.0034, -0.0103,  0.0301, -0.0326,
        -0.1126,  0.0048,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1147, -0.0325, -0.0825,  0.0068,  0.0522,  0.0309,  0.0267, -0.0193,
        -0.0019,  0.1199,  0.0762,  0.0663,  0.0284,  0.0398,  0.0701,  0.1368,
        -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([-0.0846,  0.0198,  0.0027,  0.0261,  0.0203, -0.0233, -0.0043,  0.0446,
         0.0259, -0.0250, -0.0162, -0.0299, -0.0085,  0.0046, -0.0388,  0.0250,
         0.0344,  0.0160,  0.0480, -0.0003,  0.0031,  0.0132, -0.0042,  0.0414,
        -0.0101,  0.0146,  0.0085,  0.0015, -0.0175,  0.0021, -0.0140, -0.0066,
        -0.0163, -0.0151,  0.0137, -0.0170,  0.0192, -0.0048,  0.0661, -0.0072,
         0.0140,  0.0073, -0.0130, -0.0037,  0.0175, -0.0109, -0.0516, -0.0362,
        -0.0126, -0.0388], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1055,  0.0478,  0.1266, -0.0063, -0.0274,  0.0170, -0.0212, -0.1179,
         0.0513,  0.0881, -0.1283, -0.0845, -0.0258,  0.0312,  0.0430,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1359,  0.1851, -0.0019,  0.0290,  0.0010,  0.0667,  0.0075, -0.0207,
         0.0647,  0.0231, -0.0508,  0.0068, -0.0241, -0.0222,  0.0209, -0.1885,
         0.0692, -0.0487, -0.0114, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1047, -0.0982,  0.1096,  0.1176,  0.0043, -0.0121, -0.0080, -0.0188,
        -0.0396, -0.0136, -0.0012,  0.0004, -0.0108,  0.0037, -0.0524,  0.0362,
        -0.0147,  0.0166,  0.0203, -0.0026, -0.1044, -0.0338,  0.0663,  0.0761,
         0.0013,  0.0073, -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([-0.2369, -0.2576,  0.0304, -0.0282, -0.1030,  0.0226,  0.0145,  0.0121,
         0.0917,  0.0345,  0.0606,  0.0449, -0.0441, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1587, -0.0329, -0.0178, -0.0889,  0.0612,  0.0177, -0.0677, -0.0365,
         0.0074, -0.0578, -0.0340, -0.0282, -0.1759, -0.0434,  0.0302, -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,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2319, -0.0279,  0.0212, -0.0930,  0.0082,  0.0309,  0.0091,  0.0768,
        -0.0427,  0.0266, -0.0110, -0.0093,  0.0207, -0.0092, -0.0080, -0.0852,
         0.0541,  0.0109, -0.0365,  0.0101,  0.0097, -0.0010, -0.0034, -0.0060,
         0.0138,  0.0583,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0005, -0.0678,  0.0730, -0.0288,  0.0395, -0.0089,  0.0174, -0.0640,
         0.0686, -0.0286,  0.0138,  0.0548,  0.0330, -0.0242,  0.0285,  0.0388,
         0.0339, -0.0057,  0.0276,  0.0411,  0.0150,  0.0004,  0.0377, -0.0211,
         0.0784, -0.1488,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  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, -0.0172,  0.0133,  0.0044, -0.0473, -0.1229,  0.0046,  0.0407,
         0.0590,  0.0244, -0.0338,  0.0583, -0.1100,  0.0203,  0.0066,  0.0271,
        -0.0129,  0.0356, -0.0225, -0.0345,  0.0262, -0.0332,  0.0092,  0.0139,
         0.0298,  0.0227,  0.0076, -0.0737,  0.0382,  0.0040,  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], device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([ 0.0032, -0.0706,  0.1405,  0.0946,  0.0029,  0.0561, -0.0342,  0.0120,
        -0.1002,  0.0399,  0.0292,  0.0485,  0.0699,  0.0352,  0.0172, -0.0588,
        -0.1069,  0.0227,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0743, -0.0021,  0.0612, -0.3639, -0.0852, -0.0720,  0.1293,  0.0245,
        -0.1873,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1103,  0.0779,  0.0227, -0.0248,  0.0111, -0.0517,  0.0124, -0.0266,
        -0.0076,  0.0068,  0.0402, -0.0665,  0.0336, -0.0135, -0.0400,  0.0200,
         0.0229, -0.0183,  0.0306,  0.3013, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0824, -0.0514,  0.0073,  0.0243,  0.0608, -0.0182, -0.0943, -0.0603,
         0.0134,  0.0189, -0.0322,  0.0467, -0.0117,  0.0266,  0.0051,  0.0394,
         0.0200,  0.0321,  0.0264,  0.0269, -0.0100,  0.0461,  0.0112, -0.0306,
         0.0724,  0.1086, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0111, -0.0064,  0.0064, -0.0012, -0.0204,  0.0191,  0.1499, -0.0672,
         0.0230,  0.0458,  0.0205,  0.0431, -0.0298,  0.0110,  0.0170,  0.0152,
        -0.0125, -0.0158,  0.0204,  0.0369, -0.0115, -0.0111,  0.0254,  0.0462,
        -0.0099, -0.0258,  0.0371,  0.0906,  0.0093, -0.0396, -0.0155, -0.0040,
        -0.0139, -0.0076, -0.0154, -0.0080, -0.0064,  0.0194, -0.0308,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0036, -0.0820, -0.0552, -0.0406,  0.0644,  0.0964,  0.0164, -0.1023,
        -0.0231, -0.0576,  0.0623,  0.0016, -0.1001, -0.0030,  0.0362,  0.0218,
         0.0995, -0.0841,  0.0007, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0345, -0.0898, -0.0153,  0.0662,  0.0471,  0.1273,  0.0296,  0.0021,
         0.0450,  0.0295, -0.0590, -0.0367, -0.1027,  0.0045, -0.0144,  0.0083,
        -0.0290, -0.0117, -0.0376, -0.0293,  0.0467, -0.0148, -0.1190,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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, -0.0056,  0.0594,  0.0110,  0.0469,  0.0381, -0.0179,  0.0089,
         0.0131,  0.0250,  0.0062, -0.0012,  0.0084, -0.0794, -0.0367, -0.0626,
        -0.0328,  0.0065, -0.0051,  0.0297,  0.0205,  0.0148,  0.0042,  0.0427,
         0.0253, -0.0225, -0.0313, -0.0539,  0.0171, -0.0241, -0.0043, -0.0126,
        -0.0227,  0.0009, -0.0103, -0.0339, -0.0078,  0.0368,  0.1048,  0.0115],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1047, -0.0417, -0.0051, -0.0141,  0.0988,  0.0340,  0.0247,  0.0809,
        -0.0251,  0.0083,  0.0043, -0.0022, -0.0049,  0.0112,  0.0186, -0.0094,
         0.0098,  0.0177, -0.0250, -0.0098,  0.0253, -0.0059,  0.0085,  0.0879,
         0.0160,  0.0138,  0.0892, -0.0093, -0.0035, -0.0128, -0.0019,  0.0156,
        -0.0269, -0.0214, -0.0162,  0.0955,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0603, -0.0294, -0.0332,  0.0336,  0.0244, -0.0261,  0.0458, -0.0018,
         0.0022,  0.0221,  0.0009, -0.0101,  0.0138, -0.0125, -0.0090, -0.0200,
         0.0006,  0.0504,  0.0776,  0.0046,  0.0200,  0.0264,  0.0308, -0.0120,
         0.0022,  0.0551, -0.1398,  0.0100,  0.0198,  0.0270, -0.0573,  0.1211,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2468,  0.1346, -0.0180, -0.0006,  0.0393, -0.0229,  0.2032,  0.0165,
        -0.0468,  0.2714,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1380,  0.0618, -0.2111,  0.0314,  0.1202, -0.0594, -0.0084,  0.0398,
        -0.0989,  0.0325, -0.0989, -0.0073,  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],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([ 0.1208,  0.0003,  0.0351,  0.0382,  0.1264,  0.1245, -0.1028, -0.0141,
         0.0228,  0.0131,  0.0386, -0.0610,  0.1194, -0.0380,  0.1450,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0624, -0.0142,  0.0263, -0.0424, -0.0679, -0.0220, -0.0322,  0.0149,
         0.0211, -0.0420,  0.0526, -0.0017,  0.0374,  0.0121,  0.1049, -0.0016,
         0.0712,  0.0286,  0.0178,  0.0009,  0.0122,  0.0191, -0.0575, -0.0180,
        -0.0228,  0.0017,  0.0022, -0.0347, -0.0223, -0.0792,  0.0019, -0.0341,
        -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1165,  0.1563,  0.1106,  0.1039,  0.0205,  0.3118, -0.0815, -0.0988,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  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,  0.0046,  0.0036, -0.0346, -0.0518, -0.0172,  0.0688, -0.0154,
        -0.0230,  0.0234, -0.0182,  0.0218,  0.0159, -0.0649, -0.0273, -0.0236,
        -0.0007,  0.0207, -0.0240, -0.0227,  0.0090,  0.0062,  0.0264,  0.0642,
         0.0340,  0.0211,  0.0423,  0.0083,  0.2607, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0156,  0.0948, -0.0655, -0.0309,  0.2570,  0.0266, -0.0146,  0.0271,
         0.0116,  0.0214, -0.0023, -0.0012, -0.2447, -0.0066,  0.0204,  0.0905,
         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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.3772e-01,  3.2616e-02,  5.6123e-02, -4.1578e-02, -5.5335e-02,
         1.1292e-01,  3.9109e-02, -1.9971e-02, -8.9709e-03,  2.6632e-04,
         3.3577e-02, -1.0910e-02,  1.9693e-02,  1.2205e-02, -3.4740e-02,
        -4.8887e-02, -2.8358e-03, -1.2015e-02,  6.3398e-02, -9.5892e-05,
        -4.8326e-02, -1.1023e-01, -5.1931e-02,  4.6540e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.9054e-02,  2.9583e-02, -5.0049e-02, -2.5291e-02,  4.0912e-02,
        -4.6915e-05,  3.2591e-03, -2.5134e-02,  9.4674e-03,  1.8098e-02,
         3.7670e-03, -1.0111e-02, -1.6981e-02,  2.5653e-02,  7.2086e-03,
        -2.9982e-02, -6.3591e-05,  1.4855e-02,  8.0317e-03, -1.6186e-02,
         3.3022e-03,  2.6513e-03,  1.6245e-02,  1.9077e-02, -1.7177e-03,
         2.9290e-02,  1.6957e-02, -1.9141e-02, -4.4589e-02, -1.4872e-02,
        -2.8714e-02, -1.9012e-02,  2.3791e-03,  6.4907e-02, -6.0680e-03,
         3.7701e-02, -6.8649e-02,  1.4146e-01,  6.4985e-03, -2.5654e-02,
         5.1367e-02, -1.6017e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+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,  0.0150,  0.0082, -0.0418,  0.0211, -0.0563,  0.0347, -0.0227,
         0.0095,  0.0090, -0.0271, -0.0110, -0.0065,  0.0018, -0.0076, -0.0371,
        -0.0361, -0.0028,  0.0137, -0.0107, -0.0166, -0.0434, -0.0103,  0.0094,
        -0.0085,  0.0261,  0.0291,  0.0236,  0.0134, -0.0540, -0.0073,  0.0080,
         0.0790, -0.0571,  0.0083,  0.0307,  0.0549,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.2396, -0.1012, -0.0010, -0.2928,  0.0828,  0.0385,  0.0169,  0.0435,
         0.0208,  0.1536, -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,  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,  0.2509,  0.0198, -0.0116, -0.0372,  0.0398, -0.0125,  0.0022,
        -0.0196,  0.0204,  0.0150,  0.0044, -0.0130,  0.0090,  0.0036, -0.0029,
         0.0510,  0.0072, -0.0158, -0.0034, -0.0036,  0.0113, -0.0032,  0.0008,
        -0.0026, -0.0032,  0.0009, -0.0561, -0.0035, -0.0472,  0.0557, -0.0869,
         0.1082,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.1717e-01,  4.5267e-02, -5.5551e-03, -1.7677e-02,  2.6908e-02,
         1.2940e-02,  8.9173e-03, -2.5529e-02,  1.3948e-05,  5.1066e-02,
         5.1393e-02,  1.6412e-02, -2.1732e-02, -8.8897e-03, -3.7721e-02,
        -2.4104e-02, -5.6647e-03,  4.0830e-04, -3.9874e-02,  7.0721e-03,
         8.7951e-03, -4.8170e-03, -1.7446e-03, -2.9186e-03,  2.2543e-03,
        -7.8674e-03, -1.3550e-04,  7.1774e-03,  2.7647e-02,  2.0056e-03,
         4.3491e-02, -3.9299e-02, -1.6576e-03,  3.9569e-02,  9.1851e-03,
         1.3271e-02, -3.0114e-02, -7.5885e-05,  8.7288e-04,  1.9610e-02,
        -2.4635e-02, -1.9981e-03,  5.5511e-03,  4.7159e-03, -3.1820e-03,
         1.4550e-03,  1.7261e-02,  2.7527e-02, -9.4973e-03,  5.6151e-03,
        -1.1738e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.0963e-02, -3.1282e-01, -2.8351e-02,  8.9924e-03, -1.3358e-02,
         1.1804e-02,  2.1604e-02, -1.9060e-02,  2.7349e-02, -2.9610e-02,
        -4.1622e-02, -2.2396e-03,  4.2279e-02, -4.7449e-02,  1.1558e-02,
         2.8299e-02, -6.7064e-04,  1.0217e-02, -5.9703e-03,  1.0118e-02,
        -2.1216e-02, -1.7667e-02,  1.9403e-02,  5.6101e-03,  5.0222e-03,
         1.1826e-02,  1.6243e-02,  1.2487e-03, -5.7537e-03, -4.4859e-03,
        -8.2887e-03, -2.1629e-04, -1.0622e-02, -5.2017e-03,  2.2492e-02,
        -4.5234e-02,  6.5142e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         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([-0.0625, -0.0104,  0.1471,  0.0397, -0.0137,  0.0105, -0.0093, -0.0398,
        -0.0414, -0.0179, -0.0262, -0.0149,  0.0177,  0.0070, -0.0043,  0.0175,
         0.0388, -0.0085,  0.0064,  0.0074,  0.0458,  0.0160, -0.0341,  0.0935,
        -0.0430, -0.0514, -0.0112,  0.0048,  0.0183, -0.0257, -0.0210, -0.0026,
        -0.0338,  0.0164, -0.0079, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1006, -0.0174,  0.0078,  0.0288,  0.0032, -0.0051,  0.0218, -0.0108,
         0.0111,  0.0036,  0.0215, -0.0302,  0.0035,  0.0033,  0.0023,  0.0242,
         0.0019,  0.0284,  0.0109,  0.0292,  0.0142,  0.0432, -0.0082, -0.0101,
         0.0327, -0.0048, -0.0069,  0.0103, -0.0079, -0.0042, -0.0155,  0.0652,
         0.0006, -0.0018,  0.0086,  0.0197,  0.0173, -0.0161, -0.0065, -0.0023,
         0.0211, -0.0043,  0.0071, -0.0048,  0.0073, -0.0089, -0.0140,  0.0741,
         0.0021, -0.0132,  0.0021,  0.0069,  0.0446, -0.0621,  0.0658],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1301, -0.0826, -0.0300, -0.0008, -0.0336,  0.0093,  0.0167, -0.0359,
        -0.0687, -0.0396, -0.0118, -0.0331,  0.0470, -0.0610,  0.0726, -0.0057,
         0.0054, -0.0133,  0.0032,  0.0454, -0.0120, -0.0357, -0.0155, -0.0487,
        -0.1325,  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([-0.0945, -0.0160, -0.0797,  0.0211, -0.0266, -0.0640,  0.0378,  0.0121,
         0.0105, -0.0262, -0.0117, -0.0186, -0.0190,  0.0382,  0.1459,  0.0331,
         0.0706, -0.0131, -0.0466,  0.0529, -0.0120,  0.0945,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0960,  0.1042,  0.0024,  0.1191, -0.0073,  0.0508,  0.0120,  0.0386,
        -0.0332, -0.0268,  0.0356,  0.0104,  0.0525, -0.0036, -0.0283,  0.0325,
         0.0083,  0.0381,  0.0216, -0.0252, -0.0043,  0.0188, -0.0289,  0.0248,
         0.0042,  0.0047,  0.0965,  0.0448,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1009,  0.0128, -0.1730,  0.0165, -0.1222, -0.3293,  0.2452,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0752, -0.0230,  0.1374,  0.0217,  0.0394, -0.2365,  0.0265, -0.0532,
         0.0321, -0.1089, -0.0123,  0.0043, -0.0126, -0.0430, -0.0273, -0.0036,
        -0.0075,  0.0104, -0.1250,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  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,  0.0058, -0.0649, -0.0634, -0.0143,  0.0521,  0.0216,  0.0501,
        -0.0684,  0.0109,  0.0113, -0.0405,  0.0022,  0.0253, -0.0211, -0.0729,
        -0.0117,  0.0020, -0.0086, -0.0210, -0.0412, -0.0195, -0.0256,  0.0186,
        -0.0352, -0.0227, -0.1134,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.5141e-03, -6.3919e-02, -5.8708e-02,  7.9197e-02,  5.9647e-02,
         5.3126e-02,  8.3140e-02, -6.9983e-03,  1.5693e-03, -3.3131e-02,
        -5.6506e-02, -6.0947e-03, -3.4218e-02,  1.0987e-02, -8.6065e-03,
        -2.2325e-02, -4.7304e-03, -2.4350e-02,  1.3839e-02, -1.8683e-02,
        -4.3680e-02,  1.7522e-02, -2.1734e-02, -1.0449e-02, -6.2870e-05,
        -9.1003e-04,  2.2747e-02, -3.7562e-02, -6.9277e-02,  1.5761e-03,
        -1.2819e-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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1629, -0.1104,  0.1149,  0.1511,  0.1186,  0.0593, -0.0241, -0.0590,
         0.0027,  0.0737, -0.0049,  0.0334,  0.0210,  0.0540, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0380, -0.0169,  0.1886, -0.0351, -0.1431, -0.0387, -0.0764,  0.0259,
        -0.0497,  0.0047, -0.0138, -0.0011,  0.0164, -0.0450,  0.0251,  0.1085,
         0.1258,  0.0455,  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,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0077,  0.0213,  0.0744,  0.0959, -0.1288, -0.0901,  0.0255,  0.0597,
         0.0458,  0.0234, -0.0066,  0.0113,  0.0197, -0.0040, -0.0437, -0.0479,
         0.0262, -0.0480,  0.0033, -0.0153, -0.0099,  0.0039,  0.0158,  0.0030,
        -0.0028, -0.0003,  0.0202,  0.0127, -0.0980,  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],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([ 0.0762, -0.1906,  0.0069,  0.0663, -0.0038, -0.0638, -0.0020, -0.0398,
         0.0077, -0.1176, -0.0525, -0.0181,  0.0626,  0.0160,  0.0350, -0.0741,
        -0.0357,  0.0016,  0.0589,  0.0454,  0.0119,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0499,  0.2114, -0.0032,  0.0303,  0.0282, -0.0393, -0.0692, -0.0519,
        -0.0011, -0.1006,  0.0943, -0.0340, -0.0177,  0.0304,  0.0233, -0.0753,
         0.0242,  0.0175, -0.0348, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0208, -0.2100,  0.0977, -0.0653,  0.0266, -0.0226, -0.0457, -0.0085,
        -0.0173,  0.0049,  0.0116, -0.0014, -0.0056,  0.0435, -0.0093,  0.0041,
         0.0026, -0.0143, -0.0140, -0.0180,  0.0043,  0.0345, -0.0054, -0.0557,
        -0.2562,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1211,  0.0196, -0.0346,  0.0474, -0.0259,  0.0703,  0.0423, -0.0364,
         0.0691, -0.0167, -0.0232,  0.0308,  0.0069,  0.0189,  0.0013, -0.0331,
         0.0439, -0.0211,  0.0118, -0.0521, -0.0236, -0.0195, -0.0260, -0.0013,
         0.0071, -0.0031,  0.0682,  0.0569,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0370, -0.0572,  0.0351,  0.0678,  0.0696,  0.0274,  0.0661,  0.0244,
         0.3296,  0.1432,  0.1427,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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,  0.1284,  0.0187, -0.0048,  0.0068, -0.0673, -0.0144,  0.0402,
         0.0180,  0.0284,  0.0542,  0.0242, -0.0230,  0.0474,  0.1145,  0.0349,
         0.0547,  0.0195,  0.0089,  0.0101,  0.0525,  0.0896, -0.0458,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0795,  0.0441, -0.0612,  0.0531,  0.0311,  0.0109,  0.0234, -0.0085,
        -0.0048,  0.0117,  0.0023,  0.0296, -0.0006,  0.0248,  0.0042,  0.0079,
        -0.0066,  0.0036,  0.0116,  0.0125,  0.0136,  0.0003,  0.0414, -0.0060,
         0.0175,  0.0035, -0.0222,  0.0092, -0.0305,  0.0825,  0.0056,  0.0465,
         0.0226,  0.0031, -0.0206, -0.0018, -0.0350,  0.0027,  0.0246, -0.0073,
        -0.0056,  0.0323, -0.0023, -0.0132,  0.0242,  0.0050, -0.0095, -0.0793],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1383,  0.0587,  0.0215, -0.0049, -0.0304, -0.0143, -0.0095, -0.0529,
        -0.2048, -0.0579, -0.1131,  0.0227, -0.0649, -0.0199, -0.0648,  0.0236,
         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,  0.0000,  0.0000,
         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, -0.2372, -0.0742,  0.1189, -0.0380, -0.0373,  0.0604,  0.0274,
         0.0342, -0.0876, -0.0160, -0.0725, -0.0384,  0.1175,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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, -0.1588, -0.0706, -0.0132, -0.0171, -0.0044,  0.0002, -0.0047,
        -0.0170, -0.0967,  0.0146, -0.0382, -0.0328, -0.0068, -0.0469, -0.0141,
        -0.0319,  0.0278, -0.0267, -0.0021, -0.0021, -0.0264, -0.0127,  0.0014,
        -0.0018, -0.0343, -0.0016, -0.0348,  0.0257, -0.0068, -0.0093,  0.0153,
        -0.0889, -0.1119,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1442, -0.0507, -0.0178,  0.0339, -0.0458, -0.0393, -0.0178, -0.0857,
         0.0093,  0.0008, -0.0126,  0.0086, -0.0122, -0.0034,  0.0220,  0.0827,
         0.0081,  0.0004, -0.0076,  0.0140,  0.0209,  0.0139,  0.0044, -0.0111,
         0.0065,  0.0111,  0.0023, -0.0050, -0.0051,  0.0283, -0.0010,  0.0232,
         0.0417,  0.0266, -0.0212,  0.0236, -0.0118,  0.0493,  0.0159,  0.0109,
         0.0197,  0.0241,  0.0055,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1943, -0.0274,  0.0467,  0.0278,  0.0361, -0.0027, -0.0056, -0.0190,
         0.0203,  0.0631, -0.0091, -0.0103,  0.0150,  0.0025, -0.0168,  0.0186,
        -0.0625,  0.0197,  0.0006,  0.0125,  0.0094,  0.0177, -0.0211, -0.0153,
        -0.0295, -0.0308,  0.1090, -0.0118, -0.0224, -0.0298, -0.0005, -0.0090,
         0.0022,  0.0533, -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],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #450: [tensor([ 0.0486,  0.2606,  0.0526,  0.0596, -0.0056,  0.0068, -0.0087,  0.0179,
         0.0336,  0.0753, -0.0404, -0.0642, -0.0066,  0.0267,  0.1050,  0.0512,
         0.0145,  0.0083, -0.0096,  0.0064, -0.0070, -0.0246, -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,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0337,  0.1037,  0.0778, -0.0910,  0.0336,  0.0172,  0.0167, -0.0607,
         0.0237, -0.0029,  0.0261,  0.0413, -0.0121, -0.0197,  0.0153, -0.0741,
        -0.0071,  0.0043, -0.0577, -0.0103, -0.0107, -0.0061,  0.0532, -0.0039,
         0.0130, -0.0232,  0.0086, -0.0196, -0.0029, -0.0051, -0.0083,  0.0040,
         0.0106, -0.0250,  0.0258, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.2067e-02,  3.4903e-01,  8.0330e-02,  2.9913e-02, -3.2692e-02,
         1.5822e-02, -5.1534e-02, -3.1459e-02, -2.8410e-04, -6.9049e-02,
        -7.1741e-02,  4.2563e-02, -2.5306e-02,  1.2142e-01,  6.7891e-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([-0.1489,  0.2388, -0.0102,  0.0172, -0.0123,  0.0361,  0.0715,  0.0075,
        -0.0816,  0.0708, -0.0444,  0.1829,  0.0348, -0.0297,  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,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0578,  0.0366, -0.0021,  0.0048, -0.0222, -0.0053, -0.0150, -0.0393,
        -0.0233,  0.0053,  0.0245,  0.0224,  0.0950,  0.0091,  0.0044, -0.0790,
        -0.0367,  0.0104, -0.0001,  0.0160,  0.0218,  0.0044, -0.0130, -0.0029,
         0.0057,  0.0270, -0.0007,  0.0105,  0.0347, -0.0024, -0.0100, -0.0088,
        -0.0202,  0.0175, -0.0001,  0.0421, -0.0393,  0.0404,  0.0072,  0.0025,
        -0.0115,  0.0005,  0.0076,  0.0107,  0.0374,  0.0167,  0.0018, -0.0243,
         0.0055,  0.0241,  0.0331, -0.0061], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1345,  0.2948,  0.0104,  0.0020,  0.0171,  0.0447,  0.0640,  0.0184,
         0.0757, -0.0156,  0.0187, -0.0479, -0.0094, -0.0823,  0.0089,  0.0274,
         0.0380, -0.0083, -0.0678, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0364,  0.1565,  0.1719,  0.0045, -0.0095, -0.0205, -0.0377, -0.0181,
         0.0565,  0.0301, -0.0134, -0.0194, -0.0432,  0.0012,  0.0072,  0.0177,
        -0.0179, -0.0260, -0.0472, -0.0265,  0.1448, -0.0940,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1812,  0.0986,  0.0723,  0.0976,  0.0155, -0.0579, -0.0640, -0.0326,
         0.0426, -0.0069,  0.1117,  0.0204, -0.0084, -0.0133,  0.0538, -0.0065,
        -0.0889,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0977,  0.1792,  0.0038,  0.0725,  0.0193,  0.0277, -0.0266, -0.0251,
         0.1012,  0.0403, -0.0174,  0.0696,  0.0112,  0.0162,  0.0271,  0.0137,
         0.0227,  0.0506,  0.0030,  0.0439, -0.0030,  0.0117, -0.0043,  0.0230,
        -0.0180,  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,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1173,  0.0346,  0.0411, -0.0341,  0.0385,  0.0390, -0.0002, -0.0619,
         0.0901,  0.0491, -0.0787,  0.0301,  0.0068,  0.0661,  0.0692,  0.0012,
         0.0236, -0.0291, -0.0010,  0.0845,  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,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0559,  0.0081, -0.0056, -0.0038,  0.0238,  0.0163, -0.0034,  0.0742,
         0.0691,  0.0251, -0.0124, -0.0009, -0.0043,  0.0006,  0.0185, -0.0030,
         0.0556,  0.0012, -0.0038, -0.0205, -0.0054, -0.0158, -0.0086, -0.0208,
        -0.0070,  0.0165, -0.0005,  0.0018, -0.0332, -0.0022,  0.0321,  0.0131,
         0.0083,  0.0140,  0.0174,  0.0181,  0.0248,  0.0672,  0.0029, -0.0095,
        -0.0035,  0.0045,  0.0152, -0.0034, -0.0130, -0.0628, -0.1723,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1577,  0.3195, -0.0629,  0.0199, -0.0185, -0.0679,  0.0039, -0.2050,
        -0.0743, -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,  0.0000,  0.0000,  0.0000,  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([ 0.0465, -0.1082, -0.0046, -0.0295, -0.0221, -0.0192, -0.0471,  0.0020,
        -0.0134,  0.0093,  0.0072, -0.0107, -0.0010, -0.0081, -0.0108, -0.0102,
         0.0059,  0.0098,  0.0049, -0.0199,  0.0136,  0.0270, -0.0276, -0.0474,
        -0.0210, -0.0013, -0.0841, -0.0903,  0.0101, -0.0142,  0.0020,  0.0036,
         0.0097, -0.0179,  0.0066, -0.0194, -0.0031, -0.0104, -0.0034, -0.0136,
        -0.0124,  0.0154, -0.0339, -0.0246, -0.0086, -0.0665,  0.0220,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0854,  0.1499,  0.0413,  0.0396, -0.0768, -0.0147,  0.0478,  0.0578,
         0.0177,  0.0605,  0.0244, -0.0210, -0.0010,  0.0925, -0.0813, -0.0022,
         0.0162, -0.0137, -0.0413,  0.0059, -0.0361, -0.0089, -0.0229,  0.0137,
         0.0047,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0292,  0.0584,  0.0316, -0.0158, -0.0217, -0.0330, -0.0364,  0.0307,
         0.0190, -0.0123, -0.0371, -0.0068,  0.0973, -0.0542, -0.0508, -0.0270,
         0.0210,  0.0513,  0.0037,  0.0063, -0.0018,  0.0591, -0.0017, -0.0269,
         0.0415,  0.0007,  0.0201,  0.0111, -0.0075, -0.0006,  0.0220,  0.0406,
         0.0115,  0.0291,  0.0200, -0.0235,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0587,  0.1557, -0.0877,  0.0433,  0.1170,  0.0903, -0.0075,  0.0507,
        -0.0237,  0.1072,  0.0169, -0.0449,  0.0136, -0.1088,  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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  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, -0.0159,  0.0516,  0.0056, -0.0111, -0.0613, -0.0356, -0.0281,
        -0.1324, -0.0700, -0.0803, -0.1480, -0.0333, -0.0949, -0.0595,  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,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.9990e-02,  3.0379e-01,  3.5259e-02,  2.5639e-02, -2.4548e-02,
        -1.0780e-01,  8.9032e-03, -7.6517e-02, -1.4411e-02,  1.3717e-02,
         4.6268e-02, -2.3462e-02, -1.4466e-02,  1.8240e-02, -4.5922e-03,
        -3.4517e-03,  1.7095e-03,  1.5161e-03, -3.9721e-03, -7.0309e-03,
        -1.3004e-02, -3.7972e-05,  2.6862e-02, -2.7643e-04,  1.1485e-02,
        -8.9871e-03,  2.2730e-02,  1.5709e-02, -2.8236e-02, -1.0740e-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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0971,  0.1087,  0.0610,  0.0803,  0.0117, -0.0005,  0.0106, -0.0050,
         0.0085, -0.0179,  0.0328, -0.0684, -0.0473, -0.0625, -0.0091, -0.0598,
         0.0233, -0.0224,  0.0275, -0.0097, -0.0077, -0.0070, -0.0051, -0.0097,
         0.0054,  0.0092, -0.0019,  0.0112,  0.0196,  0.0202, -0.0023, -0.0367,
        -0.0502, -0.0076,  0.0045,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-7.0728e-02,  2.9808e-01,  3.2411e-02,  3.5425e-03, -7.6902e-03,
        -1.0771e-01,  1.2063e-02,  6.4329e-03,  3.5137e-03,  2.3299e-06,
         1.9661e-02,  6.3876e-03, -3.4517e-02,  2.1267e-03, -1.2583e-02,
        -1.0193e-03, -3.2935e-03, -6.3525e-03,  1.1510e-02, -7.4910e-03,
         3.0490e-02,  1.8114e-02,  1.6650e-02,  3.2594e-03,  1.4863e-02,
         6.6272e-03, -2.8622e-03, -4.6491e-03,  6.8853e-03,  4.5634e-03,
        -1.2638e-02,  4.0509e-03,  1.1720e-02, -7.7671e-02,  1.3784e-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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1029, -0.0785, -0.1900, -0.2259, -0.0922,  0.0005, -0.1360, -0.0468,
        -0.0135, -0.0548,  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([ 0.0503,  0.0364, -0.0679, -0.1286, -0.1005, -0.0463,  0.0114,  0.0211,
        -0.0289,  0.0755,  0.0350, -0.0125, -0.0023,  0.0007, -0.0110, -0.0368,
         0.0134, -0.0559,  0.0326,  0.0192, -0.0112,  0.0054,  0.0476, -0.0456,
        -0.1040,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  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,  0.0259,  0.0330,  0.0199, -0.0351, -0.0288,  0.0512,  0.0465,
        -0.0168, -0.1186, -0.0003,  0.0368, -0.0045, -0.0014,  0.0194,  0.0332,
        -0.0159, -0.0098,  0.0345, -0.1587,  0.1358,  0.0286,  0.0658, -0.0029,
         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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1482,  0.1007,  0.0538,  0.0187,  0.0063,  0.0048,  0.0103, -0.0115,
         0.0254,  0.0098,  0.0032, -0.0173, -0.0294, -0.0065,  0.0145,  0.0284,
         0.0252, -0.0021, -0.0157,  0.0062, -0.0184,  0.0025, -0.0180,  0.0014,
        -0.0027, -0.0045,  0.0120,  0.0112,  0.0124, -0.0332, -0.0080, -0.0142,
        -0.0270, -0.0041,  0.0064, -0.0158, -0.0062,  0.0018, -0.0137,  0.0031,
         0.0030,  0.0094, -0.0032,  0.0295, -0.0009,  0.0436,  0.0248, -0.0055,
        -0.0130, -0.0209, -0.0143, -0.0133,  0.0384, -0.0259], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.1189,  0.4277,  0.0291,  0.0445,  0.0158,  0.0173,  0.0320, -0.0300,
        -0.0931, -0.0135,  0.0737, -0.0345, -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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.1097e-01,  3.2346e-01,  1.8733e-02, -1.7390e-02, -3.2449e-02,
        -1.0721e-03,  4.8119e-03,  9.9394e-02,  1.4165e-04,  2.1392e-02,
        -7.7448e-03, -4.6373e-02, -1.1814e-02, -3.1087e-02, -1.0648e-01,
        -2.2192e-02, -3.1292e-02,  2.6841e-03, -2.6146e-02,  3.8622e-02,
        -4.5750e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2015,  0.0695, -0.1781, -0.0502,  0.0965,  0.0658,  0.1090,  0.0759,
         0.1535,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  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,  0.3008,  0.0368, -0.0292,  0.0382,  0.0018,  0.0097, -0.0474,
        -0.0065, -0.0026,  0.0158, -0.0185, -0.0284, -0.0421, -0.0080, -0.0257,
        -0.0220,  0.0049,  0.0378,  0.0402, -0.0006, -0.0435,  0.0037,  0.0082,
        -0.0051, -0.0021, -0.0718,  0.1343,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  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,  0.2777,  0.0605,  0.0390, -0.0844,  0.0141, -0.0148, -0.0406,
        -0.0380, -0.0349,  0.0620,  0.0237,  0.0049, -0.0322, -0.0380, -0.0086,
        -0.0132,  0.0372, -0.0050, -0.0059,  0.0687, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0327, -0.1290, -0.0744, -0.0221, -0.0770, -0.0208,  0.0222,  0.0042,
         0.0196,  0.0431, -0.0374, -0.0195, -0.0280, -0.0347, -0.0428,  0.0068,
         0.0094,  0.0053, -0.0259,  0.0342,  0.0689,  0.0060,  0.0110,  0.0259,
         0.0358,  0.0492,  0.0089,  0.0320, -0.0096,  0.0107,  0.0507,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.0441e-02, -3.0280e-01, -1.5846e-02, -3.4590e-02,  6.5965e-04,
        -4.2032e-02, -2.0398e-03, -9.3289e-04,  3.4634e-02,  1.2397e-02,
         1.6735e-02, -6.8337e-03,  1.2712e-02,  2.7575e-02, -1.1624e-02,
        -2.1977e-02, -2.1506e-02, -4.3885e-03, -1.4938e-02, -1.3999e-02,
        -2.3001e-02,  1.3177e-03, -6.9909e-03,  2.7697e-02,  3.6951e-05,
         7.3688e-03,  2.3174e-02,  8.2293e-03, -9.1927e-03,  1.9077e-02,
         5.5678e-02,  3.4267e-03, -2.7107e-03,  1.1733e-02, -8.8001e-03,
         2.6982e-02,  2.3912e-02, -8.7036e-03,  4.1631e-03,  3.0438e-02,
         2.6427e-02,  7.0574e-03, -1.9264e-02, -2.5962e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.4861e-01,  2.8347e-01, -4.7912e-02, -1.0395e-03, -3.5815e-02,
        -8.7213e-02, -1.0413e-04,  2.8066e-02,  6.7160e-02, -2.6574e-02,
        -3.0819e-02, -6.7048e-03,  2.2310e-03, -3.8528e-02,  3.3072e-02,
        -8.9819e-02,  7.2863e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1315,  0.3885,  0.0138,  0.0438, -0.0276,  0.0753,  0.0128, -0.0215,
        -0.0042, -0.0017,  0.0086, -0.0096, -0.0078,  0.0084, -0.0248, -0.0087,
         0.0392, -0.0219, -0.0229, -0.0140,  0.0101,  0.0232, -0.0144,  0.0248,
        -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1159, -0.2005, -0.1128,  0.0256, -0.0065, -0.0196,  0.0433,  0.0049,
         0.0603, -0.0054,  0.0251, -0.0236, -0.0160, -0.0086, -0.0354, -0.0078,
         0.0739,  0.0191,  0.0073, -0.0083,  0.1340,  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,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0133,  0.3030, -0.3557,  0.0298,  0.0315, -0.1165, -0.0633, -0.0769,
        -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1057,  0.2076,  0.0508, -0.0200,  0.0210,  0.0118,  0.0496, -0.0123,
        -0.0266, -0.0065, -0.0137,  0.0142, -0.0339,  0.0073, -0.0151, -0.0329,
        -0.0088, -0.0222, -0.0052, -0.0048, -0.0128, -0.0079, -0.0053,  0.0016,
        -0.0030,  0.0647, -0.0119, -0.0057, -0.0504, -0.0323,  0.0189, -0.0530,
         0.0089,  0.0537,  0.0000,  0.0000,  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([ 7.6639e-02,  3.4477e-01,  2.0379e-02, -3.9239e-02,  1.0589e-02,
         1.3880e-02, -4.1449e-02, -1.0533e-02,  2.2129e-02,  2.5386e-02,
        -9.7332e-03, -8.9171e-03, -1.3400e-02, -4.3493e-02, -2.5036e-04,
         1.3373e-02,  3.9719e-02,  1.6541e-02, -5.3437e-03,  1.2875e-02,
         6.1609e-03, -2.3637e-02, -2.3714e-02, -6.7074e-03,  5.9708e-03,
         1.4593e-03, -1.0696e-02, -5.6596e-03, -6.1648e-03,  1.4329e-02,
        -4.9430e-03, -9.5385e-04,  1.0955e-02, -1.1431e-02, -2.6799e-03,
        -1.5046e-03, -7.8287e-03, -6.5420e-03,  9.6925e-03,  6.0113e-03,
        -9.7261e-03, -6.8953e-03, -3.3034e-02, -1.4667e-02,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2330, -0.2392, -0.0697,  0.0427,  0.0219,  0.0121,  0.0420, -0.0013,
        -0.0238,  0.0218, -0.0234,  0.0107, -0.0295, -0.0051, -0.0170,  0.0106,
         0.0154, -0.0128,  0.0179,  0.0331,  0.0401, -0.0116, -0.0428,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0521,  0.4819, -0.1314,  0.0459, -0.0502, -0.0267,  0.0284, -0.0104,
         0.0581,  0.0164,  0.0561, -0.0134,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1215,  0.4228,  0.0290, -0.0497,  0.0352,  0.0104, -0.0519,  0.0043,
        -0.0210, -0.0246, -0.0744,  0.0195, -0.0118, -0.0277,  0.0168, -0.0193,
         0.0053,  0.0138,  0.0083,  0.0206, -0.0110, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0007, -0.5684, -0.1164,  0.0899, -0.0368,  0.0335, -0.0082, -0.0506,
        -0.0372, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.0313e-02, -3.8531e-01,  5.6365e-02,  3.1671e-02, -1.6152e-02,
        -8.9763e-03,  5.2813e-04, -1.2235e-02, -5.4585e-05, -3.5817e-02,
         2.3020e-02,  1.2602e-02, -2.2825e-02, -2.5481e-02,  1.9771e-02,
        -5.4647e-03, -3.5704e-02, -1.2549e-02, -6.7192e-03,  1.0736e-03,
        -1.9535e-02, -1.2773e-01, -1.3011e-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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0404,  0.2522, -0.0535, -0.0657,  0.0131,  0.0013, -0.0282, -0.0286,
        -0.0044, -0.0082, -0.0229, -0.0050, -0.0066, -0.0131,  0.0339,  0.0003,
         0.0065,  0.0007, -0.0414,  0.0317, -0.0136, -0.0176, -0.0106,  0.0019,
        -0.0045, -0.0192,  0.0011,  0.0084, -0.0126, -0.0164, -0.0210, -0.0009,
        -0.0248, -0.0048,  0.0146, -0.0022, -0.0052, -0.0014,  0.0126,  0.0202,
         0.0100,  0.0148,  0.0206, -0.0383,  0.0450], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.2631e-02,  3.5883e-01,  1.8087e-02, -1.0289e-01, -1.5422e-02,
         1.1748e-02,  6.1030e-02, -1.7224e-02, -1.8828e-02,  1.8148e-02,
         7.7367e-02,  8.8566e-03, -1.8951e-02, -1.3772e-02, -7.4370e-03,
        -9.3878e-05, -8.6639e-02, -1.9621e-02, -2.1285e-02, -4.1141e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.4477e-02,  4.1309e-01, -1.6424e-01, -4.0694e-02,  1.6212e-02,
        -2.8956e-02, -8.0787e-03,  7.8702e-03,  8.1511e-03,  4.6090e-05,
         1.7922e-02, -1.4705e-02,  4.1504e-02,  2.4712e-02,  4.1765e-02,
        -4.7752e-03, -2.5814e-02,  2.1203e-02,  1.7529e-02, -4.7989e-03,
         1.1225e-02, -3.7960e-03, -6.2955e-03,  1.2929e-02,  5.3702e-03,
        -3.3847e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.4080e-01,  3.7261e-01,  2.0043e-02, -3.5864e-02,  5.5032e-04,
        -3.7976e-02,  1.0508e-02, -8.6466e-03, -2.0816e-02,  1.1092e-02,
         2.8375e-03, -1.2310e-02,  3.6094e-02, -5.1579e-02, -2.5226e-02,
         4.4868e-02,  1.8504e-02,  1.4622e-02,  3.8177e-03,  2.1827e-03,
        -1.3327e-05,  5.4890e-03,  2.7634e-02,  1.0607e-02,  2.5404e-03,
         8.5365e-03, -5.6539e-02, -1.7686e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1498, -0.1781,  0.0703, -0.1550, -0.0171, -0.1040, -0.1509, -0.1046,
        -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1523, -0.3574,  0.0586,  0.0682, -0.0158,  0.0181,  0.0143,  0.0282,
         0.0286,  0.0103,  0.0010,  0.0232, -0.0023,  0.0041, -0.0321,  0.0257,
        -0.0085, -0.0069, -0.0075,  0.0020,  0.0106, -0.0157,  0.0064, -0.0033,
        -0.0227,  0.0454,  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 #100: [tensor([ 4.4395e-02,  3.9992e-01,  1.4571e-02,  4.8241e-02, -7.2883e-03,
        -1.4332e-02,  1.4460e-02, -3.0278e-02, -3.6964e-03, -7.5761e-05,
        -1.1846e-02, -6.1098e-02, -1.3857e-02,  3.4796e-02,  2.4974e-02,
         5.6373e-03, -7.6642e-03,  5.8950e-03, -1.0303e-02,  3.7890e-03,
         1.5958e-02, -1.8733e-02,  3.0566e-02,  4.0221e-03, -1.8889e-02,
         2.8158e-02,  1.7889e-02,  6.4188e-03, -4.7868e-03, -1.2250e-02,
         1.5557e-02,  9.6519e-03,  5.3893e-04,  4.3522e-03,  1.1670e-03,
        -4.6377e-03, -2.2019e-02,  2.7287e-02,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.0601e-01,  3.3043e-01,  1.6642e-01, -1.1350e-02,  8.4999e-03,
         1.5021e-02, -1.3210e-02,  3.7293e-03, -9.0544e-03, -7.7628e-04,
        -1.5384e-05, -3.3576e-02, -6.7692e-03, -1.3883e-02, -1.8939e-02,
         1.6817e-04, -3.1259e-03, -1.8469e-03,  2.5831e-03, -9.8555e-03,
        -4.5474e-03, -7.2149e-03, -1.4262e-03,  1.1558e-02, -4.6873e-04,
         3.8497e-03,  7.4815e-05, -7.3249e-04,  2.1319e-03,  5.5384e-04,
        -2.3478e-02,  2.9070e-03,  8.0081e-03, -1.4972e-02,  3.9144e-03,
         7.4024e-03,  2.3450e-02,  2.8035e-02,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0136, -0.4542,  0.0568,  0.0677,  0.0183, -0.0291, -0.0180,  0.0528,
        -0.0088,  0.0527,  0.0160, -0.0024, -0.0120,  0.0028,  0.0197, -0.0114,
         0.0234,  0.0243, -0.0059, -0.0208, -0.0008, -0.0010,  0.0167, -0.0103,
         0.0157, -0.0012, -0.0023,  0.0090,  0.0125,  0.0201,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.7169e-02, -3.3607e-01, -4.0915e-02, -2.8308e-02, -1.3803e-02,
         7.1911e-03,  5.2110e-02, -5.4284e-04,  3.7831e-03,  3.5843e-02,
         5.4161e-03, -6.3477e-05, -6.9076e-03, -7.0708e-03, -4.6714e-03,
         2.1900e-03, -1.4429e-02, -2.0202e-03,  1.6818e-02, -5.7431e-02,
        -1.2068e-03, -2.2013e-03,  8.0195e-03,  2.2381e-02,  7.1256e-03,
         7.0774e-03,  2.9558e-02,  6.9346e-03, -3.5026e-03, -2.0814e-02,
        -6.9541e-02, -1.7894e-02,  1.8685e-03, -1.1653e-02,  3.1289e-03,
         4.0951e-03,  3.5237e-03,  2.2761e-03, -7.5841e-02, -1.8606e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0081,  0.2985, -0.1220, -0.0657,  0.0433,  0.0337,  0.0083,  0.0100,
        -0.0014,  0.0129, -0.0122, -0.0031,  0.0141, -0.0138, -0.0082, -0.0153,
        -0.0032,  0.0064,  0.0144, -0.0134, -0.0288,  0.0086, -0.0122, -0.0011,
        -0.0013,  0.0088,  0.0110,  0.0094,  0.0052, -0.0171,  0.0136,  0.0109,
        -0.0033,  0.0151, -0.0255,  0.0255, -0.0054,  0.0891,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0367, -0.2188, -0.2003, -0.0095, -0.0341,  0.0128,  0.0443, -0.0078,
         0.0444, -0.0150, -0.0205, -0.0568, -0.0265,  0.0010, -0.0011, -0.0783,
        -0.0321, -0.0101, -0.0272, -0.0066, -0.0225,  0.0341,  0.0575, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1083, -0.3730,  0.0844,  0.0192, -0.0322, -0.0554, -0.0088, -0.0136,
         0.0045, -0.0066, -0.0235, -0.0079, -0.0005, -0.0413,  0.0041, -0.0126,
         0.0025, -0.0049,  0.0050,  0.0027,  0.0073, -0.0198,  0.0088, -0.0152,
        -0.0066,  0.0055,  0.0124, -0.0202, -0.0066, -0.0097, -0.0314, -0.0455,
         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,  0.4107, -0.0004, -0.0494,  0.0320, -0.0662,  0.0399, -0.1177,
         0.0807,  0.0191,  0.0286,  0.0472,  0.0065,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.5962e-02, -4.0822e-01, -8.4071e-02,  5.3736e-02,  2.5257e-02,
         3.3792e-04,  9.9463e-03,  5.6521e-02,  2.9206e-03,  2.8283e-02,
         2.2120e-02, -8.6251e-03,  1.9683e-02,  4.3047e-02,  1.7437e-02,
         1.1861e-02, -1.8967e-02,  2.7292e-03,  2.3864e-02,  3.6192e-03,
        -3.2718e-03, -2.4169e-02, -3.7001e-03, -5.1650e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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, -0.4280,  0.0190,  0.0079,  0.0359,  0.1142, -0.0067,  0.0009,
         0.0025,  0.0058, -0.0194, -0.0282,  0.0098,  0.0046,  0.0083, -0.0275,
         0.0193,  0.0008,  0.0266,  0.1127, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0367,  0.3916,  0.0646, -0.0466,  0.0745, -0.0238,  0.0290, -0.0006,
         0.0142,  0.0067, -0.0066,  0.0124,  0.0293,  0.0347, -0.0079,  0.0893,
        -0.0047,  0.0269,  0.0061,  0.0204, -0.0008,  0.0229,  0.0031,  0.0013,
        -0.0035,  0.0314, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0491,  0.3996,  0.0527,  0.0617,  0.0421, -0.0676,  0.0089,  0.0093,
        -0.0032,  0.0475,  0.0463,  0.0301, -0.0069, -0.0468, -0.0048,  0.0636,
        -0.0523,  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],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([-1.2695e-02,  3.2916e-01,  1.3410e-03,  5.6247e-02,  3.4590e-03,
        -2.6043e-02,  4.2146e-03, -3.4474e-04, -6.0131e-03, -3.3449e-03,
        -4.6523e-03,  1.3910e-02,  1.3655e-02, -4.9802e-03, -8.8409e-03,
        -6.5577e-02, -2.5006e-02, -3.0919e-02,  3.1784e-03, -7.8012e-03,
        -2.1530e-03, -4.1168e-03, -4.8230e-03,  7.6506e-04, -3.2775e-03,
        -7.9958e-03,  8.9640e-03, -7.2039e-03,  4.6128e-03,  4.6404e-03,
         3.0475e-03, -5.8235e-02, -8.2950e-03,  5.6171e-03,  1.8264e-04,
        -1.5504e-02, -3.1558e-03,  7.7419e-03,  1.7059e-02,  5.5783e-02,
        -3.1260e-03, -1.7678e-02, -3.6406e-03, -1.6522e-02, -6.4671e-02,
         4.9807e-02,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1464,  0.1815,  0.0719, -0.0249,  0.0150,  0.0129,  0.0091,  0.0105,
        -0.0006,  0.0052,  0.0161,  0.0103, -0.0049, -0.0025, -0.0028,  0.0138,
         0.0035, -0.0078,  0.0257,  0.0367,  0.0179, -0.0155,  0.0248, -0.0129,
        -0.0031,  0.0084,  0.0244,  0.0005, -0.0179, -0.0054,  0.0004, -0.0225,
         0.0052,  0.0014,  0.0072, -0.0095,  0.0005, -0.0115,  0.0037, -0.0046,
        -0.0202,  0.0123, -0.0173, -0.0011,  0.0053, -0.0177,  0.0410,  0.0857],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1409, -0.1801,  0.0979,  0.0047,  0.0726, -0.0549,  0.0225,  0.0112,
        -0.0170,  0.0060, -0.0162, -0.0351, -0.0143,  0.0024,  0.0003, -0.0256,
        -0.0256,  0.0043, -0.0029,  0.0120, -0.0124,  0.0097, -0.0148, -0.0112,
         0.0136,  0.0087,  0.0019, -0.0027,  0.0419, -0.0015,  0.0107, -0.0012,
        -0.0298, -0.0932,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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,  0.6816, -0.1483,  0.0962, -0.0138, -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,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.8611e-02, -3.1625e-01,  1.4358e-01, -6.4886e-03,  1.6346e-02,
        -2.2480e-02, -2.6582e-02, -8.5330e-03, -2.4877e-02, -2.0943e-02,
         4.6292e-03,  5.0337e-03,  1.6386e-03,  3.3459e-02, -6.7341e-03,
        -2.1037e-04,  1.5508e-02,  1.5213e-02, -9.2565e-03,  6.8059e-03,
         4.8010e-02, -6.1703e-03,  2.9626e-03, -8.9527e-03, -3.2034e-02,
        -1.2041e-02,  2.0033e-02, -1.4671e-02, -5.2003e-03, -8.1408e-03,
        -5.1678e-03,  2.8710e-03, -1.3793e-02,  7.7260e-03,  1.0254e-02,
        -9.6124e-03, -1.4182e-02, -5.8857e-03,  1.9134e-04, -2.9601e-03,
         1.2926e-02,  2.8494e-03, -5.5409e-03,  2.0976e-02,  1.2669e-02,
        -1.1008e-02,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0423, -0.3420, -0.1298,  0.0266, -0.0156, -0.0401, -0.0111,  0.0265,
        -0.0157,  0.0539,  0.0148,  0.0069,  0.0032,  0.0365, -0.0335,  0.0354,
         0.0779,  0.0059, -0.0021, -0.0071,  0.0136, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0385,  0.4015,  0.1235, -0.1842,  0.0761, -0.0244, -0.0298,  0.0312,
        -0.0053, -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,  0.0000,
         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,  0.4550,  0.0167,  0.0020, -0.0783,  0.0204, -0.0483,  0.0184,
         0.0523, -0.0066, -0.0227, -0.0037, -0.0573,  0.0261, -0.0272, -0.0294,
        -0.1146,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1597,  0.3184,  0.0118, -0.0682,  0.1157, -0.0023,  0.0459,  0.0182,
         0.0207, -0.0399, -0.0229, -0.0056, -0.0054,  0.0335,  0.0213,  0.0752,
        -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2144,  0.2456,  0.0347,  0.0019,  0.0898, -0.0408, -0.0323,  0.0077,
         0.0062, -0.0689, -0.0483, -0.0732, -0.0857, -0.0068,  0.0212,  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([-0.2190, -0.4303,  0.0096, -0.0359, -0.0137, -0.0258, -0.0006,  0.0408,
        -0.0425, -0.0316, -0.0278,  0.0109, -0.0073,  0.0051,  0.0333, -0.0505,
        -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0641, -0.4384, -0.0859,  0.0247,  0.0586, -0.0278,  0.0501, -0.0431,
         0.0097,  0.0143, -0.0007, -0.0060,  0.0530,  0.0875,  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],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([-8.2341e-02, -5.2337e-01,  2.1289e-02, -1.0317e-03,  4.4306e-02,
         1.7069e-02, -9.0538e-03,  1.2988e-01,  2.8483e-04,  1.1096e-02,
         3.0853e-02,  4.4811e-03,  3.8006e-02,  1.8393e-02,  6.8541e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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, -0.3823,  0.0544,  0.0159, -0.0817,  0.0313,  0.0637,  0.0357,
         0.0046, -0.0517,  0.0426,  0.0552,  0.0109, -0.0701, -0.0216, -0.0042,
         0.0060,  0.0593, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.5820e-02, -4.6641e-01, -5.6804e-02,  1.4851e-02,  1.1430e-02,
         2.2961e-02,  7.9047e-03, -3.4486e-02,  3.3862e-02, -6.9620e-02,
         2.8106e-02,  7.8240e-02,  4.5882e-02,  1.0599e-02, -2.9445e-02,
         3.5363e-03, -4.6214e-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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.7240e-03, -3.7432e-01,  2.5361e-03, -6.6779e-02, -1.1995e-02,
        -2.3900e-02,  2.3602e-02, -9.0125e-04,  4.7196e-03, -2.9144e-03,
         3.2609e-02,  1.8358e-02,  1.8863e-02,  3.6234e-03,  1.6157e-02,
        -3.6964e-02, -3.6427e-03,  1.7133e-02, -3.0442e-02,  1.1230e-02,
         6.0174e-03, -2.5744e-02,  7.1612e-03,  1.5027e-03,  4.0977e-02,
        -2.2996e-03,  7.7633e-03,  7.1485e-03, -1.6476e-02,  2.2921e-03,
        -8.5565e-03,  4.3078e-03, -4.9211e-03, -5.5849e-03, -1.0514e-02,
         7.1462e-03, -1.6900e-03,  7.0272e-03,  2.3087e-03,  6.9806e-03,
         1.7386e-03,  3.4595e-04,  7.5576e-03, -7.4294e-03, -1.2371e-02,
        -1.2982e-02,  1.9361e-02,  1.9779e-02,  1.0890e-02, -2.0712e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0693,  0.4315,  0.0842,  0.0774, -0.0146, -0.0043, -0.0716, -0.0436,
        -0.0102, -0.0293, -0.0235,  0.0064, -0.0132, -0.0492, -0.0630,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0119, -0.6055, -0.0251, -0.0071, -0.0231, -0.0208, -0.0100, -0.0030,
        -0.0147, -0.0255,  0.0274,  0.0081, -0.0088, -0.0110, -0.0256,  0.0402,
        -0.0731, -0.0138,  0.0241, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.9269e-02, -2.0749e-01,  5.8863e-02,  7.0370e-02, -6.5309e-02,
         5.5643e-02,  5.0878e-02,  1.6088e-02, -8.1547e-03, -1.1011e-02,
        -2.6737e-02,  4.7715e-02, -1.1944e-03, -6.2729e-05,  6.2192e-03,
        -1.4823e-02, -2.9152e-02,  8.1595e-04, -4.6341e-03,  2.0482e-02,
         6.9629e-02,  3.9894e-03, -1.5935e-02, -1.0812e-02, -2.6367e-02,
        -7.2729e-02,  2.5623e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1405, -0.2501, -0.1159, -0.0583, -0.0440, -0.0599,  0.0210, -0.0059,
         0.0365, -0.0191,  0.0339,  0.1881, -0.0149, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.8963e-02, -4.5818e-01,  1.6472e-02,  4.2552e-03, -9.9981e-03,
        -1.0182e-02,  2.8107e-04, -2.6843e-04,  3.1042e-02,  1.6881e-02,
         2.4607e-02, -1.8362e-02,  2.0929e-01, -2.4333e-02,  8.5841e-02,
        -2.1042e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.7683e-02,  5.6861e-01, -4.2212e-02,  4.7369e-02,  1.5528e-02,
         2.2610e-02,  1.9300e-02,  1.9669e-02, -5.5222e-03,  8.8950e-03,
         2.3307e-03,  1.4580e-02, -1.4596e-02,  2.9154e-03,  1.3034e-02,
         2.2320e-02,  9.1221e-03, -1.2067e-02, -1.6959e-02, -2.6915e-03,
         2.8508e-04, -1.3333e-02, -6.4389e-03, -1.9347e-02, -1.8646e-02,
        -1.4465e-02, -4.9473e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.6965e-02,  4.2448e-01,  1.7860e-02, -2.1240e-02,  3.6965e-02,
         2.4766e-03, -7.2223e-03,  1.4320e-02,  1.5933e-02,  1.7997e-02,
         3.4060e-04, -2.3239e-02, -1.7030e-02,  1.0185e-02, -2.4314e-02,
        -3.2396e-02, -2.7188e-02,  5.2831e-04, -2.4123e-02,  3.5071e-02,
        -2.1150e-02,  2.7047e-04, -5.6639e-03, -2.5373e-03, -8.9331e-02,
         6.1176e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.6263e-02, -4.7338e-01,  6.9526e-03, -4.8162e-03, -4.7380e-02,
         1.0880e-02, -1.1001e-02, -9.6336e-03,  3.0030e-02, -1.0793e-02,
         3.8042e-04,  1.5914e-03, -1.9560e-02,  6.4657e-03, -2.0676e-02,
        -3.3145e-02,  4.6616e-03, -2.7423e-03, -6.6352e-03, -3.6489e-02,
         3.1631e-02,  1.8660e-02,  2.4564e-02, -1.1737e-02, -4.8538e-03,
        -3.7806e-03, -5.5759e-03, -1.1139e-02,  1.5913e-02, -7.3987e-02,
        -4.6821e-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>)]
Iter #250: [tensor([ 0.1050,  0.4864, -0.0029, -0.1585, -0.0160,  0.0474,  0.0024, -0.0096,
        -0.0367,  0.0052,  0.0258,  0.0150,  0.0061, -0.0083,  0.0076, -0.0067,
         0.0045, -0.0407, -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([ 0.2373, -0.3780, -0.0344, -0.0554,  0.0503, -0.0607,  0.0926, -0.0373,
        -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([ 2.1627e-01,  3.9894e-01,  1.3746e-02, -3.0781e-02, -2.8596e-02,
         1.1432e-03, -2.4908e-02, -5.8539e-02, -2.3979e-02, -1.1191e-02,
        -1.8169e-02,  3.6132e-02,  1.8953e-04,  4.4570e-03,  9.6109e-03,
         2.7226e-03,  4.2069e-03,  1.0162e-02, -3.7955e-03,  3.0435e-02,
         7.2027e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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,  0.5268, -0.0055,  0.0149, -0.0120,  0.0327, -0.0015,  0.0181,
        -0.0366, -0.0112, -0.0134, -0.0242, -0.0044, -0.0027,  0.0060, -0.0409,
        -0.0046,  0.0044, -0.0016,  0.0048,  0.0157, -0.0338, -0.0054, -0.0125,
        -0.0330, -0.0177,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0021,  0.2759,  0.0472,  0.0091, -0.0207, -0.0482,  0.0212, -0.0468,
        -0.0077,  0.0109,  0.0350,  0.0226,  0.0271, -0.0099, -0.0062,  0.0238,
         0.0063,  0.0234,  0.0230, -0.0012, -0.0298,  0.0106,  0.0166,  0.0115,
         0.0030,  0.0136, -0.0143,  0.0185, -0.0053,  0.0205, -0.0181, -0.0105,
         0.0168, -0.0086, -0.0296, -0.0038,  0.0066, -0.0399,  0.0539,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0179,  0.3656,  0.0408, -0.0087, -0.0377, -0.0179, -0.0140,  0.0148,
         0.0876,  0.0980,  0.0035,  0.1085, -0.0483,  0.0130,  0.0039,  0.0162,
        -0.0048,  0.0123,  0.0439,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1197,  0.3258,  0.0280,  0.0022,  0.0079,  0.0416,  0.0174, -0.0344,
        -0.0168, -0.0217, -0.0005, -0.0300, -0.0227, -0.0162, -0.0345, -0.0236,
        -0.0108, -0.0115, -0.0085, -0.0105,  0.0309, -0.0957,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.2673e-02, -4.4005e-01, -3.4587e-03, -1.6446e-04, -1.5855e-02,
        -5.2823e-02, -1.0527e-02,  2.0188e-02, -3.7239e-04, -9.1004e-03,
         1.1936e-04,  6.7008e-03,  1.4796e-02, -1.3967e-02, -3.8012e-03,
        -1.2643e-02, -2.0770e-02,  4.5624e-03,  1.7312e-04,  2.1983e-02,
         2.0506e-02, -1.6199e-02, -3.2439e-03, -1.0707e-02, -3.7172e-02,
        -6.7599e-03, -9.8616e-03, -1.6269e-02, -2.2977e-02,  3.9988e-02,
        -5.6675e-03,  4.6928e-05, -2.6770e-02, -3.2661e-02,  2.7364e-02,
         2.0339e-02,  2.7038e-03, -3.9438e-04, -8.2130e-03,  1.7433e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.8855e-02, -4.9126e-01, -6.0766e-02, -2.7309e-02, -5.3559e-03,
         3.5198e-02,  1.1000e-02, -2.2712e-02,  1.1956e-02,  9.1866e-03,
         1.3044e-03,  7.2198e-03, -2.8648e-03,  1.1991e-02,  3.0103e-03,
         7.1061e-03,  9.0187e-03,  6.4472e-03, -9.1645e-03, -1.2703e-02,
         2.4080e-02,  1.7275e-03,  1.6899e-02,  8.8713e-04, -5.4553e-03,
        -3.5697e-03, -1.1222e-02,  2.1392e-03, -1.1753e-04, -1.1752e-02,
         4.8416e-03, -1.5986e-02, -7.3437e-03,  1.8335e-02, -4.7065e-03,
        -5.6512e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2459,  0.2536,  0.0174, -0.0098, -0.0051,  0.0151,  0.0104,  0.0188,
         0.0123,  0.0144,  0.0029,  0.0032,  0.0017,  0.0037, -0.0328, -0.0059,
         0.0037, -0.0041,  0.0164,  0.0179, -0.0151,  0.0094, -0.0096, -0.0126,
        -0.0073, -0.0367, -0.0504,  0.0138,  0.0166, -0.0027,  0.0390,  0.0917,
         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, -0.2911,  0.0109, -0.0298,  0.1000,  0.0112,  0.0005, -0.0081,
        -0.1416,  0.2740,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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,  0.3807, -0.0059,  0.0343,  0.0332,  0.0473,  0.0051, -0.0066,
         0.0684, -0.0712, -0.0286,  0.1403, -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>)]
Iter #300: [tensor([ 0.1167,  0.3544, -0.0313, -0.0089,  0.0396, -0.0273,  0.0227, -0.0352,
        -0.0913, -0.0266,  0.0217, -0.0088, -0.1016, -0.0787, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0510,  0.3668,  0.0736,  0.0307,  0.0221,  0.0318, -0.0022,  0.0172,
         0.0085, -0.0189, -0.0213, -0.0118,  0.0011, -0.0075, -0.0152,  0.0006,
        -0.0069, -0.0147,  0.0184, -0.0136,  0.0236, -0.0157, -0.0121, -0.0071,
        -0.0181, -0.0121, -0.0051, -0.0247, -0.0041, -0.0103, -0.0194, -0.0304,
         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,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1419, -0.6770, -0.0640,  0.0237,  0.0492, -0.0156, -0.0171,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.5389e-02, -5.5026e-01,  1.0902e-01,  3.7086e-02,  1.9834e-02,
         1.9484e-02,  1.1397e-02,  6.1269e-03, -2.3852e-02, -1.1903e-02,
        -4.1495e-03,  7.8402e-03,  8.6826e-03, -4.5268e-03, -2.4851e-03,
         5.4582e-02, -6.8163e-03, -8.5110e-03,  8.0458e-03, -1.0436e-02,
        -2.3060e-03, -4.6877e-03,  4.9369e-03, -2.0252e-03,  9.9461e-04,
         8.8652e-03, -7.3066e-03,  1.0161e-02,  2.7754e-02,  5.3080e-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([ 0.0626, -0.3785,  0.1428, -0.0776, -0.0753,  0.0108, -0.0070,  0.0180,
         0.0253,  0.0338, -0.0221, -0.0102, -0.0008, -0.0058, -0.0145,  0.0011,
        -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0093, -0.4708, -0.1016,  0.0585, -0.0341, -0.0018, -0.0373,  0.0308,
         0.0178, -0.0043, -0.0125, -0.0109,  0.0212,  0.0007,  0.0225,  0.0130,
         0.0061,  0.0054,  0.0060,  0.0118,  0.0132, -0.0060, -0.0348,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.4759e-02,  3.2800e-01,  3.0876e-02, -1.8988e-02,  7.5247e-02,
        -5.0208e-03,  7.9215e-02, -4.5328e-02, -5.4335e-02, -3.2079e-02,
        -7.2727e-04, -7.1587e-04, -4.9214e-03, -1.3998e-02,  6.9010e-04,
        -7.4585e-03, -3.5332e-02, -8.2433e-03,  1.4870e-03,  6.9514e-03,
         1.1539e-03, -1.9764e-03, -9.4249e-04,  1.1204e-02,  7.2208e-03,
         9.1353e-04, -4.6352e-03,  1.1758e-02,  8.3402e-03,  4.9533e-03,
        -8.9077e-03,  4.3978e-03, -4.5787e-03, -6.3034e-03,  1.6705e-04,
         1.3278e-02, -1.4957e-03,  7.3048e-02, -2.1166e-03, -2.4515e-02,
         2.2643e-03, -1.1454e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.1445e-02,  3.8580e-01,  6.8760e-03, -4.4310e-03,  2.4592e-03,
        -4.3977e-02,  1.4152e-02, -1.7349e-02, -3.0657e-02,  2.7775e-02,
         2.0053e-02,  2.3135e-02, -6.9102e-03, -6.0062e-03, -6.8281e-03,
         3.5183e-04, -1.2317e-03, -1.7174e-03, -1.3996e-03,  1.3859e-02,
        -6.0009e-04, -1.2639e-02,  7.9476e-03,  2.5007e-02, -6.0024e-03,
         1.0165e-02,  9.5143e-03,  1.2915e-02,  1.3747e-02, -3.2427e-03,
         1.8823e-02, -2.8471e-02, -2.9685e-02, -4.8480e-03,  3.6714e-02,
        -3.4629e-02,  3.4022e-02, -5.4617e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1243,  0.5120,  0.0239,  0.0859, -0.0299,  0.0097, -0.0190, -0.0313,
        -0.0796, -0.0798, -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,  0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.6116e-02,  4.4476e-01, -3.9481e-02,  8.4319e-03, -3.4237e-02,
        -3.5735e-02,  2.1482e-03, -4.6438e-03, -5.9751e-02, -7.8014e-03,
        -1.4396e-02, -3.2679e-03, -1.4985e-02, -1.5625e-02, -6.4799e-03,
         1.3185e-03, -2.0818e-04,  1.1761e-02,  1.8237e-03, -8.5797e-03,
         1.9590e-02,  1.8165e-02, -9.1212e-03,  5.8525e-03, -3.7895e-02,
        -2.5602e-03, -1.9180e-02, -1.5393e-02, -1.1349e-02, -9.0182e-03,
        -4.3503e-02,  4.0556e-03,  5.2763e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.8894e-02, -3.1116e-01, -3.6614e-03, -2.0779e-02, -3.4318e-02,
         1.7064e-02, -5.2392e-03, -9.5166e-03, -2.8963e-02,  2.6032e-02,
         2.2166e-02,  1.0617e-02,  8.0863e-03,  1.0811e-02,  2.7554e-02,
        -4.5206e-03,  4.8045e-03,  9.6396e-03, -1.7260e-02,  6.1287e-03,
        -5.3638e-03, -1.0772e-02,  4.0076e-03,  6.8595e-03, -6.5317e-03,
        -3.5639e-03, -7.2227e-04,  3.6184e-03,  5.0532e-03, -8.2318e-03,
        -2.8144e-03, -1.6636e-02, -2.5126e-02, -3.1006e-02, -1.7833e-02,
        -5.7480e-03, -4.6802e-03, -3.1406e-03,  1.0861e-03, -2.1338e-02,
        -1.6132e-02, -1.2543e-03,  2.2511e-03,  1.4510e-04,  1.5042e-02,
        -1.0915e-03,  1.0219e-02, -8.2964e-03,  5.8729e-03,  2.6396e-02,
         1.0196e-01], device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.2580e-03, -4.2972e-01,  1.0495e-01, -2.4049e-03, -1.1408e-02,
         4.2075e-02,  3.6257e-03,  1.9411e-02, -1.5258e-02,  5.4954e-03,
        -1.9636e-02,  5.0993e-03, -1.0340e-02, -1.8697e-03,  5.5058e-03,
         1.5605e-02,  9.3663e-03,  2.9622e-04,  3.0992e-03, -5.6700e-03,
         1.7669e-02, -2.3763e-03, -3.3153e-02, -7.5913e-03, -2.2882e-02,
         1.4695e-03,  7.1537e-06, -1.4783e-02,  1.3082e-02, -2.2657e-03,
         2.6652e-02, -3.9523e-03, -9.9593e-05,  4.3360e-02,  1.4148e-02,
        -5.5116e-03, -7.5901e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         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([ 0.1651,  0.3521,  0.0334, -0.0422, -0.0393, -0.0125, -0.0115, -0.0138,
         0.0018, -0.0077, -0.0295, -0.0094, -0.0138,  0.0088,  0.0088, -0.0011,
        -0.0098, -0.0134, -0.0274, -0.0155,  0.0071, -0.0218, -0.0102, -0.0155,
         0.0022,  0.0197,  0.0178, -0.0032, -0.0071,  0.0063,  0.0125,  0.0011,
        -0.0112, -0.0027,  0.0256,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.3462e-02, -2.2311e-01, -2.2887e-01, -5.3353e-02,  4.2723e-02,
         4.0865e-03,  4.3395e-03, -9.9479e-03, -9.0946e-03,  5.7458e-03,
        -4.2607e-03,  1.3131e-02, -9.8365e-05,  4.3864e-03, -1.5948e-02,
         4.6421e-03, -3.1514e-03,  2.9615e-03, -2.8933e-03,  5.6705e-03,
        -8.5481e-03,  5.7702e-02,  2.4690e-04, -1.4070e-03,  3.4161e-03,
        -4.5921e-03,  1.2459e-03, -1.1151e-03, -2.4210e-02,  7.6231e-04,
        -1.1673e-02, -4.2561e-03, -1.0300e-03, -5.6585e-03, -6.4449e-03,
        -2.3415e-04, -4.3356e-03, -5.7870e-03, -4.6476e-03,  4.7596e-03,
        -5.3258e-03, -3.3451e-03,  1.2943e-04,  1.4807e-03, -3.9889e-03,
        -4.5296e-03, -2.0168e-03,  8.3903e-03,  5.1057e-03,  3.4602e-03,
        -2.0050e-03, -3.0823e-03, -2.4087e-02,  4.9369e-02, -1.9733e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.3965e-01, -3.8635e-01,  2.9215e-03, -1.8311e-02, -1.8010e-05,
        -3.3029e-03, -1.2277e-02, -1.5982e-02,  9.8061e-03, -1.6440e-02,
        -3.3024e-02, -2.5711e-02,  2.7686e-02,  6.7980e-03, -7.9274e-03,
         4.5906e-02, -1.3781e-02, -1.5702e-03, -1.5875e-02,  1.7666e-02,
         1.3371e-02,  2.0557e-02, -6.6826e-03,  4.4168e-03, -5.0926e-02,
        -3.0440e-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([-4.0474e-02, -4.1580e-01,  5.4559e-03,  2.2862e-02,  4.6664e-02,
         4.4094e-02,  1.6829e-02,  1.0301e-04, -2.6385e-03,  1.3135e-02,
         6.5800e-03,  1.7794e-02,  3.1771e-02, -1.4296e-02,  1.0574e-01,
         1.1479e-02, -1.8728e-03,  4.9592e-02,  4.7235e-02,  4.3600e-02,
         3.3312e-02,  8.4414e-04,  2.7821e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.7841e-02,  5.1349e-01,  5.6980e-02, -2.0539e-02,  1.1977e-04,
        -1.9461e-02,  3.1869e-02, -2.3004e-02,  1.7351e-02,  1.7299e-02,
        -5.9997e-03,  3.2124e-02,  2.1323e-02,  1.0150e-02,  8.3878e-03,
        -2.6391e-03, -1.9699e-02,  1.2004e-02,  9.7671e-03,  1.6369e-03,
        -1.7077e-02, -3.0101e-03,  1.7921e-04, -1.6432e-02, -7.3749e-03,
         2.1592e-02,  4.8601e-03,  2.2789e-02, -5.0024e-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([ 0.0892,  0.5233,  0.1005,  0.0475,  0.1635,  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,  0.0000,  0.0000,  0.0000,  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, -0.2662,  0.0685, -0.0245,  0.0554, -0.0277,  0.0334, -0.0660,
        -0.0171, -0.0148, -0.0485, -0.0777, -0.0246, -0.0026, -0.0315,  0.0199,
         0.0219,  0.0691, -0.1020,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.0855e-02,  5.7219e-01,  1.9223e-02, -2.3131e-03, -5.3452e-02,
        -1.7955e-02,  4.2279e-03,  4.8655e-03,  1.4702e-02,  1.4185e-02,
         4.4029e-03, -9.1038e-03,  4.7813e-04,  5.6908e-02,  3.7618e-02,
         1.5487e-02,  1.3310e-02,  1.0296e-05,  4.6427e-02,  3.1313e-03,
         1.9237e-02, -9.6715e-03, -2.4020e-03, -1.0431e-02, -1.7130e-02,
        -7.6743e-03, -3.4283e-03,  9.1829e-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([ 0.0645,  0.4081, -0.0117,  0.0566,  0.0681, -0.0148,  0.0346,  0.0016,
         0.0168,  0.0181, -0.0049,  0.0061, -0.0094,  0.0080, -0.0534,  0.0055,
        -0.0145, -0.0124,  0.0086,  0.0152, -0.0188,  0.0036,  0.0195, -0.0090,
         0.0061,  0.0101, -0.0206, -0.0173, -0.0013, -0.0227,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0144,  0.6089,  0.0266, -0.0455,  0.0263, -0.0225, -0.0753, -0.0503,
        -0.0033, -0.0506, -0.0209, -0.0036,  0.0082,  0.0235,  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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  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,  0.4497,  0.1056,  0.0519, -0.0041,  0.0067,  0.0319, -0.0553,
         0.0077, -0.0294,  0.0123, -0.0036,  0.0027,  0.0186, -0.0052,  0.0329,
         0.0334,  0.0238, -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,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0306, -0.6076, -0.0145,  0.0076, -0.0606, -0.0235,  0.0041,  0.0024,
         0.0303, -0.0028, -0.0142, -0.0092, -0.0179, -0.0085,  0.0140,  0.0115,
        -0.0068, -0.0049,  0.0061, -0.0018,  0.0079, -0.0114,  0.0018,  0.0249,
         0.0079,  0.0042, -0.0138, -0.0061,  0.0192, -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],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #400: [tensor([-2.0349e-01, -3.2453e-01,  7.2221e-02, -6.4107e-02, -3.5028e-02,
        -6.6045e-03,  5.6530e-02,  2.8046e-02,  1.9463e-02,  2.9817e-03,
         3.1153e-02,  3.3642e-02, -4.0520e-03, -6.7216e-03, -1.8325e-02,
         8.0090e-04,  1.1733e-02,  1.6872e-02, -2.0246e-03,  2.3291e-04,
         4.0203e-02, -2.1244e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.2359e-02,  5.8018e-01,  2.9897e-02, -8.4284e-03,  3.2593e-02,
        -2.7150e-03, -3.3705e-02,  7.5431e-03, -1.7405e-02, -2.2615e-02,
        -8.1413e-02,  1.8261e-02, -4.6318e-02,  4.0886e-04, -1.4953e-02,
        -9.5362e-03,  2.4388e-03, -3.1915e-02, -6.7822e-03, -5.3712e-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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.1897e-02,  4.9117e-01,  6.5247e-02,  1.2264e-02,  7.2430e-02,
         4.4419e-03, -9.7002e-03,  2.9194e-03,  1.9880e-02,  1.1429e-02,
         3.3833e-02, -1.0878e-02,  1.2463e-03, -2.9085e-02,  2.2691e-02,
        -3.3524e-04,  2.7535e-02,  4.3789e-02,  2.4182e-02,  4.4243e-03,
        -1.0923e-03,  2.6897e-02, -7.4205e-03,  1.7953e-02, -1.7262e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 9.7935e-02,  3.5926e-01,  6.1914e-02, -9.0074e-02, -6.6956e-03,
         1.2535e-02,  5.7423e-03,  1.3167e-03,  1.2098e-02, -1.2095e-02,
         5.1753e-02,  3.0293e-02,  6.6261e-03,  1.7770e-03,  1.2431e-02,
         3.4854e-02, -7.9725e-03,  2.5097e-05, -1.6900e-02, -9.6866e-03,
         1.6271e-02, -2.0913e-02,  9.8332e-03, -2.3177e-03, -2.3093e-02,
         6.3027e-03,  2.4748e-02,  2.8782e-02, -3.5756e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+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,  0.6018, -0.0348, -0.0482, -0.0094, -0.0151,  0.0414, -0.0140,
        -0.0721,  0.0097,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0532,  0.4058,  0.0283,  0.0137,  0.0251,  0.0297,  0.0084, -0.0210,
         0.0343, -0.0204,  0.0204,  0.0141, -0.0135,  0.0099,  0.0074, -0.0097,
         0.0391,  0.0147,  0.0951,  0.0250,  0.0102, -0.0012, -0.0092, -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([-9.1096e-03, -3.9043e-01, -9.7330e-02,  2.8878e-02,  3.1311e-02,
        -4.4625e-03, -2.7162e-02,  1.1752e-02,  2.3494e-02,  8.0142e-03,
         8.3641e-04,  9.2562e-03,  5.2586e-03,  7.1123e-03,  4.4715e-03,
        -5.0506e-03,  5.4431e-03, -2.1983e-04, -3.7757e-03, -3.0057e-03,
         8.7399e-03,  5.2848e-03,  3.1337e-03, -9.7248e-03, -6.9869e-03,
         5.7350e-03,  2.3207e-03, -1.3468e-02, -8.4312e-03,  1.4115e-02,
        -4.3423e-02, -5.6710e-03,  1.3942e-03,  1.4264e-03,  1.2749e-02,
         8.4469e-03,  3.0874e-02,  1.0723e-02,  5.4475e-03,  9.3788e-03,
         8.0002e-03, -2.1311e-02, -3.6229e-03, -9.8410e-04,  1.2062e-02,
         1.3530e-02,  1.3010e-02, -4.3637e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1341, -0.4547, -0.1159, -0.0029, -0.0193,  0.0374,  0.0204,  0.0446,
         0.0491, -0.0015,  0.0168,  0.0112,  0.0057,  0.0220,  0.0295, -0.0244,
         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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0036, -0.6489,  0.0558,  0.0467, -0.0092, -0.0230, -0.0194, -0.0401,
        -0.0091, -0.0346, -0.0273, -0.0080,  0.0373, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.4890e-02,  5.2203e-01,  3.2327e-02,  1.5367e-02,  2.7002e-03,
         4.1510e-05,  1.0331e-03,  3.6263e-03,  9.8616e-03,  2.3729e-02,
        -9.8519e-03, -2.0008e-02, -3.7922e-03,  2.5553e-03, -2.0172e-02,
         7.5829e-03, -1.5594e-02,  2.6919e-03,  1.4717e-02, -1.4935e-02,
        -1.1766e-02,  9.7363e-03,  8.8479e-03, -4.2647e-03, -2.0994e-02,
        -1.5327e-02,  1.6387e-03, -8.1005e-03,  4.0309e-03, -4.2193e-03,
         4.9598e-03, -2.2231e-02,  3.7251e-02,  3.9124e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.4330e-02,  4.7533e-01, -2.5519e-02,  3.9798e-02,  3.3265e-02,
        -1.7899e-03, -2.1240e-02,  1.3753e-02,  1.1512e-02, -9.3543e-03,
        -2.2428e-03, -1.1728e-02,  2.2573e-03,  4.6372e-03,  1.0751e-02,
         2.5977e-02, -1.5074e-04, -3.8190e-03, -2.7760e-03, -7.3228e-03,
        -3.6612e-03, -3.3487e-03, -2.2526e-03,  4.4344e-03, -2.5476e-02,
        -2.3204e-02, -8.8157e-03, -9.0820e-05,  3.7320e-03, -1.0876e-02,
         2.0350e-03, -1.8050e-03, -1.5787e-02,  6.3138e-03,  1.0611e-03,
        -7.1638e-03, -3.4459e-03, -4.9426e-03, -1.9452e-02, -2.0765e-03,
        -3.8318e-03,  8.4991e-02,  4.3653e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.5525e-02,  5.6035e-01,  7.8899e-02,  9.2997e-03, -9.0654e-03,
        -3.4295e-03,  3.1980e-03, -7.1456e-03,  6.5119e-03, -6.0483e-03,
        -2.9737e-02,  2.4004e-03, -1.0793e-02,  1.6996e-02, -1.9466e-03,
         4.4364e-03,  5.4594e-03,  7.6543e-03, -1.3095e-03,  8.1900e-03,
         1.0825e-02, -3.0077e-04, -2.0148e-02,  3.8346e-03,  1.6735e-04,
        -2.6568e-03, -1.1957e-02, -5.7962e-04, -1.7008e-04,  1.3056e-02,
         2.5282e-02,  3.9287e-03, -3.7102e-02, -5.0655e-02, -3.0944e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  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([ 6.4355e-02,  5.2274e-01, -1.5208e-02, -4.3863e-02, -5.0030e-03,
        -2.7044e-02,  1.5299e-02, -4.1749e-03,  2.7102e-04, -4.5194e-03,
        -9.2949e-03,  8.6863e-03,  4.2451e-04,  1.8390e-02,  5.6597e-02,
         2.2666e-02, -2.2994e-03, -4.2158e-03, -1.5915e-02, -1.6337e-02,
         7.8883e-03, -4.6842e-02, -8.7970e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.7089e-02, -4.3400e-01, -8.8787e-02,  5.0190e-02,  5.9387e-02,
        -4.8530e-03,  1.4809e-02,  6.0005e-03,  2.9202e-02,  1.2913e-02,
         1.1193e-02,  1.5379e-02, -1.7657e-02,  4.2272e-03,  8.6810e-03,
         4.0590e-03, -1.7270e-02,  4.9528e-05,  3.7112e-02,  1.3212e-02,
        -3.6746e-03, -6.8924e-03,  1.8846e-02,  1.5634e-02,  8.5815e-03,
         2.9567e-03, -5.5013e-03,  1.2786e-02,  1.3114e-02,  1.7190e-03,
        -5.2177e-03,  4.9191e-04,  7.5827e-03,  4.0210e-03, -7.0953e-03,
        -9.8172e-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([ 0.1674,  0.4927,  0.0168, -0.0584,  0.0057, -0.0195,  0.0294,  0.0090,
         0.0211, -0.0210,  0.0086, -0.0329,  0.0131, -0.0713, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0288, -0.5937, -0.0374, -0.0140,  0.0109,  0.0122,  0.0608, -0.0163,
         0.0040, -0.0336,  0.0190,  0.0414,  0.0275,  0.0474, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0495, -0.3926, -0.0143,  0.0086,  0.0104, -0.0152, -0.0081,  0.0214,
         0.0129, -0.0050,  0.0204, -0.0066, -0.0272, -0.0023,  0.0125, -0.0165,
        -0.0118, -0.0136, -0.0029,  0.0108, -0.0121,  0.0103,  0.0113,  0.0097,
         0.0212,  0.0325, -0.0014, -0.0048, -0.0091, -0.0080, -0.0038, -0.0058,
        -0.0138, -0.0068,  0.0120,  0.0219,  0.0028,  0.0075, -0.0068,  0.0036,
         0.0133, -0.0274, -0.0053, -0.0020, -0.0106, -0.0163, -0.0055,  0.0082,
        -0.0012,  0.0108, -0.0181,  0.0135], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1075, -0.5443, -0.0139,  0.0370,  0.0278,  0.0305,  0.0062,  0.0121,
         0.0208, -0.0535,  0.0085,  0.0064,  0.0049,  0.0035,  0.0064, -0.0058,
        -0.0032, -0.0270,  0.0658, -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([-0.0012, -0.4099, -0.0068,  0.0327, -0.0228,  0.0251,  0.0259, -0.0132,
        -0.0333, -0.0040,  0.0139, -0.0280, -0.0381, -0.0011,  0.0061,  0.0138,
         0.0092, -0.0283,  0.0731,  0.0289, -0.0956, -0.0890,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  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,  0.4164,  0.0930, -0.0036,  0.0969,  0.0394,  0.0082,  0.1318,
        -0.0200,  0.0187, -0.0655,  0.0451,  0.0023,  0.0134,  0.0019,  0.0346,
         0.0021, -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,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0979, -0.4897, -0.0066, -0.0122,  0.0010, -0.0084,  0.0379,  0.0577,
         0.0401, -0.0026,  0.0013, -0.0027,  0.0203, -0.0009, -0.0011,  0.0031,
        -0.0148,  0.0041, -0.0034, -0.0146,  0.0200,  0.0115,  0.0130, -0.0047,
         0.0906,  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([-1.3167e-01, -4.2391e-01, -3.7207e-02,  4.1686e-02, -1.9927e-02,
        -1.5814e-02, -3.4506e-02, -3.8048e-04, -4.2969e-03, -9.9648e-03,
        -6.9860e-02,  9.7041e-03,  1.3426e-02,  3.2970e-03,  9.8295e-03,
        -1.2598e-02, -7.5537e-03, -2.0218e-02, -1.4357e-02,  5.4863e-02,
        -6.4933e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1178, -0.1574, -0.0223,  0.0349,  0.0059, -0.0389, -0.0840,  0.0211,
        -0.0022,  0.0508,  0.0138,  0.0028,  0.0050,  0.0467,  0.0040,  0.0338,
         0.0255, -0.0097, -0.0034, -0.0191,  0.0179, -0.0036, -0.0228, -0.0114,
         0.0292,  0.0090, -0.0002, -0.0030, -0.0141, -0.0006, -0.0399, -0.0083,
         0.0019, -0.0040,  0.0009, -0.0090, -0.0052,  0.0028,  0.0106, -0.0071,
         0.0024, -0.0060,  0.0003,  0.0240,  0.0040,  0.0120,  0.0506,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1061,  0.2976,  0.1372,  0.0049,  0.0197, -0.0588, -0.0433, -0.1417,
        -0.1405,  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], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #500: [tensor([ 8.4640e-02, -4.7657e-01, -4.5265e-02, -2.4655e-02, -1.6815e-02,
         1.5646e-02,  9.1401e-03,  2.7508e-03,  4.3848e-03, -1.9684e-02,
        -1.0381e-02, -2.2234e-04,  4.8712e-03,  1.8087e-03,  1.8945e-03,
         2.4564e-02,  2.8709e-04,  5.2016e-05,  1.0027e-04,  9.1065e-03,
         1.1051e-02,  1.9901e-02,  1.4796e-02,  5.3065e-03, -2.0318e-03,
        -4.2396e-03,  1.8178e-02, -5.1721e-03, -1.9369e-03, -2.0042e-03,
         8.1288e-03,  1.5558e-03,  7.2821e-03,  6.6573e-03,  6.2912e-03,
         2.8122e-02,  8.2058e-03,  1.5650e-02,  1.6026e-02, -1.0076e-02,
         1.7156e-03,  7.5904e-03,  3.3493e-03, -7.1963e-03, -5.5545e-03,
         7.5430e-03, -2.1603e-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([-0.0167,  0.5489,  0.0645,  0.0134,  0.0826,  0.0118, -0.0292,  0.0359,
         0.0029,  0.0035, -0.0034, -0.0173, -0.0055, -0.0229,  0.0142,  0.0098,
        -0.0058, -0.0039,  0.0201,  0.0053, -0.0034, -0.0128, -0.0008,  0.0152,
         0.0156, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0316, -0.3466, -0.0369, -0.0465,  0.0064,  0.0613,  0.0020, -0.0056,
        -0.0054, -0.0067,  0.0379, -0.0140, -0.0108,  0.0101,  0.0106,  0.0406,
         0.0051, -0.0015,  0.0101,  0.0076,  0.0091, -0.0259,  0.0311,  0.0054,
         0.0101,  0.0101,  0.0032,  0.1050,  0.0044,  0.0048, -0.0085, -0.0034,
        -0.0027, -0.0007, -0.0035, -0.0462, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0911,  0.3466,  0.0937, -0.1196, -0.0006,  0.0386, -0.0739,  0.0106,
        -0.0160, -0.0181,  0.0055,  0.0268,  0.0339, -0.0759,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0134, -0.5910, -0.0120,  0.0358,  0.0535,  0.0308, -0.0085,  0.0066,
        -0.0410,  0.0151,  0.0227, -0.0253, -0.0141,  0.0292,  0.0413, -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,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.4877e-02, -5.7691e-01,  1.3427e-02,  2.0011e-02, -9.1927e-03,
         1.5448e-02,  1.3997e-02,  1.7754e-02, -5.4683e-03, -1.1333e-02,
         2.7212e-03,  8.5866e-03, -3.1918e-02, -2.7723e-02,  8.3980e-03,
        -2.8972e-02, -3.1169e-05,  2.0527e-03,  5.7008e-03, -1.2720e-02,
         1.9249e-02, -2.1213e-02, -2.4979e-02,  1.6110e-03,  4.1626e-03,
         2.9115e-03, -6.1809e-03, -7.8776e-03,  1.5040e-03,  4.3069e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.3242e-01,  5.4332e-01, -1.2719e-02,  1.3985e-02,  1.9546e-04,
         8.4059e-03, -1.4156e-03, -8.8134e-03, -2.3706e-02, -3.9019e-02,
        -1.9772e-02, -2.5674e-02, -3.3478e-04, -1.0871e-02,  1.5266e-03,
        -9.6818e-03, -5.4674e-03,  1.1058e-02,  6.1326e-03, -1.0835e-02,
         4.2211e-03, -3.0577e-03, -9.1111e-03,  4.5923e-04, -8.3315e-03,
         6.1256e-03, -3.2611e-04,  2.2116e-03,  1.8696e-02,  8.0308e-03,
         4.9101e-03,  1.6688e-02,  1.6261e-03, -6.3258e-03,  2.2716e-02,
        -1.8127e-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([ 1.8593e-02,  4.9282e-01, -1.3946e-02,  2.9130e-02, -1.2812e-02,
        -1.6230e-02, -8.2749e-02, -3.6962e-02, -1.1609e-02,  5.9254e-03,
        -5.3993e-03, -3.5603e-03, -1.3009e-02,  8.0836e-03, -7.6908e-03,
        -1.6476e-03, -1.4220e-02,  2.6205e-03, -2.4518e-03, -5.7862e-03,
        -1.4910e-03, -2.8503e-04, -1.8185e-03,  1.3901e-02,  3.0590e-02,
         1.1984e-02, -1.8398e-03,  6.8372e-04,  7.4550e-03, -8.0396e-03,
        -4.9868e-04,  1.6638e-02, -3.0158e-03, -7.0989e-02, -4.5529e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.2147,  0.2943,  0.0247,  0.1145,  0.0654,  0.0064, -0.0202,  0.0241,
         0.0547, -0.0468,  0.1343,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  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, -0.3654, -0.1451, -0.0893, -0.0453,  0.0422,  0.0065, -0.0092,
         0.0059,  0.0173,  0.0103,  0.0058, -0.0078, -0.0193,  0.0184,  0.0147,
         0.0096, -0.0041,  0.0144,  0.0230,  0.0125,  0.0389,  0.0029, -0.0193,
         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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0252,  0.4683,  0.0275, -0.0053,  0.0015, -0.0505, -0.0241,  0.0221,
        -0.0067,  0.0245,  0.0256,  0.0359,  0.0145,  0.0124,  0.0076, -0.0052,
        -0.0077, -0.0126, -0.0071,  0.0623, -0.0181, -0.0239,  0.0105,  0.0154,
         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([ 3.1761e-02,  4.4016e-01,  6.0385e-02,  2.0740e-02,  1.0018e-02,
         9.4798e-03,  1.7686e-02,  1.3231e-03, -5.6370e-04,  1.1952e-02,
         8.4920e-05, -9.2047e-04, -5.0120e-03,  1.4868e-03, -1.9518e-03,
        -6.9678e-04, -4.6648e-04, -6.4263e-03,  1.2279e-02, -1.1105e-02,
         4.9808e-03, -7.3906e-03, -6.9676e-03, -3.9805e-03,  4.1573e-03,
         1.3885e-02, -3.3662e-02, -8.8159e-03,  3.0673e-03, -1.3602e-02,
        -2.0699e-02, -2.2977e-02, -5.3446e-02, -3.6929e-03, -1.0691e-02,
         5.3782e-03,  2.7035e-03,  8.9052e-03,  8.8538e-03,  3.0471e-03,
        -1.0404e-02,  5.8884e-04, -9.8952e-03,  7.2669e-03,  6.6342e-03,
         8.4022e-03,  4.2832e-04, -1.3024e-02, -8.3737e-04, -1.9515e-02,
         3.1082e-03,  2.7354e-03,  2.0403e-02,  1.1357e-02], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #550: [tensor([ 0.0221, -0.7113,  0.0097, -0.0452,  0.0508, -0.0135, -0.0040,  0.0044,
        -0.0079, -0.0139,  0.0179, -0.0384, -0.0541,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0286, -0.4282,  0.0422, -0.0308, -0.0318, -0.0286,  0.0097, -0.0359,
        -0.0110,  0.0062,  0.0131,  0.0208,  0.0258, -0.0173,  0.0116, -0.0265,
         0.0347, -0.0071,  0.0364, -0.0192,  0.1344,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  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, -0.5893, -0.0151,  0.0497,  0.0215,  0.1838,  0.0019, -0.0423,
        -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,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1769, -0.3946, -0.0005, -0.0112, -0.0089,  0.0024,  0.0231,  0.0608,
         0.0017, -0.0032, -0.0465,  0.0011,  0.0099,  0.0207,  0.0240, -0.0232,
        -0.0038,  0.0024, -0.0450, -0.0191, -0.0236, -0.0018,  0.0010, -0.0008,
         0.0197,  0.0100,  0.0527, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0973, -0.4300, -0.0659, -0.0253,  0.0259, -0.0295, -0.0105,  0.0238,
        -0.0162,  0.0172,  0.0288, -0.0054, -0.0078, -0.0011, -0.0150, -0.0223,
        -0.0259,  0.0131,  0.0243,  0.0096, -0.0738,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1383, -0.4285,  0.1091, -0.0353, -0.0163, -0.0136, -0.0139, -0.0021,
         0.0089, -0.0004,  0.0033,  0.0160,  0.0127, -0.0042, -0.0126,  0.0105,
        -0.0039, -0.0046, -0.0033, -0.0125,  0.0023, -0.0007, -0.0044,  0.0020,
        -0.0055, -0.0024,  0.0024, -0.0017, -0.0040,  0.0014,  0.0371, -0.0861,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.5728e-02, -4.6494e-01,  4.3114e-03,  3.5270e-02,  1.7104e-02,
        -1.2115e-02, -7.5991e-04,  2.7093e-02,  1.6228e-02,  6.9644e-03,
        -1.8445e-02, -1.4105e-02,  6.6603e-03, -5.7750e-03, -2.4889e-02,
        -2.2891e-02, -1.5215e-02, -1.1257e-02,  1.1716e-03,  1.5433e-02,
        -4.4253e-03, -3.2574e-03, -8.0420e-03,  5.1404e-05,  7.6577e-03,
        -3.9257e-03, -4.1500e-03, -5.7388e-03, -7.0863e-03,  6.9836e-04,
         5.1485e-02,  9.5441e-04,  1.4602e-02, -3.4480e-03, -2.0992e-02,
        -3.6350e-03,  4.0686e-02,  1.1022e-02,  1.4773e-03,  3.1095e-03,
         1.0125e-02, -7.9015e-03,  3.5000e-04,  3.8820e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0678, -0.6143, -0.0125, -0.0034,  0.0078,  0.0396,  0.0314,  0.0190,
        -0.0282,  0.0524,  0.0013,  0.0322,  0.0007,  0.0267, -0.0031, -0.0101,
        -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,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.6676e-02,  6.2234e-01, -2.4540e-02,  9.3121e-03,  3.1754e-02,
        -1.8907e-02, -8.7759e-03,  3.2144e-04, -1.5920e-02,  9.0598e-03,
        -7.4963e-03, -9.3430e-03, -1.5654e-02, -3.0840e-04,  7.8799e-03,
         1.0852e-02,  9.9222e-03,  2.3998e-03, -2.2444e-02, -2.1054e-02,
         5.1580e-03,  1.8715e-02,  1.3651e-02,  3.0760e-02, -6.7547e-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([-4.4943e-02, -5.0184e-01, -5.0249e-02, -1.5207e-02, -2.1570e-02,
        -1.4943e-02, -2.7355e-03, -2.5314e-02,  9.0734e-03, -3.0092e-02,
         2.9841e-02, -2.1281e-02, -2.6326e-02, -1.8706e-03, -6.5857e-03,
         4.9661e-04, -3.1941e-03, -4.2714e-03, -3.4874e-02, -5.3578e-02,
         2.3638e-02,  7.8077e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0627, -0.4959,  0.0624, -0.0358, -0.0031, -0.1483, -0.0057, -0.0947,
         0.0915,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.5635e-02, -4.3451e-01,  2.9223e-02, -4.0092e-02,  4.5911e-03,
         1.4430e-02,  2.7266e-02,  8.4793e-03, -3.3491e-03, -1.8960e-03,
         7.1553e-03, -1.1952e-02,  2.6251e-02, -2.1507e-03,  1.6515e-02,
         3.3699e-02,  2.7669e-03, -4.4049e-02,  1.7452e-02,  8.7325e-03,
         3.5680e-03,  1.3437e-02,  1.2778e-02, -1.0876e-02,  9.3948e-03,
         4.0948e-02,  8.0189e-03, -5.3500e-03, -2.2780e-02,  6.0196e-03,
         4.0956e-04,  1.2341e-02, -1.2528e-02, -2.1359e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  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([-6.7930e-03,  4.7898e-01, -5.3483e-02, -3.7174e-02,  1.1854e-02,
         1.0609e-02, -3.8876e-02,  4.0175e-03, -5.1532e-03,  4.7817e-03,
         4.0398e-03, -5.0212e-03,  1.1215e-02, -4.9078e-03, -6.9315e-03,
        -3.2160e-02, -1.9168e-02,  9.5306e-03,  2.1870e-02, -1.7190e-03,
         1.0243e-02,  4.6117e-03, -1.5541e-02, -2.9966e-03,  1.7057e-02,
         6.4052e-03, -1.4177e-04,  6.1411e-03,  3.6075e-03,  1.0247e-02,
        -1.4798e-02,  9.5741e-03, -8.7038e-03, -6.0793e-03, -1.6155e-02,
         7.9157e-03,  6.3614e-03, -9.5685e-03, -6.2878e-03, -1.7873e-03,
        -6.2330e-03, -2.2528e-03, -5.2336e-02,  6.6678e-03,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0493,  0.6334, -0.0259, -0.0547,  0.0027, -0.0075, -0.0086, -0.0137,
         0.0018,  0.0050,  0.0171, -0.0096,  0.0106,  0.0061, -0.0014, -0.0201,
        -0.0257,  0.0168, -0.0017, -0.0206, -0.0080,  0.0161, -0.0099,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0086, -0.5535, -0.1383,  0.0674,  0.0225,  0.0317,  0.0010,  0.0028,
        -0.0057,  0.0617,  0.0379,  0.0212,  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,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0010,  0.4115, -0.0060, -0.0672,  0.0052, -0.0322, -0.0111, -0.0292,
        -0.0236, -0.0466, -0.0111,  0.0500,  0.0202,  0.0039, -0.0230, -0.0185,
        -0.0179, -0.0032, -0.0055,  0.0039, -0.0864, -0.1230,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.7950e-01,  6.6197e-01,  1.4104e-02,  2.7171e-02, -4.2309e-03,
        -6.1966e-04, -2.1267e-02, -1.0130e-02, -1.3025e-02, -6.7980e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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, -0.3771,  0.1286, -0.0044, -0.0493,  0.0171,  0.0277, -0.0022,
        -0.0062, -0.0197,  0.0307, -0.0007,  0.0104, -0.0146, -0.0078, -0.0033,
        -0.0284, -0.0057, -0.0218, -0.0090, -0.0257, -0.0956,  0.0754,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-9.4932e-02, -3.4521e-01, -4.8405e-02, -2.2010e-02,  3.8974e-03,
         8.9853e-03,  1.9138e-02, -2.7353e-02, -2.8205e-03,  1.1270e-02,
        -9.3192e-03, -6.5370e-04,  7.8216e-03,  1.0070e-02, -5.4183e-05,
         8.6499e-03,  5.7023e-03,  1.9495e-02,  1.0936e-02,  2.9388e-02,
         5.7939e-03,  8.7424e-03,  3.2483e-03,  8.3806e-03,  1.1032e-02,
        -5.0014e-03,  1.7400e-03,  1.2594e-02,  4.3397e-05,  9.6242e-03,
         2.1595e-02,  1.8000e-02,  1.9364e-02, -5.3288e-03,  1.9356e-02,
         7.6387e-03,  3.7631e-03,  6.1639e-04, -5.3567e-03, -8.4460e-03,
        -1.2727e-02, -4.3192e-03, -1.3868e-02, -2.8723e-02, -7.8590e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0994,  0.6034, -0.0157, -0.0349, -0.0010, -0.0083,  0.0403, -0.0334,
        -0.0053, -0.0084, -0.0280, -0.0069,  0.0092, -0.0167, -0.0100, -0.0123,
        -0.0197,  0.0095,  0.0032, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1682, -0.3905,  0.0362, -0.0050,  0.0101,  0.0124,  0.0199,  0.0263,
         0.0376,  0.0209,  0.0156,  0.0280,  0.0165, -0.0069,  0.0155,  0.0211,
         0.0498, -0.0045, -0.0220, -0.0083,  0.0080, -0.0203,  0.0218,  0.0220,
         0.0052, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.3511e-02, -6.1066e-01, -5.0349e-02,  7.0725e-03,  3.8216e-02,
        -8.7730e-03, -6.4405e-03, -6.0007e-03, -3.8484e-02, -8.7771e-03,
        -1.2463e-02,  9.1018e-03, -4.2776e-03, -1.5294e-02, -1.0544e-03,
        -4.9165e-03,  4.1586e-03,  9.8522e-03, -6.1830e-03, -3.9627e-03,
        -4.0083e-03,  5.6955e-05, -1.3053e-02,  5.8852e-03,  8.2577e-03,
         5.6697e-03,  1.0537e-02, -2.2989e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2270, -0.5881, -0.0505,  0.0136, -0.0484, -0.0086, -0.0062,  0.0470,
        -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.1682e-02, -4.8111e-01,  2.7944e-02,  1.0181e-01, -2.2064e-02,
        -2.5636e-02,  4.4804e-03,  2.3784e-02,  5.6402e-03, -3.7591e-02,
        -1.1275e-02, -6.0421e-03, -2.2383e-03, -1.2318e-02, -6.1116e-02,
         1.3520e-02, -1.2976e-03, -7.5916e-03,  2.6591e-02, -9.7498e-03,
        -6.2796e-03,  2.9399e-02,  5.0776e-03,  8.4410e-03, -1.6044e-02,
        -4.6946e-04,  2.0812e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  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([ 4.2646e-02,  6.3369e-01,  3.0248e-02, -4.5273e-02,  1.2525e-02,
        -1.7917e-02, -6.7477e-03,  1.2566e-02,  4.8792e-03, -3.3297e-03,
        -9.0251e-03,  3.2276e-02,  4.2387e-03,  8.0858e-03,  1.0025e-03,
        -1.2229e-03, -3.8646e-03,  3.7383e-03,  2.4189e-04, -1.2783e-02,
         3.2895e-03, -7.2137e-03,  7.7600e-03,  1.6922e-02, -1.4011e-03,
        -9.6873e-03,  9.2257e-03,  1.9994e-03,  4.4482e-03,  2.2079e-02,
         1.0506e-03,  7.0543e-03, -4.0517e-04,  1.3391e-03, -6.7560e-03,
        -8.8447e-04, -6.0976e-03, -6.0845e-03,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.8962e-02,  4.0358e-01,  1.1380e-01,  1.9226e-02, -1.2418e-02,
        -4.2284e-03, -1.5531e-02,  3.2780e-03,  1.3858e-02,  8.2197e-03,
         7.0237e-03,  1.2479e-02, -3.8355e-03, -9.0959e-03,  1.4026e-02,
         8.6305e-03, -1.6603e-03,  1.4317e-04, -2.3105e-02,  1.5973e-03,
        -1.1974e-02,  1.5294e-02,  1.4940e-03, -1.2857e-02, -1.0465e-02,
         3.9893e-03,  9.5740e-03,  8.8298e-03, -6.5869e-04,  2.3925e-02,
        -1.9795e-02,  1.0484e-02,  1.0186e-03, -6.8799e-03, -3.1036e-03,
         9.5669e-03,  6.1329e-02, -1.4062e-02,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.8687e-02, -6.3589e-01, -3.9540e-02,  3.5381e-04,  1.9995e-02,
        -1.6732e-04, -6.5363e-03,  1.3909e-02,  1.9609e-02,  4.7002e-03,
         1.9815e-03,  2.2075e-03, -6.2572e-02,  5.1403e-03,  9.1486e-03,
        -1.0656e-02,  3.6521e-03,  4.7854e-03, -1.4940e-02, -2.0236e-03,
        -6.9509e-03, -2.8495e-03,  2.5381e-02, -4.8661e-04,  1.6289e-03,
         9.0176e-03,  6.3364e-03, -1.8697e-02,  1.1762e-02, -4.0392e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.7178e-02, -4.6288e-01,  1.9692e-01, -8.6875e-03, -1.7554e-02,
         9.6233e-03,  1.6606e-03,  2.9854e-02,  6.4956e-05,  9.5553e-03,
         3.5798e-03, -5.3146e-03,  1.6651e-03, -8.3883e-03,  5.0481e-04,
         1.1525e-02, -3.6953e-03,  2.8013e-03,  1.6893e-04, -1.2460e-03,
         6.8065e-03,  2.3063e-03,  3.3594e-03,  2.4032e-03,  1.1536e-02,
        -3.2087e-03, -7.3817e-03, -7.6678e-03, -6.2689e-03,  7.6084e-03,
        -3.3342e-03,  9.1146e-03,  4.8936e-03, -1.1814e-03,  1.9459e-03,
        -1.5150e-03,  5.6709e-03,  4.6943e-03, -2.5176e-02, -2.1059e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.3065e-04, -4.4080e-01,  1.1637e-02,  9.1115e-03, -3.1379e-02,
         1.2524e-02,  3.4589e-03, -2.1525e-02, -1.4651e-02,  1.9177e-03,
        -4.6784e-02,  8.4958e-03,  1.8468e-02, -1.9722e-02,  6.9799e-03,
         2.0551e-03,  3.7547e-03, -9.4048e-03, -4.0803e-02,  3.6598e-02,
        -3.0533e-02, -2.8914e-02,  3.4224e-03,  1.3742e-02,  5.4325e-03,
         2.3494e-03, -3.4308e-03, -4.3204e-03, -4.1793e-03,  3.7000e-03,
         1.3017e-02,  6.5912e-03,  1.2962e-02,  1.5028e-02,  4.1886e-02,
         1.2536e-02, -2.0380e-02, -3.7074e-02,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0073,  0.5331, -0.0051,  0.0442, -0.0017,  0.0124, -0.0189, -0.0524,
        -0.0204, -0.0542,  0.0011, -0.0312,  0.0087, -0.0243, -0.0272,  0.0066,
         0.0086, -0.0044, -0.0110, -0.0038,  0.0157, -0.0772, -0.0123,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.7275e-01, -5.4541e-01, -5.2007e-02,  1.7184e-03, -2.4284e-02,
         2.3018e-02, -2.8110e-03,  2.0559e-03,  1.1016e-02, -4.2060e-03,
        -8.8794e-03, -9.3366e-03, -8.1458e-03,  2.3356e-02,  4.8570e-03,
         5.8624e-03, -4.8350e-05, -1.3709e-03,  1.9886e-03,  3.3470e-03,
        -5.3408e-03, -1.4112e-02, -4.5187e-03, -5.5113e-03, -6.6398e-03,
         4.8101e-03, -1.6428e-02,  9.8803e-03,  1.2751e-02, -7.9553e-03,
         2.6829e-03, -2.9056e-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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0312, -0.5099, -0.0326,  0.0165, -0.0111, -0.0214, -0.1311,  0.0312,
         0.0540,  0.0037, -0.0937,  0.0076,  0.0329,  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([ 3.5565e-02,  5.5114e-01,  2.8232e-02, -3.9169e-02, -3.4869e-02,
        -3.0022e-02,  3.7780e-04, -4.8450e-02, -2.9608e-02, -6.0786e-03,
         1.7922e-02, -5.0399e-03,  7.7631e-03, -1.5468e-02, -4.2614e-03,
        -1.0846e-02, -2.6394e-03,  2.5553e-02, -4.1859e-03, -2.8421e-02,
         1.7175e-02, -1.9777e-02, -2.4861e-02, -1.2578e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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, -0.6092, -0.0022, -0.0200,  0.0244,  0.0741, -0.0097, -0.0013,
        -0.0167, -0.0011, -0.0101,  0.0210, -0.0014,  0.0114, -0.0014, -0.0199,
        -0.0103, -0.0081,  0.0092, -0.0501,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1920, -0.4154,  0.0074,  0.0009, -0.0316,  0.0007, -0.0250, -0.0020,
        -0.0111,  0.0013, -0.0061,  0.0399, -0.0030, -0.0297,  0.0061, -0.0024,
        -0.0048,  0.0147,  0.0013, -0.0214,  0.0046, -0.0270, -0.0005,  0.0128,
         0.0057, -0.0094, -0.1231,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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,  0.5306, -0.0540,  0.0238, -0.0219, -0.0029, -0.0163, -0.0045,
        -0.0745, -0.0074, -0.0105, -0.0435, -0.0090, -0.0107,  0.0225,  0.0158,
         0.0094,  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],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #150: [tensor([-9.5609e-03,  6.0365e-01, -6.3271e-02,  1.6344e-02, -1.6230e-02,
         1.9390e-02,  7.3578e-03, -7.8136e-03, -5.0116e-03, -2.0891e-03,
         2.0723e-02,  9.4211e-03, -1.8560e-03, -5.7069e-03,  3.9969e-03,
         1.0375e-02,  6.0400e-03, -1.8996e-03, -5.2758e-04,  7.9594e-03,
         5.7489e-03, -1.9867e-02, -7.0841e-04,  1.9625e-03, -6.9134e-03,
        -1.0592e-02,  7.9931e-04,  3.3585e-03,  1.8399e-03,  1.0743e-03,
        -8.3292e-03, -1.5870e-02, -1.5387e-02, -1.4346e-03, -8.6990e-04,
        -5.5085e-04, -5.1844e-03, -2.5151e-04,  1.0801e-02,  6.8302e-03,
        -6.6316e-04, -2.6410e-03, -2.4367e-03,  5.0098e-03,  4.4497e-02,
         7.1590e-03,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.1525e-02,  3.5354e-01,  2.8322e-02, -3.4669e-02, -3.2609e-02,
        -5.1857e-03, -1.6576e-02, -6.9959e-03,  3.9491e-03, -9.2687e-03,
        -3.2631e-03, -1.7916e-02, -6.3425e-03, -8.2728e-03, -2.1705e-03,
        -8.9149e-03, -7.9783e-03, -7.1239e-03,  3.0204e-03, -3.1601e-02,
        -1.6744e-02,  4.1655e-02,  1.2231e-02,  4.3594e-05,  1.0503e-02,
         5.8388e-03, -6.4441e-03,  6.5010e-03, -2.2846e-03,  1.7831e-03,
         1.2257e-02, -2.4594e-02,  5.5094e-03,  6.3830e-03,  4.0242e-03,
        -9.0462e-03,  2.3032e-02,  3.5937e-02, -3.3046e-03, -5.7731e-03,
         2.3049e-03, -7.9485e-04, -9.9318e-03, -1.3237e-03,  2.2089e-02,
        -2.3819e-02, -4.0951e-03,  4.2506e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0116,  0.5329,  0.0105, -0.0229, -0.0019, -0.0114,  0.0020, -0.0153,
        -0.0256, -0.0130,  0.0372,  0.0132,  0.0082,  0.0114,  0.0027, -0.0009,
        -0.0058, -0.0237, -0.0016, -0.0046, -0.0129,  0.0047,  0.0373, -0.0059,
         0.0018,  0.0038,  0.0040,  0.0009, -0.0276,  0.0088, -0.0211, -0.0006,
         0.0812,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.4663e-04,  7.9802e-01, -1.2108e-02,  5.8867e-02,  1.0680e-01,
         2.3956e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 6.6165e-03,  4.4919e-01, -1.2614e-01, -1.4187e-02, -2.2376e-02,
        -2.6402e-03,  5.0428e-03,  8.8537e-03, -1.6073e-02, -4.3084e-03,
        -1.2055e-02,  6.8580e-03, -7.4102e-03,  1.9274e-02,  4.7929e-03,
        -7.4235e-04,  1.9542e-03, -9.6866e-03, -4.9367e-03,  2.1704e-02,
         1.3120e-02, -7.6273e-03, -7.7396e-03, -5.8212e-03,  7.8999e-03,
         2.6984e-03, -1.5296e-04,  7.7325e-03,  1.9075e-03, -6.3852e-03,
        -7.1560e-03,  1.7615e-04, -5.3134e-04, -3.2052e-03, -6.0304e-03,
        -4.9029e-03,  7.3395e-04, -7.2821e-03,  5.1941e-04, -4.1330e-03,
         1.0165e-02, -1.0527e-03,  6.9393e-03, -2.7097e-04,  7.0007e-02,
        -7.0972e-02,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.6923e-02, -4.5756e-01,  9.6382e-02, -1.5322e-02, -9.4083e-04,
        -6.0582e-03, -2.0802e-02, -9.2183e-03, -2.3457e-02, -1.4662e-02,
        -2.4609e-02,  1.0155e-02, -4.2094e-04, -2.5458e-02, -1.1254e-02,
         2.6797e-02,  8.2227e-02,  3.3234e-02,  1.0768e-02, -2.6761e-02,
        -5.3014e-02, -3.3979e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+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, -0.7057, -0.0588,  0.0249, -0.0725,  0.0063,  0.0244, -0.0097,
         0.0132,  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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1067, -0.5564,  0.0306, -0.0297, -0.0010,  0.0052,  0.0113, -0.0333,
        -0.0303,  0.0137,  0.0066, -0.0106,  0.0072,  0.0120,  0.0316,  0.0396,
         0.0387, -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([-0.0024, -0.6089, -0.0503, -0.0529,  0.0476, -0.0318, -0.0825,  0.0078,
         0.0007, -0.0095,  0.0450,  0.0163, -0.0014,  0.0205, -0.0095, -0.0122,
         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([-0.1785, -0.4355, -0.0169,  0.0124, -0.0312,  0.0366,  0.0040, -0.0036,
         0.0139,  0.0394, -0.0184,  0.0374,  0.0258,  0.0332, -0.0111, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1725,  0.5535, -0.0711,  0.0058, -0.0114,  0.0178,  0.0233,  0.0057,
         0.0012,  0.0044, -0.0205,  0.0045,  0.0149,  0.0056, -0.0083,  0.0183,
        -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0881, -0.6236, -0.0718, -0.0019,  0.0075, -0.0125, -0.0022,  0.0213,
         0.0037,  0.1092,  0.0033,  0.0032, -0.0078,  0.0283, -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],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #200: [tensor([-0.1097, -0.5110,  0.0133,  0.0096,  0.0311, -0.0112,  0.0109,  0.0062,
         0.0353,  0.0195, -0.0243, -0.0194, -0.0007,  0.0222,  0.1756,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.3729e-02, -5.5914e-01,  1.6073e-03, -2.6274e-04, -1.7626e-02,
         7.7144e-03,  6.7362e-02,  7.2817e-03,  1.5755e-02,  2.0990e-02,
         6.6004e-03,  5.0376e-02, -1.2480e-02,  5.3890e-02, -2.3970e-02,
         1.1734e-02, -1.5913e-02,  6.3242e-02, -3.3055e-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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1148,  0.6013,  0.0146, -0.0082, -0.0076, -0.0279, -0.0170, -0.0164,
        -0.0421,  0.0230,  0.0341,  0.0121,  0.0148, -0.0065, -0.0132,  0.0279,
         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,
         0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.2705e-02, -5.2269e-01, -2.4491e-02, -1.0820e-02,  2.5532e-02,
         1.3233e-02,  1.3087e-02,  2.7986e-02, -2.3691e-03,  2.4658e-03,
        -1.0272e-02,  1.1592e-02, -7.9003e-03,  3.1544e-03,  5.2036e-03,
         1.4074e-02,  5.4980e-02,  6.4758e-03,  4.1238e-03,  2.3890e-03,
         6.5007e-03, -1.8720e-03,  2.4836e-02, -1.8097e-03, -1.5483e-03,
         4.7951e-03, -7.7681e-03,  4.5116e-03, -4.7958e-03,  6.3393e-03,
        -7.6966e-03,  4.5168e-03,  1.1411e-02, -6.3097e-03, -1.1123e-02,
         7.7398e-03, -2.5859e-03,  3.5281e-04,  5.5932e-03, -4.2906e-03,
        -1.5285e-02,  6.0959e-03,  5.7577e-03, -4.3721e-03, -5.8350e-03,
        -5.8982e-04, -2.7189e-03,  5.6776e-03, -1.0955e-02, -3.0777e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0659,  0.5982,  0.0779, -0.0216,  0.0145, -0.0090, -0.0312,  0.0171,
         0.0293, -0.0127, -0.0479, -0.0120,  0.0017,  0.0248,  0.0214, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0042, -0.6877, -0.0504, -0.0268, -0.0217, -0.0159,  0.0086, -0.0171,
         0.0011,  0.0119,  0.0179, -0.0045, -0.0032,  0.0076,  0.0142,  0.0143,
        -0.0482, -0.0081, -0.0145, -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,  0.0000,  0.0000,
         0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.4339e-02, -4.8727e-01, -1.1508e-01,  1.4284e-02,  1.0741e-02,
         1.1540e-02, -1.1426e-02, -3.6226e-02, -2.2285e-02, -1.3343e-02,
         2.6513e-02,  1.9673e-02, -1.6426e-02, -1.1937e-04,  4.1060e-03,
        -4.1211e-03, -8.9537e-03,  1.3188e-03, -9.2081e-04,  1.8882e-02,
         1.1813e-02, -6.2946e-03, -1.5696e-03, -3.9406e-04,  5.2423e-03,
         8.7082e-02,  4.0037e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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, -0.5888, -0.0025, -0.0218, -0.0703, -0.0174,  0.0401,  0.0474,
         0.0172, -0.0319, -0.0139,  0.0078,  0.0207, -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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  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, -0.4701,  0.0019, -0.0605, -0.0042, -0.0056, -0.0088,  0.0054,
         0.0345,  0.0242, -0.0354, -0.0098, -0.0506, -0.0576,  0.0900,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.0926e-01,  5.1612e-01, -4.0351e-02, -4.5056e-02,  3.8799e-02,
         2.3521e-02,  5.5787e-03, -3.4768e-03,  7.8469e-04,  9.2191e-03,
         1.0594e-02, -7.7387e-04,  3.8527e-03,  1.3005e-02,  5.1830e-03,
         6.6488e-03,  1.8522e-02, -7.2758e-03, -2.2983e-02, -2.1447e-03,
        -1.2824e-04,  2.3043e-02,  2.2895e-03, -4.8721e-03, -6.4002e-03,
         1.5952e-02, -6.4164e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.5915e-03,  4.5917e-01,  3.9041e-02, -2.7405e-02,  2.0456e-02,
         1.0436e-03,  1.0354e-02,  1.5660e-02,  1.1030e-01,  1.2555e-02,
         7.2089e-03,  1.5102e-02,  3.2843e-03,  2.2258e-02,  4.2796e-03,
         4.8838e-02,  4.2464e-02,  3.3137e-04,  1.6083e-02,  5.5344e-02,
        -1.5233e-02, -1.2740e-03,  1.2134e-02,  1.3343e-02,  2.2596e-02,
        -2.1648e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.2852e-02, -5.6415e-01, -2.2333e-02,  8.2021e-03, -2.1006e-02,
         1.9281e-02,  4.1738e-03, -8.6914e-04, -3.5759e-03,  1.2011e-02,
        -1.6024e-02,  1.0211e-02, -1.6602e-02,  1.1470e-02,  7.7554e-03,
        -8.9321e-03,  5.9194e-03, -1.1240e-02, -3.7563e-04,  7.2608e-03,
         1.3591e-02,  3.2008e-02, -1.9227e-02,  3.9729e-03,  1.2895e-02,
         6.4761e-03, -2.0083e-02, -1.1847e-02, -1.6367e-02,  6.3530e-02,
        -2.5754e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  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([-2.0385e-02, -6.3120e-01,  4.1096e-02,  2.5845e-02,  1.6417e-02,
         1.2936e-02,  6.4454e-02, -7.4977e-03,  3.2550e-02, -9.3245e-06,
         2.4320e-02, -3.5196e-02, -9.7902e-03,  1.4317e-02,  9.3479e-03,
        -3.3312e-02,  1.0600e-03,  1.3908e-03, -1.8877e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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,  0.6955,  0.0886, -0.0179, -0.0370,  0.0041, -0.0056,  0.1099,
         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([ 0.0618, -0.5708, -0.0574,  0.0516,  0.0388, -0.0174,  0.0222,  0.0071,
         0.0038,  0.0312, -0.0069, -0.0016,  0.0120,  0.0063,  0.0191, -0.0193,
        -0.0075,  0.0112,  0.0068,  0.0169, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0179,  0.6345,  0.0128,  0.0193, -0.0110, -0.0151, -0.0290, -0.0034,
        -0.0253,  0.0101,  0.0008, -0.0162, -0.0134,  0.0279,  0.0015, -0.0050,
        -0.0285, -0.0137,  0.0032, -0.0052, -0.0033, -0.0023,  0.0099, -0.0251,
         0.0072,  0.0293,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0561,  0.5327,  0.0274,  0.0044, -0.0050, -0.0075, -0.0280,  0.0425,
        -0.0034, -0.0190,  0.0102,  0.0135, -0.0038,  0.0013, -0.0238,  0.0028,
         0.0131,  0.0059,  0.0042,  0.0062, -0.0083, -0.0044, -0.0085, -0.0025,
         0.0178,  0.0011, -0.0144, -0.0058, -0.0070,  0.0265,  0.0075, -0.0042,
         0.0042, -0.0116, -0.0036, -0.0064, -0.0029,  0.0185,  0.0339,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0695, -0.5263,  0.0116,  0.0165,  0.0231,  0.0012, -0.0089, -0.0045,
        -0.0472, -0.0985, -0.0149, -0.0220,  0.0037,  0.0146, -0.0064, -0.0357,
        -0.0084,  0.0285, -0.0245,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.8673e-02, -4.4092e-01, -4.7274e-02,  2.2408e-02, -4.7518e-02,
        -1.1676e-02,  1.2063e-02,  8.7520e-02,  1.7678e-02, -1.7904e-02,
        -2.3189e-02, -4.7078e-03, -1.9088e-02, -1.9665e-02, -3.7290e-02,
        -2.7439e-03,  2.8620e-02,  3.8501e-03, -6.7239e-03,  6.6338e-05,
         3.7855e-02,  4.8889e-02, -1.3682e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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, -0.3637,  0.1097,  0.0184, -0.0018, -0.0256, -0.0157, -0.0183,
        -0.0053,  0.0134, -0.0155, -0.0103, -0.0038, -0.0069, -0.0060, -0.0431,
        -0.0149,  0.0167,  0.0015, -0.0007,  0.0129,  0.0048,  0.0097, -0.0246,
         0.0315,  0.0106,  0.0124,  0.0222,  0.0036,  0.0271,  0.0073,  0.0095,
        -0.0047,  0.0070,  0.0179, -0.0022,  0.0322,  0.0103, -0.0134,  0.0248],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0174, -0.5858, -0.0130, -0.0318, -0.0186,  0.0114,  0.0227,  0.0174,
        -0.0039, -0.0146,  0.0128,  0.0102,  0.0130,  0.0036,  0.0049,  0.0008,
         0.0007,  0.0072,  0.0092, -0.0008,  0.0135,  0.0085, -0.0050, -0.0089,
         0.0071,  0.0165,  0.0227, -0.0011, -0.0040,  0.0068, -0.0043,  0.0087,
         0.0096,  0.0186,  0.0231, -0.0419,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0230,  0.5446,  0.0065, -0.0371,  0.0031, -0.0071,  0.0181,  0.0068,
        -0.0262,  0.0054,  0.0162,  0.0184,  0.0048, -0.0092, -0.0020,  0.0100,
         0.0071, -0.0413, -0.0197,  0.0052, -0.0090,  0.0024,  0.0073,  0.0175,
         0.0073, -0.0074,  0.0021,  0.0321, -0.0114, -0.0089, -0.0131,  0.0696,
         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, -0.7421,  0.0426, -0.0186,  0.0201, -0.0149, -0.0226, -0.0360,
         0.0087,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1251,  0.0629, -0.1254, -0.0201, -0.0215, -0.0460, -0.0285,  0.0110,
         0.1252,  0.0051,  0.0421, -0.1364, -0.2508,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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([ 0.0815,  0.4674, -0.0024, -0.0329, -0.0187,  0.0581, -0.0535,  0.0133,
        -0.0024,  0.0230,  0.0484,  0.0051, -0.0537, -0.0676, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.3723e-02, -4.8294e-01, -4.7075e-02, -4.4593e-02, -2.5172e-02,
        -2.3662e-02, -2.6519e-03,  1.4842e-02,  3.4134e-03,  3.7530e-03,
         3.2039e-02,  7.7614e-03,  7.4177e-03,  9.6337e-03,  1.2996e-03,
         8.2622e-05,  7.1000e-04,  9.1544e-03,  2.3977e-02,  6.7937e-03,
        -9.2759e-03,  1.5111e-02, -2.4469e-02, -6.3693e-03,  4.3606e-03,
         6.0707e-03,  2.2180e-03, -9.2904e-03,  9.6361e-04, -1.3024e-02,
         1.7336e-02, -3.6331e-02, -3.4489e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1064, -0.6700,  0.0089,  0.0358, -0.0213,  0.0979, -0.0214, -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,  0.0000,
         0.0000,  0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.7390e-02, -5.2006e-01, -9.0354e-02, -5.2493e-03, -1.6551e-02,
         4.7979e-03,  2.5972e-02,  4.5624e-03,  7.5458e-03,  1.1784e-02,
        -7.3393e-03,  3.6322e-03,  2.5793e-03, -1.2160e-02, -1.1865e-04,
         3.4158e-02,  5.7214e-03,  4.8039e-03, -8.2245e-03, -7.3376e-04,
         9.2826e-03,  7.3516e-03,  7.3278e-03,  2.3314e-02,  7.2660e-03,
         1.1953e-02, -8.8749e-03,  5.9421e-03,  2.7397e-02, -5.7559e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+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, -0.4816, -0.1275, -0.0732,  0.0082, -0.0185, -0.0040,  0.0047,
         0.0069, -0.0030, -0.0007, -0.0394, -0.0210, -0.0060,  0.0462,  0.0594,
         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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0813, -0.4221,  0.0655,  0.0347, -0.0063, -0.0613, -0.0026,  0.0138,
         0.0084, -0.0052, -0.0262,  0.0014,  0.0186, -0.0031,  0.0685,  0.0090,
        -0.0203,  0.0045,  0.0087, -0.0379, -0.0241, -0.0512,  0.0223, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0161,  0.5035,  0.0501,  0.0282, -0.0104, -0.0036,  0.0439, -0.0108,
         0.0295, -0.0053,  0.0083,  0.0136,  0.0075, -0.0077,  0.0181, -0.0080,
        -0.0407, -0.0012, -0.0061,  0.0120, -0.0045,  0.0024,  0.0047,  0.0089,
         0.0011,  0.0104,  0.0037, -0.0054, -0.0166,  0.0081,  0.0031,  0.0052,
         0.0011, -0.0098, -0.0126,  0.0025, -0.0260,  0.0103,  0.0040,  0.0101,
         0.0189, -0.0060,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.4750e-02,  4.4306e-01,  1.5566e-01,  2.1502e-02,  2.1810e-02,
        -3.0557e-02, -1.6853e-02,  4.2849e-02,  2.2313e-02, -9.3016e-03,
        -4.1806e-04,  1.0940e-02, -6.7413e-03, -1.3681e-02,  5.4588e-03,
         2.1942e-03,  1.3685e-03, -6.1130e-03, -2.9635e-03,  5.3779e-03,
        -5.8857e-03,  6.8030e-03,  6.2123e-04, -1.9404e-02,  5.3660e-04,
         9.4065e-04,  5.6685e-03,  9.3949e-03,  2.2573e-03,  1.8037e-02,
         3.1599e-03, -1.5040e-02, -3.3397e-03, -8.1259e-03, -1.3841e-02,
        -8.7837e-03, -1.3019e-03, -1.2945e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0770, -0.6662,  0.0302,  0.0017,  0.0525, -0.0232, -0.0534, -0.0130,
        -0.0309,  0.0015, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0518,  0.5965,  0.0335, -0.0192,  0.0074, -0.0262, -0.0186,  0.0009,
        -0.0092, -0.0036, -0.0476, -0.0162, -0.0007,  0.0039,  0.0119,  0.0036,
         0.0117,  0.0095,  0.0072, -0.0276,  0.0007,  0.0030, -0.0044,  0.0081,
        -0.0060,  0.0008,  0.0021, -0.0137,  0.0147, -0.0024,  0.0038,  0.0188,
        -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.4494e-02,  4.7023e-01,  1.4188e-03,  8.6357e-03,  1.3120e-02,
        -6.5744e-03, -7.2523e-03,  2.2046e-02, -5.9402e-03, -4.8675e-02,
        -2.7648e-02, -2.0342e-02, -1.3360e-02, -1.2789e-02,  7.9861e-03,
         1.3008e-02, -1.0900e-02, -8.0704e-03,  1.1508e-02, -1.4848e-02,
        -7.9731e-03, -1.2569e-03, -1.2390e-03,  1.1640e-02,  7.3547e-03,
         5.8766e-03,  2.6695e-03,  4.2696e-03,  3.6061e-03, -4.3388e-03,
        -4.4722e-03,  1.3311e-03, -2.1480e-02,  1.1157e-02,  1.1451e-02,
         1.8044e-02,  1.8393e-02,  1.6604e-03,  3.9048e-03,  2.2415e-03,
         2.9446e-02,  9.1680e-03, -3.6627e-04, -8.6971e-03,  7.7780e-03,
        -3.8611e-03,  6.6755e-04, -2.4061e-03, -4.7391e-03, -1.9410e-03,
        -1.7728e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0057e-01,  5.5865e-01, -1.1430e-03, -1.2121e-02,  3.7315e-03,
        -2.1299e-02, -2.7571e-03,  1.3885e-02, -1.4759e-02, -1.0276e-02,
         1.2312e-03, -2.9329e-03, -1.2732e-02, -5.4687e-03, -4.3163e-04,
        -1.4057e-03, -5.3875e-03,  2.0834e-03, -5.4220e-03, -1.4762e-03,
        -5.6473e-03, -2.5063e-02, -1.4782e-02, -3.9843e-03, -1.2124e-02,
         3.4536e-03, -2.4165e-02,  7.0256e-03, -4.2392e-03, -1.0422e-02,
         4.7727e-03, -2.0665e-03, -6.4818e-03, -2.4612e-02, -3.9491e-03,
        -2.0988e-02,  4.8466e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         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([-4.1445e-02, -5.1987e-01,  5.2322e-02, -4.3310e-02,  7.6326e-03,
         1.3372e-03,  6.7703e-03,  1.7970e-02,  3.3648e-03,  4.0168e-04,
         2.9182e-02,  1.0515e-02, -1.5275e-02, -1.1681e-02, -6.9230e-03,
        -4.0031e-03,  1.0269e-02,  9.8051e-03, -2.3199e-02,  5.5453e-03,
        -1.3817e-02, -8.8163e-03,  7.9080e-03,  1.4267e-02, -4.3100e-03,
        -1.3057e-03, -5.6998e-03, -3.3568e-03, -1.3997e-02, -3.6540e-03,
        -5.0233e-03, -3.7558e-03,  9.7099e-03,  7.2112e-03,  1.4295e-02,
        -6.2051e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1552e-02,  4.3184e-01,  1.8998e-01,  6.5069e-02, -1.2978e-03,
        -7.1701e-03,  1.4350e-03,  2.3264e-03,  1.8576e-02, -1.2456e-02,
        -9.0082e-03, -5.0695e-04,  1.4469e-03, -8.5537e-03,  2.5071e-04,
        -1.0235e-02,  3.8758e-03,  7.4279e-04,  9.3043e-04, -9.1042e-03,
        -6.1798e-03, -8.0526e-03,  6.5331e-04, -3.0160e-03, -7.3689e-03,
        -1.9428e-03, -2.4969e-03,  3.8685e-03, -1.3117e-02,  2.3955e-03,
         2.1215e-03, -1.1258e-02,  1.9236e-03, -5.7390e-03,  2.4556e-03,
        -3.2206e-03, -3.9886e-03,  7.3969e-03, -1.1308e-03,  2.3035e-04,
        -1.5156e-03,  2.0864e-02, -1.2451e-02, -4.4188e-03,  4.3014e-03,
         5.8659e-03, -4.9946e-03, -9.0407e-03, -7.4037e-03,  6.1479e-03,
         8.1613e-03, -8.4023e-04,  1.0240e-02,  1.9383e-03,  2.6901e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0169, -0.4273, -0.0553, -0.0145, -0.0167, -0.0047,  0.0078, -0.0115,
        -0.0239, -0.0018,  0.0079,  0.0108,  0.0437,  0.0231,  0.0624,  0.0246,
         0.0059, -0.0070,  0.0408,  0.0237,  0.0271, -0.0257,  0.0397,  0.0020,
         0.0016, -0.0737,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  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,  0.6012,  0.0955, -0.0269, -0.0063, -0.0139,  0.0288,  0.0051,
         0.0127, -0.0081, -0.0117, -0.0138, -0.0168, -0.0144, -0.0100, -0.0068,
         0.0308, -0.0060, -0.0254,  0.0187,  0.0042, -0.0105,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0304,  0.6811, -0.0127, -0.0277, -0.0035, -0.0252,  0.0162, -0.0058,
        -0.0099, -0.0160,  0.0020, -0.0013,  0.0087,  0.0059,  0.0126,  0.0084,
        -0.0011, -0.0103,  0.0092,  0.0086, -0.0047, -0.0016, -0.0110, -0.0139,
         0.0078, -0.0149, -0.0014,  0.0343,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1310, -0.5536, -0.1042, -0.0395, -0.0720, -0.0300, -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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  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, -0.5628,  0.0903,  0.0016, -0.0100,  0.0282,  0.0279,  0.0205,
        -0.0070,  0.0202, -0.0048, -0.0504,  0.0010, -0.0113, -0.0329,  0.0184,
         0.0076, -0.0836,  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,  0.0000,  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, -0.5526, -0.0397, -0.0389, -0.0289,  0.0271,  0.0074,  0.0344,
        -0.0412, -0.0099, -0.0036, -0.0038,  0.0153, -0.0378, -0.0244,  0.0029,
         0.0077,  0.0010, -0.0018, -0.0007, -0.0050,  0.0113,  0.0025,  0.0108,
         0.0026, -0.0020,  0.0432, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0756, -0.4790,  0.0030, -0.0443,  0.0069,  0.0116,  0.0188,  0.0046,
        -0.0294,  0.0077,  0.0082,  0.0050,  0.0021,  0.0178,  0.0170,  0.0108,
        -0.0084,  0.0256,  0.0316,  0.0039,  0.0120,  0.0134,  0.0032,  0.0198,
         0.0247,  0.0033,  0.0035,  0.0055,  0.0412, -0.0288, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1052, -0.4152, -0.0224, -0.0267, -0.0215,  0.0265, -0.0304,  0.0810,
        -0.0034, -0.0124,  0.0067,  0.0174,  0.0134, -0.0453,  0.1725,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0921e-01,  5.8153e-01, -3.7241e-02,  1.0231e-02,  7.9506e-03,
         2.8823e-02, -3.9732e-02, -5.7948e-02,  1.6379e-02, -5.1072e-03,
         1.4001e-02, -2.1174e-04, -9.9684e-03,  6.9339e-03,  8.8720e-03,
         2.7030e-02,  1.2635e-02,  9.3011e-03,  1.6892e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 6.6826e-03,  6.3743e-01,  7.3584e-03, -4.2524e-02,  2.5375e-03,
        -3.3503e-04, -2.6574e-02, -2.7227e-02, -6.5682e-03,  6.9789e-03,
         7.4788e-03,  1.1918e-03,  1.0402e-02,  9.1079e-03, -9.8187e-03,
         2.4301e-02,  4.5936e-02,  8.6379e-04, -5.5732e-04, -1.3779e-02,
         1.5290e-02,  2.2313e-03,  9.0008e-03, -1.9714e-03, -8.3697e-04,
         5.7400e-03, -1.1486e-02,  7.1033e-03, -2.8147e-02,  3.0541e-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,  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([-0.1309, -0.4372,  0.0346,  0.0125,  0.0154,  0.0200,  0.0534,  0.0056,
         0.0543,  0.0388,  0.0461,  0.0074,  0.0123, -0.0065, -0.0079, -0.0111,
         0.0111,  0.0105, -0.0009,  0.0085,  0.0628, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.6970e-03, -7.6340e-01, -1.6017e-02, -6.3998e-04, -1.0681e-02,
        -6.3984e-03,  1.0089e-02, -3.0087e-03, -6.2912e-03,  1.7381e-03,
         5.5893e-02,  1.9628e-02,  2.3105e-02,  4.5480e-03, -1.8430e-02,
         2.4482e-03,  2.3411e-03,  1.3516e-02, -8.3535e-03, -3.0778e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0157,  0.6802,  0.0131,  0.0324,  0.0424,  0.0035, -0.0258,  0.0127,
        -0.0029,  0.0052, -0.0193, -0.0134, -0.0050, -0.0197,  0.0100, -0.0032,
        -0.0095,  0.0015,  0.0065, -0.0075,  0.0025,  0.0019,  0.0015,  0.0078,
         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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0278, -0.6037,  0.0199, -0.0066,  0.0239,  0.0138, -0.0091,  0.0074,
        -0.0057,  0.0317, -0.0266,  0.0359,  0.0172, -0.0102, -0.0086, -0.0062,
        -0.0188,  0.0152, -0.0085,  0.0132,  0.0062, -0.0117, -0.0121,  0.0058,
        -0.0054, -0.0015,  0.0042, -0.0303, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1344, -0.5318,  0.0757, -0.0164,  0.0031,  0.0345, -0.0420,  0.0054,
        -0.0076,  0.0090, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.3547e-02, -5.1459e-01,  1.6455e-02,  3.1833e-04,  9.9697e-03,
        -4.9961e-03,  2.3904e-02, -1.8917e-02, -3.0083e-02, -1.2740e-02,
         5.7664e-03, -2.4012e-02, -1.4962e-02, -2.7447e-02, -7.6943e-02,
        -4.8210e-03, -2.2197e-03, -1.1010e-02, -2.7043e-02,  8.2352e-03,
        -1.9563e-02, -6.5188e-03, -5.0099e-02,  3.5842e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.0128e-02,  5.2477e-01,  4.3459e-02, -4.8801e-03,  4.0458e-03,
        -3.0162e-03,  1.4732e-02,  1.0842e-02, -5.0240e-03,  9.5436e-03,
         8.4943e-03,  1.1543e-02,  1.3105e-02,  4.3623e-04,  7.7885e-03,
        -3.7451e-03, -4.9808e-03,  5.2183e-05,  3.3868e-03,  3.0102e-03,
         5.0454e-03, -6.0816e-03, -4.1982e-03,  2.3807e-02,  1.1907e-02,
        -3.4238e-03,  1.1766e-02,  1.0485e-02,  1.7267e-03, -6.0191e-03,
        -6.4384e-04, -8.6188e-03, -3.1825e-03, -7.8924e-03,  2.2163e-03,
         3.9064e-03, -1.8629e-03,  6.8319e-03,  3.3298e-03, -5.0997e-03,
         2.4195e-03,  4.1337e-03,  6.6054e-03, -5.6452e-03, -4.2204e-04,
        -2.1031e-02,  2.9709e-02,  6.5003e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0560,  0.5910, -0.0023, -0.0168, -0.0324, -0.0139, -0.0448, -0.0211,
         0.0588,  0.0026, -0.0085, -0.0173, -0.0207, -0.0162,  0.0074, -0.0035,
         0.0866,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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,  0.5287, -0.0368,  0.0088,  0.0108,  0.0406, -0.0697,  0.0166,
         0.0467,  0.0634,  0.0007, -0.0368, -0.0240, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.5010e-02,  6.4604e-01,  1.9154e-02, -3.6620e-02, -8.8536e-03,
         4.0800e-03, -2.9321e-03, -7.5186e-03, -1.0621e-02,  3.5468e-02,
         5.5918e-03, -3.6968e-03, -5.6401e-03, -3.9621e-03, -5.4725e-03,
        -1.7047e-03,  4.6283e-03,  1.9651e-03,  1.0802e-02,  2.1356e-02,
         1.1484e-03,  2.5318e-03, -6.5060e-03,  3.0022e-03, -8.5971e-03,
        -4.8553e-03,  5.2039e-03, -7.6144e-03,  8.4256e-04, -6.7513e-03,
        -2.2182e-04, -3.5675e-03,  1.2668e-02,  1.5369e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.5656e-02, -5.4795e-01,  1.4317e-01, -3.1905e-02, -9.2560e-03,
         5.2572e-03,  7.7970e-03, -2.2045e-02,  1.2762e-02,  6.5952e-03,
         5.2709e-04,  5.0568e-03, -4.4095e-03, -2.3137e-03,  1.5291e-04,
         1.1914e-02,  1.5640e-03,  8.1012e-03,  2.2442e-03, -1.0589e-02,
         1.1013e-02,  3.9648e-04, -1.4206e-03, -4.2094e-04, -4.5709e-03,
         1.9594e-03,  9.8796e-05, -2.6292e-03, -4.2121e-03,  8.9545e-03,
         9.1341e-04, -6.2468e-04,  3.1609e-02, -2.1257e-04,  3.0111e-03,
        -3.2086e-03,  6.8203e-03,  6.5502e-03,  3.9207e-04,  8.4869e-03,
         4.2365e-04, -2.8278e-02, -4.5236e-03,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.0616e-02,  5.7649e-01,  4.9000e-02,  3.3794e-03,  2.2823e-03,
        -3.6058e-03, -4.4873e-03, -1.3259e-02, -2.5939e-02, -1.6242e-02,
        -2.0263e-02,  4.6091e-04,  1.3622e-02, -8.4771e-03, -1.0613e-02,
        -6.3509e-03,  1.2832e-02, -3.1247e-03, -9.9504e-03,  3.2221e-03,
         1.0890e-02,  1.3863e-02, -9.4723e-03, -1.6466e-02, -2.6594e-02,
         6.9072e-03,  7.6642e-03,  3.0719e-03, -7.4579e-03,  3.8814e-03,
        -4.2807e-03,  2.4577e-03, -5.9433e-03,  5.0382e-03,  3.1800e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  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([-0.0597, -0.5072, -0.0269,  0.0492, -0.0009,  0.0388,  0.0089,  0.0394,
         0.0108, -0.0106, -0.0194, -0.0146, -0.0101, -0.0445,  0.0602, -0.0039,
         0.0054, -0.0120, -0.0101, -0.0057, -0.0259,  0.0016, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.4226e-04, -5.4575e-01, -7.7467e-02,  2.5182e-02,  3.7742e-02,
        -1.4767e-02,  2.8540e-02,  9.6301e-03,  7.1915e-04,  4.2731e-03,
         9.1118e-03,  2.8749e-03, -9.2783e-03, -1.7663e-03,  8.7927e-03,
         6.1134e-03,  3.8439e-03,  4.3007e-03,  7.9540e-02,  2.5443e-02,
        -1.2461e-02,  9.7509e-03,  4.2682e-03, -2.1325e-03, -1.5712e-02,
         1.2027e-03,  7.7319e-03, -4.0209e-03, -6.9709e-03,  1.6278e-02,
        -4.4518e-03, -4.8644e-03,  2.4602e-03, -7.4521e-03, -3.0530e-03,
        -1.6097e-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([ 0.1137,  0.5313,  0.1422,  0.0112,  0.0209, -0.0358, -0.0040,  0.0277,
        -0.0100,  0.0032,  0.0154, -0.0007,  0.0194, -0.0194, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0585, -0.6970,  0.0024, -0.0009,  0.0115,  0.0345,  0.0367, -0.0033,
         0.0173,  0.0118,  0.0512,  0.0208, -0.0025, -0.0183, -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,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.8980e-02, -4.0774e-01, -6.0638e-02,  1.2852e-02, -1.2136e-02,
        -4.6402e-03,  2.3862e-02,  7.6680e-03,  6.5953e-03,  1.1930e-03,
         1.4626e-02, -2.6157e-03,  2.8683e-02, -2.7740e-03, -1.4031e-02,
        -2.5118e-04, -7.8471e-03, -4.3859e-03, -2.3760e-03, -7.5018e-03,
        -4.0509e-03, -5.6282e-03, -9.7202e-03,  4.1365e-02, -2.2280e-02,
         3.3632e-03, -3.3678e-03,  9.1189e-03,  3.1700e-02, -9.9012e-03,
        -3.6659e-03, -1.6159e-02, -2.1031e-02, -1.2427e-02, -4.2846e-03,
        -1.2170e-02,  7.2202e-03, -2.4482e-03, -2.1270e-03,  1.5954e-03,
        -4.3224e-05, -1.4644e-02, -1.0875e-03, -2.4695e-03,  1.7781e-02,
        -4.5072e-03, -3.0084e-03, -1.3261e-03, -7.1058e-03,  6.4033e-03,
         2.0395e-02,  4.6210e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0726, -0.6468,  0.0278,  0.0193,  0.0334,  0.0357,  0.0086,  0.0064,
         0.0025,  0.0118, -0.0066, -0.0077, -0.0026,  0.0170,  0.0160, -0.0046,
         0.0150, -0.0125,  0.0033, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.6212e-02,  5.9844e-01,  1.2418e-01,  1.7307e-02,  4.8786e-04,
        -1.2734e-02, -1.3960e-02,  1.0762e-02, -4.8296e-03, -1.3072e-02,
         1.0072e-02,  1.2733e-02,  1.4946e-02,  8.1852e-03, -2.5055e-04,
        -1.6800e-03, -3.7474e-03, -3.0770e-03, -2.4274e-02, -2.3047e-03,
        -1.2314e-02, -7.4438e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0458,  0.6267,  0.0559,  0.0213,  0.0028,  0.0135,  0.0158,  0.0186,
         0.0264,  0.0210, -0.0172, -0.0021, -0.0160,  0.0017, -0.0125,  0.0098,
        -0.0141, -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,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.7039e-02, -7.0314e-01, -1.5631e-02, -2.9018e-02,  8.4156e-03,
        -2.6181e-02,  1.1146e-02,  4.1857e-02,  3.2402e-03, -9.3475e-03,
         1.8418e-03, -1.6269e-02,  2.3913e-03,  4.8716e-05,  3.3444e-03,
        -1.0557e-02, -2.5809e-02, -1.7197e-03, -7.2596e-04,  7.8838e-04,
         4.0231e-03,  8.2208e-03, -2.1617e-02,  3.5969e-03, -2.3924e-02,
         1.0112e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.6732e-02, -6.3378e-01, -4.1666e-02,  1.7325e-02, -2.4997e-03,
        -1.9170e-02,  4.6161e-05, -8.7229e-03, -2.3522e-02, -5.4054e-02,
        -1.1516e-02,  1.0324e-02, -3.6699e-02,  4.5802e-03, -3.0071e-02,
         1.2596e-02, -1.1872e-02, -2.8898e-02, -4.5670e-03, -2.1633e-02,
         9.7253e-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([-7.6884e-03, -4.9628e-01,  5.4268e-02, -1.2053e-02,  8.4727e-03,
         1.9870e-02,  1.9845e-03,  2.6094e-02, -5.6073e-03, -2.2579e-02,
        -7.9571e-04, -2.0009e-02,  1.1821e-03,  1.3223e-02, -1.1160e-02,
         1.3575e-03,  1.1238e-02,  4.7702e-04, -8.0737e-03,  2.8470e-03,
         9.4171e-03, -5.6879e-03, -9.5454e-03, -4.0276e-03, -1.3922e-02,
         2.0631e-02,  3.0035e-03, -6.1553e-03, -2.7500e-02,  1.2788e-02,
        -2.9850e-02,  5.0565e-03,  3.8791e-03, -5.7871e-03,  5.5558e-03,
        -8.4655e-03,  1.9291e-03,  7.3138e-03,  2.1293e-02,  4.1142e-03,
        -5.3999e-03, -4.8239e-03,  7.5067e-04, -2.2284e-03, -1.9487e-03,
         5.3594e-02, -7.4990e-05,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1950, -0.4907,  0.0820,  0.0509,  0.0326,  0.0201,  0.0078,  0.0592,
         0.0559, -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], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #500: [tensor([-3.3494e-02, -6.3400e-01, -3.2922e-02, -3.8797e-02, -3.9230e-03,
         8.7305e-03,  3.1732e-03,  1.9215e-03,  3.6977e-03,  1.4526e-02,
        -6.4402e-03, -4.1254e-03,  1.9602e-03,  5.7832e-03, -5.3764e-03,
         6.8984e-03,  3.6415e-03,  1.0581e-03, -3.2359e-04,  4.5470e-03,
        -1.2039e-03,  3.1248e-02,  6.9210e-03,  6.8885e-03, -8.5614e-03,
        -5.7893e-04, -8.0324e-03, -4.0401e-03, -6.8690e-03,  2.6374e-04,
         8.2385e-03,  6.1115e-03, -9.1129e-03, -4.5632e-03, -3.7017e-03,
         4.9324e-03,  5.9954e-04,  1.9012e-03, -6.9406e-03, -8.2348e-03,
         5.0175e-03, -1.4155e-03,  1.1630e-03,  1.8916e-03, -1.6035e-02,
         1.8466e-02, -1.1734e-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([ 5.1122e-02, -5.5430e-01, -3.3252e-02,  7.8963e-02, -3.2531e-03,
         9.5863e-03,  3.0512e-02,  5.4254e-02,  9.5690e-03, -8.3367e-05,
         4.5270e-05,  3.1794e-02,  1.0690e-02,  1.7066e-02,  3.0716e-02,
        -7.6276e-03,  7.6311e-03, -3.1901e-03,  8.7785e-03,  1.1077e-03,
         1.4664e-02,  8.5913e-03, -2.8270e-03, -1.0872e-02, -9.6024e-03,
         9.8995e-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([-6.3100e-03, -4.9441e-01, -5.8108e-02, -4.6645e-03, -8.1798e-03,
        -2.9087e-02, -8.9623e-03, -6.8966e-03,  1.1198e-02,  4.8299e-02,
         2.2190e-02, -3.7348e-03, -8.3204e-03, -1.4530e-02, -9.2200e-03,
         8.1010e-03, -1.0307e-02, -1.1231e-02, -3.9410e-05,  3.2491e-03,
         4.2283e-03,  4.4720e-02, -2.1722e-03, -1.5535e-04,  4.5952e-03,
         7.3864e-03, -7.8401e-03, -2.8592e-02,  2.1805e-03, -7.1683e-03,
         7.3886e-03, -1.2068e-02, -4.5081e-03, -1.2785e-05, -2.1362e-02,
        -4.1586e-02,  3.7003e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1549,  0.4261, -0.1790, -0.0483,  0.0091,  0.0097, -0.0361, -0.0024,
        -0.0158,  0.0359, -0.0073,  0.0011,  0.0383, -0.0142,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.2676e-02, -6.0736e-01, -2.9713e-02, -1.6434e-02, -2.7378e-02,
        -4.4969e-02, -3.4079e-04, -1.3853e-02, -2.8465e-02, -8.7476e-03,
        -1.1118e-02, -2.7512e-02, -1.9248e-02,  3.2437e-02, -3.5306e-02,
         2.4440e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.6764e-02, -5.7297e-01, -3.6545e-02,  2.5006e-02, -4.9045e-04,
         6.0139e-02,  8.9644e-03,  2.3238e-02,  2.0341e-02, -1.4346e-02,
         3.0445e-02,  9.6032e-03, -1.1232e-02, -8.5183e-04,  8.0835e-03,
        -1.3064e-02,  1.7778e-02, -1.0725e-03,  6.3097e-03,  5.2185e-04,
         2.3965e-02,  2.7985e-03,  1.0312e-02,  2.1672e-03,  1.7105e-03,
         1.5653e-02, -4.9522e-03, -2.4204e-03, -4.6614e-02, -1.6427e-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([-0.0587, -0.5650, -0.1063,  0.0246,  0.0011, -0.0015, -0.0111, -0.0087,
        -0.0060,  0.0190,  0.0130,  0.0115,  0.0039, -0.0062,  0.0096, -0.0008,
         0.0087, -0.0099, -0.0013,  0.0043,  0.0023, -0.0018, -0.0056,  0.0125,
        -0.0049,  0.0052, -0.0161, -0.0007, -0.0020, -0.0023,  0.0196, -0.0091,
         0.0169,  0.0048,  0.0144, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-4.9288e-03, -6.4919e-01, -2.5240e-03,  6.9160e-03,  1.8139e-02,
         1.1403e-02,  6.6341e-03,  1.9058e-02,  5.4515e-02, -2.3890e-02,
         1.9831e-03, -1.1383e-03,  2.0005e-02,  1.2661e-03,  1.2379e-03,
        -2.1541e-03,  7.4813e-03, -6.5978e-03,  6.9328e-03,  5.4871e-03,
        -1.0599e-02, -6.8043e-03, -4.4421e-03, -2.7652e-02, -2.3879e-03,
        -5.6108e-03,  1.4380e-02, -1.0017e-03, -1.2713e-02,  4.6690e-04,
         7.5277e-03, -1.6276e-02, -3.5495e-03,  2.9915e-02, -5.1941e-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([ 0.0431, -0.3889,  0.0262, -0.2094, -0.0639, -0.0315,  0.0293,  0.0480,
        -0.0474,  0.0305, -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,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.2453e-01, -5.7348e-01,  9.0174e-03,  2.6078e-02, -5.2334e-03,
        -6.3700e-03,  1.7435e-02, -8.4844e-04,  3.4166e-04, -3.7819e-06,
         1.4958e-02,  1.0974e-03, -3.5584e-02,  4.6233e-03,  5.3975e-03,
         2.5791e-02,  2.0569e-02, -3.2555e-03,  9.0096e-03,  1.0517e-02,
        -1.3681e-02,  2.2275e-02, -1.1789e-02, -3.5931e-02,  2.2185e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0218,  0.5818,  0.0717, -0.0041,  0.0191, -0.0124, -0.0008, -0.0035,
        -0.0031, -0.0139,  0.0033, -0.0019,  0.0012,  0.0076,  0.0221,  0.0174,
        -0.0274,  0.0095,  0.0231,  0.0244,  0.0330,  0.0346,  0.0107, -0.0222,
        -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([ 0.0367,  0.4815,  0.0172,  0.0080,  0.0160,  0.0053, -0.0025,  0.0014,
         0.0066,  0.0300, -0.0115, -0.0118, -0.0033,  0.0072,  0.0013,  0.0018,
        -0.0165, -0.0039, -0.0062, -0.0149,  0.0127, -0.0094, -0.0129,  0.0065,
        -0.0017, -0.0150,  0.0094,  0.0216, -0.0048, -0.0284, -0.0211, -0.0116,
         0.0257, -0.0013,  0.0019, -0.0029, -0.0005,  0.0027,  0.0121,  0.0025,
        -0.0020, -0.0018, -0.0025, -0.0032, -0.0043, -0.0105, -0.0046,  0.0043,
         0.0034, -0.0054,  0.0167, -0.0041, -0.0013,  0.0478], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.0387, -0.6647, -0.0695, -0.0157,  0.0271, -0.0025,  0.0406,  0.0021,
        -0.0205,  0.0181, -0.0019,  0.0168, -0.0283, -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,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.6092e-02, -6.1766e-01, -2.0493e-02,  1.5013e-02, -2.3228e-02,
        -2.9108e-02, -2.5371e-02, -2.6452e-02, -9.2209e-03, -3.9857e-04,
        -1.8819e-02,  1.1269e-02,  8.6281e-03, -4.7820e-03,  7.9628e-04,
         3.8636e-02, -2.5771e-03, -1.1105e-02,  1.2058e-02, -2.3962e-03,
         6.5893e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1424, -0.6174,  0.0089,  0.0308, -0.0212,  0.0078,  0.0758, -0.0432,
         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,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0007, -0.6554, -0.0070,  0.0361, -0.0175,  0.0186, -0.0038,  0.0371,
         0.0100, -0.0158,  0.0194, -0.0106,  0.0125,  0.0230,  0.0027,  0.0025,
         0.0108, -0.0038,  0.0015, -0.0027,  0.0062,  0.0156,  0.0061,  0.0233,
         0.0224, -0.0013,  0.0069,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.5657e-02, -6.6811e-01, -4.0477e-02,  1.4445e-03,  1.7041e-02,
         1.4242e-02,  8.2780e-03,  1.9173e-02,  1.2798e-02,  2.0454e-02,
         6.4936e-03, -1.2660e-02,  1.4066e-04,  6.2755e-03,  4.0098e-03,
        -2.1813e-02, -2.2627e-02, -1.2680e-02,  1.0173e-02,  4.2230e-02,
        -3.1252e-02,  1.9716e-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([ 1.0154e-01,  6.2351e-01, -4.5782e-02,  1.3343e-02, -6.0250e-03,
         1.1926e-02,  1.6489e-04,  3.6885e-03,  1.2490e-03,  1.2568e-02,
         4.5282e-03,  9.2275e-03, -2.0954e-02,  6.7820e-03, -1.3422e-02,
        -1.4180e-02,  6.3595e-03,  2.1557e-03, -5.1192e-03,  2.8723e-03,
        -4.6241e-03,  6.9953e-03, -3.4887e-03,  3.7529e-04,  1.5397e-03,
         7.7228e-03,  7.5608e-04,  2.4704e-03,  9.1936e-03, -9.3465e-03,
         4.4641e-03,  4.3627e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-6.3929e-02,  5.2259e-01,  5.5782e-02,  4.1612e-03,  2.2059e-02,
         1.4786e-02, -5.4079e-04, -6.7855e-03,  4.0695e-04, -2.2549e-02,
         1.0345e-02, -9.0066e-03, -1.5880e-03,  2.2229e-03, -7.1034e-03,
         6.5960e-03, -3.9916e-03,  1.6366e-03,  2.6998e-03, -5.4719e-03,
        -7.2118e-03,  9.5802e-03, -3.1623e-03, -2.0565e-03, -5.7441e-03,
         1.5450e-03,  1.8927e-02,  5.4622e-03, -4.7033e-03,  4.3910e-03,
         1.7022e-02, -8.2847e-03, -1.1072e-02,  2.4569e-03, -2.9412e-03,
         5.2514e-03, -2.7718e-02, -4.4815e-03,  2.6499e-03,  3.7184e-03,
        -2.0799e-02,  7.3534e-04,  6.3500e-02, -2.3395e-03], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0363, -0.6038, -0.0470,  0.0139, -0.0340, -0.0313,  0.0105, -0.0329,
        -0.0040,  0.0454, -0.0240,  0.0159, -0.0122,  0.0159, -0.0022, -0.0403,
        -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.0937e-01,  6.3433e-01, -1.2249e-02,  3.8160e-02,  2.9582e-02,
        -7.9372e-03, -5.9721e-04,  7.8427e-03, -1.4787e-02,  2.4000e-02,
        -1.0595e-02,  1.2073e-05, -8.8939e-03,  1.0786e-02, -1.1433e-02,
        -7.1119e-04,  4.8817e-03, -8.3067e-04,  1.0269e-02,  1.4066e-02,
        -6.9811e-03,  1.9553e-02,  7.4806e-03, -2.9502e-03, -1.1703e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0440, -0.6013, -0.0557, -0.0133, -0.0022,  0.0377, -0.0281,  0.0077,
        -0.0090, -0.0238, -0.0091, -0.0061, -0.0047, -0.0064, -0.0067, -0.0081,
        -0.0225, -0.0087, -0.0167, -0.0369,  0.0425,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0461, -0.6556,  0.0742, -0.0488,  0.0739,  0.0189, -0.0144,  0.0192,
        -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0599, -0.4576,  0.0684, -0.0292, -0.0286, -0.0173,  0.0125, -0.0227,
        -0.0087, -0.0142, -0.0197,  0.0230,  0.0156,  0.0122, -0.0139,  0.0172,
         0.0062, -0.0142,  0.0031,  0.0072, -0.0026, -0.0124,  0.0209,  0.0065,
         0.0067,  0.0142,  0.0158,  0.0059, -0.0052,  0.0049, -0.0158,  0.0120,
         0.0038, -0.0218,  0.0000,  0.0000,  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([-4.1158e-02,  4.5365e-01,  2.2214e-02,  7.7070e-02,  1.1581e-02,
         1.4960e-02, -5.8286e-02,  1.9032e-02, -1.8132e-02, -1.7184e-03,
         1.2923e-03,  1.3537e-02,  5.6478e-03,  1.2447e-02,  1.6934e-02,
        -7.6829e-03,  2.5549e-03,  1.1858e-02,  8.9080e-03,  4.9925e-03,
        -5.8282e-03, -5.0640e-03,  2.8739e-02, -1.1115e-02,  9.0302e-03,
         3.6814e-03,  5.6289e-03,  3.2272e-03, -1.6361e-03,  6.7313e-03,
        -8.3312e-03, -5.0153e-03,  7.3126e-04, -9.0613e-03, -1.7395e-02,
         1.7793e-03, -3.8875e-04, -1.6187e-02, -8.2261e-03,  4.6493e-03,
        -1.8385e-03, -8.0372e-04,  2.1360e-02, -1.9892e-02,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.9273e-02, -6.0063e-01,  4.2975e-02,  3.1663e-02,  4.4324e-03,
        -6.9730e-03, -7.3583e-03, -8.9121e-03, -4.3463e-03,  1.3175e-02,
        -4.4051e-03, -5.4961e-03,  1.0588e-03,  2.6634e-03, -1.4081e-02,
         1.4227e-02,  1.3234e-03,  1.2305e-02,  1.7579e-03,  1.1747e-02,
         2.6313e-02, -9.7293e-05, -4.8039e-02, -3.6743e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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,  0.4794,  0.1854,  0.0928, -0.0079,  0.0201, -0.0090,  0.0025,
        -0.0029, -0.0158, -0.0148,  0.0306, -0.1049,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.1379e-01, -5.7896e-01,  3.6114e-03,  6.0720e-03, -1.7598e-02,
         5.3560e-02, -6.1904e-04,  1.7325e-02, -4.5784e-04,  8.3799e-03,
        -9.0234e-03, -2.6814e-04, -1.8405e-03, -2.8525e-02,  1.0889e-02,
        -1.6215e-02,  3.9957e-02, -5.7607e-04,  8.5990e-04,  3.3304e-03,
         8.7214e-02, -9.2260e-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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0612, -0.6960,  0.0732,  0.0173,  0.0039, -0.0164,  0.0667, -0.0379,
        -0.0148,  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,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.1900e-03, -5.8221e-01,  1.0348e-01, -3.9030e-02, -3.5812e-03,
         3.3697e-03, -9.5748e-03, -6.9862e-03, -1.0397e-02, -2.6194e-02,
         3.4438e-02,  7.1904e-03,  4.2118e-02, -2.3394e-03,  1.2621e-02,
        -7.9711e-03,  6.5668e-03, -1.9088e-03,  5.0400e-04,  1.1590e-02,
         1.9069e-02, -5.7580e-02, -6.0905e-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([-1.0191e-01, -5.7273e-01, -4.9037e-02,  1.3381e-02,  9.3192e-03,
        -6.0495e-04,  3.7313e-03, -3.3311e-03, -1.7874e-03,  3.3330e-03,
         1.1283e-02, -9.5847e-03,  8.7765e-03, -7.9825e-03,  1.6856e-02,
         4.9089e-03,  1.3689e-02, -2.7637e-03,  1.3756e-02,  1.9042e-03,
        -5.5265e-03,  1.1278e-03, -1.0368e-02, -7.2366e-04,  1.5458e-04,
         2.2900e-03,  7.2866e-03,  3.9931e-04, -2.3389e-03,  3.7197e-03,
        -1.4839e-02, -6.0369e-03,  3.5599e-03,  1.0720e-02,  3.3023e-03,
        -1.0623e-02, -6.5359e-03,  3.9578e-03, -1.6126e-02, -6.6583e-03,
        -5.4643e-03, -5.5693e-03, -3.2734e-03,  1.4805e-02, -3.9222e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0254, -0.6173,  0.0207,  0.0522, -0.0011, -0.0055, -0.0162,  0.0246,
        -0.0137,  0.0012, -0.0036, -0.0020, -0.0013,  0.0300,  0.0127,  0.0175,
        -0.0073, -0.0312,  0.0603,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 2.5707e-02,  5.7740e-01, -2.2685e-02, -2.7621e-02,  2.1236e-02,
         7.5325e-03, -1.0222e-03, -3.7072e-03, -1.2121e-02, -2.7762e-04,
        -3.8076e-03,  1.0275e-02, -4.7127e-02,  1.0672e-02, -1.2666e-02,
        -2.5739e-04, -1.4244e-03,  4.0287e-03, -1.1845e-02,  3.8560e-03,
        -2.0674e-02,  1.1869e-02,  4.8011e-02,  4.4572e-03,  1.9832e-02,
        -8.9889e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 9.1651e-02,  7.3289e-01,  5.5609e-03,  1.7679e-02,  8.7081e-04,
        -3.6275e-03,  5.0561e-04,  1.0006e-02,  1.3103e-02, -5.8116e-04,
         4.9308e-03,  2.1152e-02,  1.0841e-02,  1.3441e-02,  2.5524e-03,
         2.7722e-03,  2.7152e-03,  8.2550e-03,  9.3056e-04, -2.8806e-04,
         4.6503e-03,  6.1581e-03,  5.9567e-03,  1.0597e-02,  2.1446e-03,
         2.8702e-03,  1.0659e-03, -2.2209e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.2090, -0.5356, -0.0402, -0.0126,  0.0315,  0.0173, -0.0363, -0.0568,
         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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0620, -0.4996, -0.0148,  0.0132, -0.0434,  0.0047,  0.0069, -0.0112,
        -0.0126, -0.0073, -0.0266, -0.0139, -0.0105,  0.0071, -0.0228,  0.0043,
         0.0205, -0.0497, -0.0199, -0.0300, -0.0043,  0.0222, -0.0047,  0.0041,
        -0.0073,  0.0527,  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], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #100: [tensor([ 0.0172, -0.5871, -0.0519, -0.0168, -0.0012,  0.0078, -0.0101, -0.0089,
        -0.0042,  0.0071,  0.0097, -0.0220, -0.0153, -0.0011, -0.0062,  0.0022,
         0.0022, -0.0023,  0.0017,  0.0091, -0.0222,  0.0094, -0.0195, -0.0082,
        -0.0550,  0.0222,  0.0012,  0.0032,  0.0042,  0.0151, -0.0027, -0.0053,
         0.0058, -0.0107, -0.0025,  0.0008, -0.0134,  0.0147,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.7472e-02,  5.5005e-01,  7.2734e-02,  9.0493e-03, -1.0216e-02,
        -4.9684e-03, -1.2183e-02, -5.8045e-03, -1.9504e-02, -7.5905e-03,
         5.2975e-03, -1.1517e-02, -7.8387e-04,  3.3162e-03,  3.9592e-03,
        -3.9368e-03, -1.6485e-02,  4.0683e-04,  4.6020e-03,  2.5544e-03,
         4.2865e-03,  4.9556e-04,  5.1847e-04,  6.0818e-03, -2.4297e-03,
         4.3929e-03,  3.1634e-03, -5.1581e-03,  1.9560e-03,  1.3797e-02,
         1.2430e-02,  2.1596e-02,  2.1953e-02, -1.2107e-02,  8.1815e-04,
         1.8145e-02,  2.5983e-02,  5.2257e-02,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.3693e-02,  5.3217e-01,  3.8094e-02,  7.4710e-03, -5.9041e-03,
         2.9857e-02, -4.1718e-03,  3.0874e-02,  1.8452e-03,  3.9790e-02,
         2.5132e-03, -6.4716e-05, -1.7055e-02, -1.2515e-02, -6.0118e-03,
         8.6915e-03,  7.6368e-03, -1.2395e-02,  2.4861e-02,  8.5210e-03,
         1.5899e-02,  1.1109e-02, -1.4592e-02, -6.5490e-03,  7.4939e-03,
         1.1766e-02, -8.6968e-03,  4.8395e-02, -6.1394e-03, -6.5225e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.9229e-03, -6.2894e-01,  8.4207e-02,  1.4002e-03,  5.5633e-03,
        -1.1282e-02,  1.9011e-02, -6.8712e-03, -1.1254e-02, -2.5725e-02,
        -5.4175e-03, -5.4813e-03, -4.0842e-03, -7.2453e-03, -4.8628e-03,
        -1.2272e-03, -2.8199e-02,  3.5353e-04, -1.8819e-02, -4.0759e-03,
        -2.2819e-03,  1.9881e-03, -4.3573e-03, -9.0122e-03, -2.3260e-02,
         2.2297e-03,  5.0071e-03, -1.1812e-02, -1.8005e-03, -1.7063e-03,
        -7.0221e-03, -1.3210e-03,  7.3163e-03, -1.5990e-02,  6.4171e-03,
         2.7237e-04,  3.0625e-03,  6.2680e-03, -6.7643e-04,  1.1258e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0456, -0.5065,  0.0545,  0.0120, -0.0117, -0.0048, -0.0070, -0.0119,
        -0.0171, -0.0070,  0.0123, -0.0006,  0.0061, -0.0057, -0.0035, -0.0007,
        -0.0098,  0.0174, -0.0086, -0.0019, -0.0005, -0.0047,  0.0234, -0.0115,
        -0.0022, -0.0146, -0.0099, -0.0028, -0.0032,  0.0238,  0.0315, -0.0021,
         0.0015,  0.0039, -0.0048, -0.0047, -0.0426, -0.0677,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.1449e-02, -6.3784e-01,  5.5556e-02,  2.8772e-02, -9.6406e-03,
         1.3781e-03,  1.4588e-02,  5.9709e-03,  2.1316e-02,  1.0566e-02,
        -4.8419e-03,  1.6087e-02, -1.5762e-03,  1.9094e-02, -1.7982e-02,
        -8.2949e-03, -7.7710e-03,  7.8863e-04, -1.5312e-02, -1.2508e-02,
         6.2872e-05,  4.4137e-02, -3.7013e-02, -1.7460e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.3060e-02, -5.1509e-01,  7.4442e-02, -7.0274e-03, -1.2906e-02,
         1.8619e-02, -8.9241e-03,  6.6387e-03,  3.0425e-03, -6.6222e-03,
        -2.5458e-02, -4.5158e-03, -5.6382e-04,  5.7127e-02, -3.4700e-02,
         9.0412e-03,  3.3711e-03, -3.1189e-03,  4.1690e-03, -8.5500e-03,
        -9.0178e-03, -5.7819e-04,  3.4328e-02, -1.0695e-04, -8.8859e-03,
        -2.8199e-05,  1.3469e-02,  1.3995e-03,  1.0213e-02, -7.0368e-03,
        -4.7465e-02, -4.8697e-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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.3220e-01,  6.4823e-01,  2.3782e-03, -2.5675e-02,  1.0999e-02,
        -4.5302e-02, -1.9517e-02, -2.3220e-02, -1.9942e-02, -2.0787e-02,
         1.0602e-02, -6.6663e-03, -3.3874e-02,  6.0019e-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([ 0.1039,  0.5659,  0.0378, -0.0127, -0.0686, -0.0085, -0.0074,  0.0116,
        -0.0137, -0.0208, -0.0087,  0.0031, -0.0075, -0.0015, -0.0096, -0.0127,
        -0.0062,  0.0067, -0.0032,  0.0040,  0.0142, -0.0072, -0.0164,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.0097e-02, -7.4733e-01,  1.2267e-02,  1.4806e-02,  2.1833e-03,
         2.5377e-02,  8.0978e-03,  4.8994e-03, -2.2411e-02, -6.2770e-03,
         6.7680e-03,  2.4090e-02,  7.2355e-03, -6.8744e-03, -4.3759e-03,
         3.3267e-04,  1.2075e-02, -3.6571e-02, -3.2860e-02, -1.0453e-02,
        -4.6160e-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([ 3.9492e-02,  5.8453e-01,  2.5222e-02, -1.7607e-02,  4.8623e-02,
        -1.4411e-02, -8.9216e-03,  8.1300e-04, -2.7298e-02,  2.6875e-02,
        -2.7965e-02, -3.4683e-02,  9.0766e-05,  3.2144e-03, -2.5327e-02,
        -4.6112e-03,  5.0682e-03,  5.4017e-03,  2.4880e-03,  1.4520e-02,
         2.4236e-03,  1.5887e-02,  7.3428e-04, -4.7936e-03, -1.0242e-02,
         3.2102e-02, -1.6657e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 7.9216e-03,  7.2813e-01,  6.3853e-02,  2.3752e-02,  2.1187e-02,
         9.1754e-03, -3.2004e-03,  8.7298e-03, -5.3330e-02,  4.8567e-03,
        -1.7744e-02,  2.1444e-02, -6.7629e-04,  1.4080e-02, -5.0520e-04,
        -8.5572e-03,  6.0621e-04,  1.2253e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  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([-1.7583e-03,  6.7090e-01,  7.3430e-04,  6.6688e-02, -1.8135e-02,
         9.5935e-03,  1.0505e-02,  3.4444e-03, -2.2877e-03,  3.1904e-03,
         3.8383e-03,  6.2407e-03, -5.2555e-03, -1.2864e-02, -5.4861e-03,
        -2.8512e-03,  5.2700e-04,  7.9570e-03, -2.4541e-03, -4.8309e-03,
        -5.5811e-03,  1.5848e-02,  5.8852e-04,  3.0782e-03,  8.0267e-03,
        -1.3511e-02,  4.8291e-03, -6.2705e-03,  1.7936e-03,  7.7447e-05,
        -3.1439e-03, -8.4054e-03, -6.8039e-03, -6.6684e-04,  1.7075e-05,
        -5.7880e-03, -6.2618e-03, -1.5286e-03,  3.5386e-03, -6.0532e-03,
        -4.5424e-03, -5.6460e-03, -1.0324e-02,  6.1249e-03,  2.7217e-02,
         4.7907e-03,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.2179e-01,  4.7256e-01,  2.8015e-02, -2.6311e-02, -1.4054e-02,
         3.4814e-03,  7.0942e-03,  6.5908e-03,  4.7268e-03,  3.5950e-03,
         4.1492e-03, -2.9649e-04, -2.5296e-03,  8.8268e-03, -6.5498e-03,
         6.3278e-03, -5.3016e-04, -1.1308e-02,  1.8371e-02, -2.3236e-02,
        -8.1000e-03,  8.5620e-03,  1.4115e-02, -6.9382e-03, -3.8840e-02,
         2.3756e-03,  1.1972e-02,  4.2147e-03, -3.9952e-03, -8.4527e-03,
         7.2578e-03, -1.6219e-02, -6.7818e-03, -1.4648e-03,  2.7133e-03,
        -6.7596e-03, -1.9675e-02, -1.0430e-02,  2.2922e-04, -1.1282e-03,
         8.8343e-04,  4.3861e-03, -6.1246e-04,  6.5394e-03,  1.1517e-02,
        -1.2243e-02,  6.5208e-04,  1.2598e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.3678e-02, -5.6784e-01,  4.5921e-02,  4.1607e-02, -1.6453e-02,
         6.1536e-03,  4.5416e-03, -4.5373e-03,  9.2680e-05,  4.0672e-03,
        -1.4794e-02, -1.5909e-02,  5.0215e-03, -1.5960e-03, -6.9033e-03,
        -1.7428e-03,  1.8001e-02,  7.5901e-03, -2.6821e-03, -1.2797e-02,
         3.5338e-02, -2.8649e-02,  7.0750e-03, -8.6509e-03, -1.2425e-03,
        -2.4304e-03,  5.4959e-03, -6.2975e-03,  4.3112e-03, -1.5657e-02,
         3.4292e-03, -1.5665e-02, -2.1295e-02, -3.2534e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.1819,  0.5132, -0.0584, -0.0935,  0.0946,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.3418e-05,  6.1290e-01,  4.0953e-02, -1.3223e-02,  2.3182e-02,
         1.5422e-02,  1.4340e-02, -5.4498e-03,  5.2034e-03, -8.6701e-03,
         3.7353e-03,  1.5712e-03, -6.4418e-04, -1.1205e-02, -2.2019e-03,
        -2.9516e-03,  5.9047e-03, -1.9636e-02, -1.3715e-02, -7.6258e-03,
        -1.7865e-02, -1.1878e-03,  1.1710e-02,  1.7083e-02,  1.1105e-02,
        -1.8227e-03, -3.4486e-03,  6.0923e-03, -4.6466e-03, -1.0435e-02,
         6.3296e-03,  4.0683e-03, -1.9538e-03, -9.4818e-03, -1.8561e-04,
         5.2863e-03,  4.4067e-03, -1.0238e-03,  3.8649e-03,  2.3727e-03,
         4.2286e-03,  4.1405e-03, -1.5475e-03,  3.2899e-03, -3.9937e-02,
        -1.3875e-02,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0661, -0.4686,  0.0026,  0.0566,  0.0123,  0.0102,  0.0195,  0.0431,
        -0.0179,  0.0018, -0.0042,  0.0133, -0.0045, -0.0130, -0.0317, -0.0184,
        -0.0446,  0.0227, -0.0205, -0.0368,  0.0741, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0314, -0.6370,  0.0149,  0.0567,  0.0120, -0.0037,  0.0606,  0.0877,
         0.0377,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.2878e-01, -6.0334e-01, -5.8706e-02,  8.2738e-03, -8.4137e-03,
         7.6709e-03, -1.0340e-02, -2.2787e-02, -9.9890e-03,  8.2067e-03,
         1.6327e-03, -7.9396e-03, -5.4725e-02, -6.2586e-03,  3.0951e-04,
        -1.1332e-02,  3.7293e-02, -1.4000e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+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, -0.5975, -0.0329,  0.0179,  0.0592, -0.0359,  0.0194,  0.0435,
        -0.0255,  0.0264,  0.0194,  0.0082,  0.0095, -0.0407,  0.0080,  0.0397,
        -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.5307e-01,  4.8851e-01,  7.1841e-02, -3.9035e-02,  3.2283e-02,
        -2.4166e-02, -1.7097e-02,  5.2221e-03, -3.0764e-04, -4.9533e-02,
         8.2552e-03,  5.2174e-03,  5.9556e-04, -3.6640e-02, -4.7919e-02,
        -2.0304e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.1223e-01,  6.5806e-01, -1.3828e-02,  4.3707e-02, -1.2915e-02,
         1.4485e-04,  3.5058e-02, -7.4042e-03,  1.3544e-02,  2.6387e-02,
        -2.5473e-02,  1.1956e-02, -2.9455e-03, -1.4408e-02, -1.5601e-02,
         3.8417e-03,  2.4930e-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([-0.0458, -0.6399,  0.0129,  0.0246,  0.0431, -0.0016,  0.0196, -0.0197,
        -0.0232, -0.0149,  0.0046, -0.0110, -0.0039, -0.0559, -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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         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([-0.1074, -0.5526, -0.0384,  0.0121,  0.0047, -0.0327, -0.0197,  0.0378,
        -0.0140,  0.0081, -0.0014, -0.0384, -0.0104,  0.0567,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0362, -0.5166, -0.0500,  0.0253,  0.0115,  0.0359,  0.0256, -0.0132,
         0.0169, -0.0046,  0.0108,  0.0305,  0.0124,  0.0619,  0.0055,  0.0031,
        -0.0165,  0.0681, -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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0336, -0.6140,  0.0022,  0.0336,  0.0087,  0.0393,  0.0273,  0.0318,
         0.0271, -0.0259,  0.0132,  0.0324,  0.0076,  0.0110, -0.0347, -0.0027,
        -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,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.2516e-02, -4.5380e-01, -1.3928e-02, -4.8464e-02,  5.0456e-03,
        -5.5268e-03,  1.7563e-02, -3.4740e-02,  2.0779e-04, -7.8691e-03,
         2.5168e-02,  1.1799e-02,  3.1888e-03,  1.7388e-03, -1.8424e-02,
        -9.2339e-03, -3.2925e-04,  4.1146e-03, -4.8435e-04,  3.2681e-03,
        -1.6302e-03, -1.4541e-02,  3.0504e-03, -1.9416e-03, -4.7672e-03,
        -7.5872e-03, -1.3500e-02,  6.0695e-03, -5.9150e-03, -5.9884e-04,
        -1.0584e-03,  7.4289e-04, -4.5380e-03,  2.6665e-03, -2.5184e-02,
         8.4553e-04, -1.1169e-02, -9.5485e-03, -1.0051e-02, -1.0381e-02,
         5.0611e-03, -9.4229e-03, -1.0273e-02, -1.5649e-02, -1.0480e-02,
        -3.9648e-03,  1.4178e-02, -6.7626e-03, -4.8395e-02,  2.6219e-03],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0582, -0.5732, -0.0590,  0.0398,  0.0062,  0.0053,  0.0168,  0.0270,
        -0.0127, -0.0170, -0.0479, -0.0135,  0.0123,  0.0170,  0.0125, -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([-0.0280, -0.7505, -0.0392, -0.0050, -0.0037, -0.0316, -0.0054,  0.0080,
        -0.0133, -0.0068,  0.0208, -0.0021, -0.0098, -0.0068, -0.0109, -0.0098,
        -0.0139, -0.0056,  0.0188,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.4255e-02,  6.2858e-01, -3.4979e-02,  2.0528e-02, -4.1834e-03,
         9.3023e-03, -2.4957e-02,  2.5829e-02,  4.8359e-03, -3.5433e-04,
        -3.1794e-02,  1.4837e-02, -1.1217e-02, -3.1165e-04, -2.0718e-03,
         9.8047e-03,  3.0622e-02, -8.4248e-03,  2.9583e-02, -8.0752e-03,
         1.3492e-02,  1.5920e-03, -6.1319e-03, -2.4453e-03, -1.2465e-02,
        -1.7573e-02,  1.1758e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0515, -0.7006, -0.0129,  0.0040,  0.0208, -0.0196,  0.0067,  0.0538,
         0.0386, -0.0133, -0.0060,  0.0062, -0.0337,  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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.1396, -0.4632,  0.0005, -0.0293,  0.0098,  0.0114,  0.0110, -0.0069,
        -0.0030,  0.0108, -0.0166, -0.0103,  0.0805, -0.0525,  0.1044,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0277,  0.6007, -0.0025,  0.0031, -0.0126, -0.0059,  0.0281, -0.0286,
         0.0077,  0.0111,  0.0011,  0.0153, -0.0299,  0.0012, -0.0451,  0.0011,
        -0.0031, -0.0051, -0.0130, -0.0074, -0.0021,  0.0019,  0.0092, -0.0164,
        -0.0323,  0.0332, -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([-5.4292e-02, -5.7459e-01, -4.3361e-02,  4.2566e-02,  7.5786e-03,
         3.0535e-02,  2.6165e-04, -6.4363e-03,  2.1915e-02,  2.8287e-04,
        -6.8820e-03,  1.5051e-04,  1.0258e-02, -1.1895e-02,  1.9276e-02,
        -4.7169e-03,  1.1499e-03, -2.3804e-02,  8.0906e-03, -2.2844e-03,
         1.6700e-02,  7.7497e-03,  3.3955e-06, -5.0099e-03,  2.9066e-03,
         9.7302e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.3256e-02, -5.8700e-01,  2.1719e-02,  3.3372e-02, -4.0088e-02,
        -1.1174e-02, -6.8942e-03,  6.9786e-04, -6.6316e-03, -5.1294e-03,
         1.1413e-02,  1.2047e-02, -1.3914e-02,  5.1812e-03,  2.0766e-03,
        -2.3141e-03,  1.4073e-02,  6.0575e-03,  7.0250e-03,  1.0928e-02,
         1.5687e-02,  3.6922e-03, -8.4957e-03,  4.9583e-06, -4.1811e-03,
         1.1471e-02, -2.9828e-02, -1.8976e-03,  1.8193e-02, -1.5373e-02,
         6.0184e-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],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #250: [tensor([-3.3725e-02, -6.9485e-01,  6.1720e-03, -3.7886e-02, -7.6313e-04,
        -3.5133e-03,  5.3483e-03, -2.0065e-02,  3.0289e-02,  1.6197e-04,
        -6.6925e-03,  1.5651e-02, -8.1615e-03,  6.3113e-03,  4.3097e-04,
        -1.4177e-02, -1.6956e-02,  5.2884e-02, -4.5958e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0259, -0.7756, -0.0321,  0.0218,  0.0194,  0.0095,  0.0688, -0.0345,
        -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,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0050, -0.7419,  0.0066, -0.0313, -0.0013, -0.0220,  0.0083,  0.0050,
         0.0228,  0.0313,  0.0093,  0.0357,  0.0082, -0.0096,  0.0083,  0.0017,
        -0.0257, -0.0076,  0.0013, -0.0046,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-2.4937e-03, -7.1519e-01, -3.7293e-05,  4.4742e-03,  8.4227e-04,
         5.0560e-03,  2.5494e-02,  6.9166e-03,  1.0781e-02,  1.1874e-02,
        -1.1701e-02,  1.5807e-02, -7.0776e-03,  8.7523e-03, -1.5360e-02,
        -1.1160e-02,  1.2700e-02,  1.3848e-02,  3.2765e-03,  1.0380e-02,
        -5.7406e-04,  4.2194e-03, -7.1259e-03,  2.4383e-02, -1.7724e-03,
        -2.0545e-02,  4.8156e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.3270e-04,  6.3422e-01,  5.0131e-02,  1.1491e-03, -2.2311e-03,
         1.6583e-02, -1.6570e-02,  1.6720e-03, -9.6297e-03, -1.4240e-02,
         1.1370e-02, -2.1896e-02, -2.3175e-02, -3.5777e-03,  1.7249e-02,
        -3.5686e-03,  1.1989e-03, -1.2980e-02, -5.0189e-03,  3.2287e-03,
         4.5314e-04, -1.1242e-02,  1.1371e-02,  9.5336e-04,  3.8898e-03,
         9.7575e-03, -1.0845e-02,  3.8752e-03, -1.1622e-02,  1.1539e-02,
         1.6437e-03, -2.0099e-03, -7.4411e-03, -3.2307e-03, -2.7979e-03,
        -3.8732e-03,  3.4914e-03,  2.5844e-02,  2.4200e-02,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.9680e-03, -6.7661e-01,  2.9554e-02,  2.1668e-02, -4.3452e-03,
        -3.2708e-02, -1.3649e-02, -2.2291e-02, -1.5786e-02, -3.9935e-02,
         2.2691e-02, -1.2208e-02,  1.7196e-02, -3.3823e-02,  3.5206e-04,
        -3.1022e-04, -6.5463e-03, -1.3293e-03,  9.1416e-03,  3.0886e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.0865e-02,  6.7557e-01, -4.3660e-02,  1.3303e-02, -6.9633e-03,
         3.0796e-02, -1.8446e-02,  1.9483e-03,  4.3318e-04, -8.8846e-03,
         1.4016e-02, -5.9191e-03,  8.1739e-03, -8.9711e-04, -2.3750e-02,
         6.2398e-03, -2.2107e-02, -8.0238e-03,  1.3733e-03, -5.3236e-03,
         1.3821e-03, -2.0696e-02, -4.1225e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 4.4952e-02,  6.4161e-01,  9.3172e-03, -5.5617e-03,  1.8056e-02,
        -4.8471e-03,  5.4351e-03,  3.8037e-03, -5.9529e-03, -4.7895e-04,
         3.4187e-03,  4.1598e-03,  4.0194e-03,  8.1606e-03, -3.5131e-03,
         1.3184e-02, -1.3023e-02,  6.1647e-04, -1.3658e-02,  2.2229e-03,
        -1.3786e-03,  2.3337e-03, -1.6290e-03,  2.6837e-02,  1.8408e-02,
        -2.1044e-02, -7.1198e-03, -3.5713e-03,  2.9264e-03,  4.5982e-03,
        -6.1507e-03,  5.0421e-04,  8.6431e-03, -1.5191e-03,  1.0034e-03,
         5.7732e-03,  2.2520e-02, -7.5049e-04, -1.2850e-02, -4.4448e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-8.0629e-03, -5.9731e-01, -6.1284e-02, -2.6199e-02, -2.4473e-02,
        -1.9203e-02, -6.2925e-03,  3.2180e-03,  2.1594e-03, -1.7213e-02,
        -6.7718e-03, -1.2358e-02,  5.2295e-03,  4.7211e-03, -1.3017e-03,
        -4.5965e-03, -2.3258e-03, -1.6497e-02, -6.6140e-03, -1.3990e-02,
        -6.9227e-03, -6.9563e-03, -8.9241e-04, -2.7993e-04, -6.8401e-03,
        -1.8796e-03, -2.9735e-03, -4.4485e-03, -1.7500e-04, -7.1200e-03,
        -6.0085e-04,  9.3963e-03,  1.4156e-02,  1.6668e-03, -2.1359e-02,
        -7.4510e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1472e-01,  4.9861e-01, -6.8054e-03, -2.1975e-02,  1.2541e-02,
        -4.6633e-02, -2.9135e-02, -2.6883e-02,  1.9368e-03, -1.9867e-02,
        -8.5715e-03,  4.1811e-03, -4.5890e-03, -2.7238e-03,  1.2309e-02,
        -2.2313e-02, -2.9305e-04, -1.1169e-04,  1.1068e-02, -9.6426e-03,
         8.3254e-03, -3.5188e-03, -3.3358e-03, -1.7324e-02, -1.2750e-02,
         1.5338e-03, -1.9407e-02,  4.5273e-03, -1.9313e-02, -1.6873e-02,
        -3.7948e-02,  2.3966e-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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 1.8990e-01, -5.8930e-01, -4.9618e-05,  4.2500e-03,  4.4508e-02,
        -6.3264e-03,  3.1766e-02,  2.9743e-02,  1.8499e-02,  8.5660e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0035, -0.7280,  0.0038, -0.0401,  0.0130,  0.0039, -0.0303,  0.0097,
        -0.0013,  0.0048,  0.0067,  0.0446,  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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>)]
Iter #300: [tensor([-0.0038, -0.6804,  0.0289, -0.0237, -0.0480,  0.0232, -0.0216, -0.0464,
         0.0102,  0.0193, -0.0138,  0.0083, -0.0306, -0.0168,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-4.5878e-02,  4.9607e-01,  4.0686e-02,  5.9827e-02,  1.9049e-02,
         1.0745e-02, -7.0481e-03, -4.0202e-03, -9.4700e-05,  1.2034e-02,
         6.1019e-03,  4.6445e-03, -3.9815e-02, -3.2313e-02,  1.2892e-02,
         3.1367e-02,  2.1076e-02, -9.5846e-04, -5.6343e-04, -1.0352e-03,
         4.3194e-03, -1.6020e-02, -4.6773e-02, -9.5671e-03,  1.1424e-02,
        -6.2564e-03,  5.2026e-03,  1.2955e-02, -5.1406e-03, -1.8847e-02,
         3.8754e-03,  1.5924e-03,  1.1809e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.5542e-03,  8.7607e-01,  5.8470e-03, -1.3506e-02,  4.0604e-04,
         6.1829e-02,  2.8712e-02,  7.0723e-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], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.4748e-02, -5.8731e-01,  1.0588e-02, -1.5688e-02, -8.4558e-03,
         1.6719e-02,  3.2348e-02,  1.2139e-02, -6.1148e-04, -2.0238e-02,
        -1.5051e-02,  8.9009e-03,  1.4586e-02, -1.8203e-02, -1.1189e-02,
         1.9757e-02,  1.0480e-02,  1.0179e-02, -1.4232e-02, -3.4183e-05,
        -4.9725e-03, -1.4653e-02, -1.5440e-03,  8.7058e-03,  2.4780e-03,
         1.0018e-02, -2.3709e-02, -1.2491e-02,  2.5892e-02,  1.4082e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-9.0771e-02,  5.0002e-01, -9.8541e-02, -6.0513e-02, -5.8831e-02,
         8.9579e-03, -1.4405e-02, -1.5151e-02, -1.6288e-02,  2.1781e-04,
        -5.8042e-04,  2.2575e-02, -4.7256e-02, -2.9126e-03,  1.9917e-02,
        -1.5467e-03, -4.1514e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-5.6578e-02, -6.1631e-01,  2.6414e-02, -5.6565e-02,  5.4042e-02,
        -8.0335e-03, -7.1121e-03,  1.6277e-02,  9.1165e-03, -1.3633e-02,
        -2.0164e-03, -5.9572e-03,  1.5150e-02, -2.0082e-02, -8.7393e-03,
         4.4413e-04,  1.1194e-03,  2.0301e-03, -2.9103e-03, -1.4817e-02,
        -9.2659e-03, -1.3341e-03, -4.2773e-02, -9.2777e-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([ 5.6144e-02,  5.4922e-01,  3.7493e-02, -2.8572e-03,  3.0714e-02,
         1.3698e-03,  7.4280e-03, -1.2310e-02,  3.5801e-02,  1.1578e-02,
         6.7956e-04, -7.5917e-03, -8.1289e-03,  2.4449e-03,  2.5941e-03,
         4.1823e-04, -3.1303e-02, -1.2200e-02,  1.6050e-03, -5.7495e-03,
         5.1871e-03, -3.5437e-03, -2.6848e-03,  9.4390e-03,  6.2870e-03,
        -3.7422e-03,  6.3939e-03,  2.3608e-02, -6.2132e-03, -3.6423e-03,
        -6.3143e-03,  8.5050e-03, -4.3897e-05,  2.2345e-02, -6.2897e-03,
         7.8600e-05,  5.9763e-04,  3.7281e-02,  1.4342e-02,  8.7471e-03,
        -2.7678e-03,  4.3162e-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([-3.4959e-02,  5.7041e-01, -5.5364e-02,  1.5958e-02,  2.7596e-03,
         8.2124e-03,  1.7388e-02, -2.5318e-05,  3.8391e-02,  2.1154e-02,
         9.3040e-03,  8.8966e-03,  2.9385e-04,  1.1557e-02,  1.6043e-02,
         1.1187e-02,  2.7620e-03,  4.9643e-03, -2.1855e-04,  6.8599e-03,
         2.9724e-03, -5.4851e-03, -3.1264e-03, -9.0574e-03,  7.9674e-03,
        -1.0852e-02, -2.5158e-04,  1.2076e-02,  1.7465e-02,  7.4021e-03,
         2.0292e-02, -9.9599e-03,  1.0105e-02,  8.8604e-03,  2.5818e-02,
        -2.0319e-03,  7.3240e-03, -2.2487e-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>), tensor([-0.0524, -0.7386,  0.0085, -0.0800, -0.0336, -0.0195, -0.0038, -0.0116,
        -0.0132,  0.0041,  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], device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.1311e-01,  6.2332e-01, -9.4458e-03, -5.8841e-03,  1.3728e-02,
        -1.0713e-02, -4.8582e-03, -3.2785e-03,  3.8548e-03, -3.6878e-03,
        -3.3006e-02, -2.1962e-03, -5.7639e-04, -7.8198e-03,  6.2605e-03,
         1.2939e-02, -2.7003e-02, -8.8733e-03, -4.6363e-03,  5.5644e-03,
         9.6173e-03,  2.3860e-04,  3.7132e-03,  6.8829e-03, -5.8529e-04,
        -2.7135e-04,  7.1536e-03, -6.0554e-03,  6.9131e-03, -4.4188e-03,
         2.4086e-02,  2.3194e-02,  6.1191e-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([-9.2245e-02,  4.2649e-01, -3.6606e-02, -7.7810e-03, -1.5150e-02,
        -8.0717e-03, -3.8652e-05, -4.7383e-03,  5.8076e-03, -2.1020e-03,
         5.1685e-03,  7.1282e-05,  6.4344e-03, -6.6762e-03,  3.7188e-03,
        -1.6767e-02, -1.9443e-02, -5.8620e-03, -4.8292e-02, -1.1994e-02,
        -3.9511e-03, -6.6207e-03, -1.0178e-03, -4.7114e-04, -1.4460e-03,
         3.6177e-03, -2.6924e-03, -5.1805e-03, -7.2117e-03, -8.7979e-03,
         1.8089e-02, -4.9341e-03, -3.0034e-02,  2.6385e-03, -2.0185e-02,
        -9.5940e-04, -6.1395e-03, -5.7788e-03, -4.5752e-03,  5.2068e-03,
         1.0396e-02,  8.3041e-03,  6.3294e-03,  1.0521e-03,  1.9388e-03,
        -8.1644e-03, -5.6980e-04, -1.0244e-02,  2.7241e-03, -5.6848e-02,
         3.0425e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.0232e-01,  4.5607e-01,  1.7117e-02, -1.3547e-02, -3.6200e-02,
         1.3408e-02,  4.2768e-03,  1.5473e-02, -1.9961e-02, -2.8586e-03,
        -4.9184e-03,  1.3595e-03, -9.4079e-03,  7.0060e-03, -6.7387e-03,
        -1.2301e-03, -2.8755e-03, -5.8845e-04,  6.5249e-04,  9.3254e-04,
        -2.3185e-02, -6.8011e-03, -5.6884e-03,  6.4922e-04,  1.0669e-02,
         8.7404e-03, -2.5800e-02, -1.1122e-03, -5.3288e-03, -5.4844e-03,
         1.9236e-03, -3.5747e-03, -7.0650e-03, -9.6549e-03, -1.5685e-02,
         4.4142e-04,  5.1257e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         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([-5.2662e-02, -5.5459e-01,  2.9467e-02,  6.8779e-02,  9.9155e-03,
         3.1615e-03,  7.1118e-03, -1.3242e-02, -1.3917e-02, -5.8037e-04,
        -4.0464e-05,  3.3144e-03, -1.0429e-02,  2.6997e-05,  4.6286e-03,
         8.9407e-03,  5.3538e-03,  2.8945e-03, -1.0459e-02,  8.8441e-03,
         2.2252e-02,  2.3657e-02,  8.8412e-03,  5.7695e-03,  8.8184e-03,
         1.3118e-02,  1.0240e-02,  2.1498e-02, -1.8014e-02, -5.3498e-03,
        -7.8287e-03,  1.3873e-02,  7.2535e-03,  4.4462e-03,  4.4211e-03,
        -1.6264e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.9549e-02, -6.4484e-01, -5.6465e-03,  1.3592e-02,  1.6020e-02,
        -8.9174e-03, -9.0085e-04,  1.2680e-02, -1.8295e-02,  1.1594e-02,
         3.6875e-02,  3.6410e-03,  3.3084e-03,  2.9799e-03,  7.2050e-04,
        -7.5969e-03,  8.2455e-04, -1.4743e-03,  2.4631e-04,  3.2317e-03,
         2.8163e-03,  6.0865e-03,  7.7178e-03,  8.9594e-03,  8.1896e-03,
        -5.7066e-03, -2.1985e-04, -7.0625e-04,  1.1816e-02,  2.3973e-05,
         2.6914e-03, -5.0591e-03, -3.5459e-03, -4.4823e-03,  2.9404e-03,
         2.8196e-03,  4.9116e-03,  1.0006e-03,  9.9884e-04, -8.0800e-04,
         1.2029e-03,  6.2036e-03,  3.1069e-03, -7.4379e-04, -6.5997e-03,
        -2.8826e-03,  3.7656e-03, -5.2911e-03, -1.7945e-03,  3.7620e-03,
         4.5285e-03,  2.5677e-03, -2.4088e-03, -1.5918e-02, -3.4793e-02],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.2705e-01, -5.5279e-01, -2.7447e-02, -1.5073e-02, -1.0970e-02,
        -1.5497e-02,  6.4872e-03,  2.3210e-03, -6.1353e-03, -3.9167e-04,
        -1.6382e-02, -3.9802e-03,  1.0952e-02,  2.3597e-02, -2.9604e-04,
         5.2599e-02,  3.4396e-03,  2.5083e-03, -7.0412e-04,  1.1047e-02,
         2.5126e-02,  1.9503e-02, -2.4765e-02, -2.2351e-02, -1.8368e-02,
        -2.2392e-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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 5.3464e-03,  7.7287e-01,  5.4968e-02,  4.7725e-03, -3.4713e-03,
        -5.2242e-03,  1.8856e-02,  2.9863e-03,  3.5767e-03, -1.1600e-03,
         3.3899e-04, -5.8194e-03,  5.7683e-03, -2.1659e-02, -1.9164e-02,
         1.0278e-03,  1.1319e-02,  5.2819e-03, -1.7780e-03, -1.0701e-02,
        -1.3314e-03,  2.7639e-03,  3.9816e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-6.3445e-02,  5.9567e-01,  6.5476e-02,  1.4531e-02,  7.6258e-03,
         1.4233e-03,  2.2973e-02,  7.5217e-04, -1.1567e-02, -5.8559e-03,
        -1.4394e-02,  1.6183e-02,  4.1147e-02, -2.5388e-03, -6.1387e-04,
        -1.7729e-02, -3.9841e-03,  4.8465e-04,  1.5819e-02,  5.9430e-03,
         5.2357e-03, -4.6057e-03, -4.0637e-03, -1.6618e-03,  2.0091e-02,
         9.0431e-03, -5.4549e-03,  2.9458e-02, -1.2230e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.1142,  0.7718,  0.0263,  0.0195,  0.0283, -0.0068,  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,  0.0000,  0.0000,  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,  0.5778, -0.0600,  0.0157,  0.0282, -0.0353,  0.0180, -0.0049,
         0.0018, -0.0236,  0.0073,  0.0380,  0.0066,  0.0036,  0.0123,  0.0076,
         0.0256,  0.0506,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 3.8391e-02,  6.4198e-01,  9.3961e-03,  1.7425e-02,  2.2231e-02,
        -3.3816e-02, -1.2922e-02, -9.9389e-03,  2.1681e-02,  1.1294e-02,
        -8.5151e-03,  8.4681e-03, -1.6632e-02,  3.6274e-02, -1.0571e-03,
         2.2434e-03,  1.0999e-02,  7.5778e-03, -1.9257e-03, -3.0701e-03,
         6.1474e-03,  4.1614e-04,  9.8331e-03, -1.1490e-03, -1.3239e-02,
         4.3545e-03,  2.0623e-02,  2.8405e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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, -0.6295,  0.0103,  0.0188,  0.0102, -0.0153, -0.0235, -0.0016,
         0.0068, -0.0028,  0.0052,  0.0135, -0.0071,  0.0071, -0.0044, -0.0077,
        -0.0193, -0.0074,  0.0176, -0.0067,  0.0176,  0.0200,  0.0008, -0.0012,
        -0.0036,  0.0020,  0.0136, -0.0028, -0.0186, -0.0686, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.6494e-02, -5.7497e-01, -1.4809e-02,  1.5467e-02, -3.4774e-02,
         3.5496e-02, -1.0654e-02,  6.7864e-02, -2.2866e-04, -1.1151e-02,
         2.6601e-02,  3.9231e-04,  3.8946e-02,  1.1200e-02,  8.0950e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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,  0.7060, -0.0329,  0.0226, -0.0105, -0.0015,  0.0193,  0.0115,
         0.0144, -0.0124,  0.0022,  0.0074,  0.0041, -0.0475, -0.0057, -0.0067,
        -0.0256,  0.0127, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-1.3828e-02,  6.9201e-01, -4.6642e-02,  1.6212e-02, -2.5428e-02,
        -9.9992e-03, -2.4871e-02, -3.3613e-03, -3.8226e-03,  6.6887e-03,
         8.1118e-03,  5.0815e-04, -3.5243e-04,  8.7426e-03, -1.9076e-02,
         2.7721e-02,  1.1002e-02,  5.6719e-04,  4.1212e-03, -5.6027e-03,
         4.8328e-03, -3.1167e-03, -6.2013e-03,  1.2636e-03,  5.8167e-03,
         5.3539e-03, -1.0279e-02,  1.0486e-02,  2.1964e-02,  2.0172e-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>)]
Iter #400: [tensor([ 0.1205,  0.6224, -0.0154,  0.0254, -0.0152,  0.0070, -0.0011, -0.0016,
        -0.0305, -0.0261, -0.0046, -0.0176, -0.0232, -0.0065,  0.0088,  0.0129,
        -0.0092,  0.0089, -0.0083, -0.0064, -0.0162, -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0064, -0.7049, -0.0536,  0.0255,  0.0036, -0.0085, -0.0042,  0.0056,
         0.0011,  0.0130,  0.0156, -0.0267,  0.0221,  0.0092,  0.0050, -0.0043,
        -0.0041,  0.0036,  0.0072,  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,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0130, -0.7010,  0.0530, -0.0253,  0.0202,  0.0060, -0.0054, -0.0023,
        -0.0042, -0.0041, -0.0066,  0.0042,  0.0032,  0.0139, -0.0124,  0.0020,
        -0.0037, -0.0136,  0.0065, -0.0165,  0.0023, -0.0011,  0.0039, -0.0133,
        -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.8479e-02, -5.9797e-01,  8.1379e-02,  3.2112e-02, -3.0010e-03,
         1.9080e-02, -1.4011e-02, -1.5439e-03, -9.4170e-03,  2.3057e-02,
        -1.8537e-02,  1.6477e-02,  4.3715e-03, -2.0938e-03, -2.1040e-03,
         1.4128e-02, -9.1861e-03, -6.3469e-03, -5.5425e-03,  1.0753e-02,
         1.4262e-02, -3.2947e-03, -8.3438e-03,  5.1126e-04,  5.9867e-03,
         3.6902e-03, -1.6502e-02, -1.3616e-02,  3.4199e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0567,  0.7152,  0.0575,  0.0135, -0.0290, -0.0050,  0.0210,  0.0096,
         0.0346,  0.0505,  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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([ 8.4017e-05,  5.5912e-01,  3.9058e-02,  2.5617e-02,  9.5429e-03,
        -3.1726e-02,  8.3179e-03,  1.3931e-02,  1.0671e-02,  7.7992e-03,
        -2.4239e-02,  7.5291e-03,  2.3524e-02,  1.3107e-02,  3.8496e-02,
        -3.4677e-03, -1.5087e-03, -6.6407e-04,  1.1674e-02, -1.2528e-02,
         5.2284e-03,  3.1755e-02,  5.8479e-02, -6.1935e-02,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 7.7115e-02,  5.8298e-01, -4.9125e-02,  7.7066e-03, -1.1516e-02,
         1.3756e-03, -1.0434e-03, -1.4321e-04,  1.3253e-03,  2.1056e-03,
        -2.5990e-02, -7.7620e-03, -4.8639e-03, -4.2193e-05,  7.9628e-03,
         2.3839e-04, -5.8849e-03,  2.2060e-03, -8.5039e-03, -3.6449e-03,
         4.8569e-03,  4.5615e-04,  5.4902e-03,  1.2379e-02,  9.5499e-03,
        -1.4430e-03, -4.2967e-03,  5.1414e-03,  4.2406e-03, -7.5369e-03,
        -2.4271e-03,  1.0206e-03, -2.4157e-03, -6.3975e-03, -7.7619e-04,
        -3.2978e-03, -1.7443e-02, -1.7349e-03, -4.4609e-03,  2.5842e-03,
        -3.5578e-04,  1.2653e-04,  9.7880e-03, -1.0791e-02, -2.2386e-03,
        -9.1287e-03,  1.7832e-02,  5.0258e-02], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0776, -0.6826,  0.0398,  0.0174, -0.0164,  0.0137, -0.0037, -0.0192,
        -0.0086,  0.0054,  0.0236,  0.0203,  0.0075, -0.0025, -0.0146,  0.0266,
        -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],
       device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0441,  0.6220, -0.1070, -0.0167, -0.0013, -0.0056, -0.0802,  0.0221,
         0.0111,  0.0363, -0.0227, -0.0128,  0.0119, -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([ 6.5114e-02,  6.5259e-01,  3.4669e-03, -3.3188e-04, -3.3453e-03,
         8.3201e-04,  1.4750e-02,  9.3409e-03,  1.8347e-02,  5.6753e-02,
        -2.5695e-03,  7.6966e-03,  1.6922e-03, -3.0944e-03, -5.6278e-03,
         2.1604e-03,  4.8476e-04, -6.1244e-05,  5.0561e-04,  2.3300e-02,
        -1.3641e-02,  1.1097e-02,  4.4549e-03,  1.0658e-02,  8.5412e-03,
        -1.2481e-02,  6.1954e-03,  9.4594e-04,  2.4422e-03, -2.9290e-03,
         1.5517e-02, -1.1083e-02,  2.5110e-02,  2.8375e-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([-1.3487e-02,  5.5429e-01,  2.5654e-02,  4.7429e-02,  2.4277e-02,
         7.8469e-03,  1.2959e-04, -2.2939e-02,  3.0421e-03, -5.1224e-03,
        -4.8913e-03, -1.7717e-02,  2.5913e-04, -1.1840e-03, -7.4999e-03,
        -1.2926e-02,  4.5582e-03, -1.3102e-02,  1.1519e-03,  1.4513e-02,
         5.5266e-04,  2.3898e-03, -1.8222e-03,  4.4823e-04,  2.4229e-02,
         6.7222e-04, -1.5682e-03, -4.5125e-03,  1.1799e-02, -2.5671e-02,
        -1.2875e-03, -6.7315e-03, -2.5925e-02, -6.7103e-03,  1.2238e-03,
         5.1664e-03, -1.8371e-02,  3.0505e-02, -5.2462e-03, -3.7787e-03,
         7.5018e-04,  2.3235e-02,  1.5385e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-8.9818e-02,  6.2759e-01,  8.1381e-02,  7.0231e-03, -5.4160e-03,
         6.0988e-04,  7.0501e-03, -7.8984e-04,  2.4063e-02, -7.1500e-03,
         6.8459e-03,  1.0384e-02,  6.1362e-03,  2.7463e-03, -2.8252e-03,
        -7.5603e-04,  6.5149e-03,  5.4645e-03, -3.1838e-03,  6.1025e-03,
        -5.0669e-03, -6.6229e-03, -1.2734e-02,  8.2331e-03,  1.2995e-03,
        -5.4274e-05,  7.4223e-03, -3.2836e-03, -8.2497e-04,  1.0107e-02,
         3.9386e-03,  4.5101e-04, -1.4488e-02,  4.7093e-03,  1.8917e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  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([-0.0036,  0.7129, -0.0098, -0.0448,  0.0064,  0.0120,  0.0089, -0.0037,
        -0.0357,  0.0188,  0.0098, -0.0304, -0.0065,  0.0205, -0.0119, -0.0010,
        -0.0017,  0.0112,  0.0027, -0.0136,  0.0081,  0.0008,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.7415e-02,  6.0374e-01,  8.4879e-03,  7.2338e-04, -2.4916e-02,
         3.9817e-03, -1.0827e-02, -1.7908e-03, -1.4558e-02, -9.2475e-03,
        -1.4100e-02, -5.8717e-03, -3.1516e-02, -5.5979e-03, -3.6837e-03,
         5.8743e-03,  6.5558e-04, -9.5567e-03,  1.0837e-02, -1.0864e-03,
         1.8352e-03, -1.0089e-02, -3.4067e-04, -1.0221e-02,  4.4806e-03,
        -6.4802e-03, -3.2991e-03,  6.1435e-03, -1.1561e-02, -2.3663e-02,
        -5.6737e-05,  1.5285e-02,  6.8971e-04,  2.9442e-03, -8.2162e-03,
         7.0228e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 2.1381e-01,  5.1356e-01,  8.2546e-02, -1.7782e-02,  5.7936e-03,
        -3.4861e-02, -3.8728e-03,  8.7842e-03,  5.7349e-03,  4.8720e-03,
        -5.1297e-04, -1.3773e-02,  6.0356e-03,  2.8209e-02, -5.9859e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0114, -0.6821,  0.0204,  0.0094,  0.0131,  0.0847,  0.0293,  0.0122,
         0.0051, -0.0175,  0.0241,  0.0214,  0.0123, -0.0450,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 3.9055e-03,  6.3313e-01, -1.4941e-03,  1.1722e-02,  2.1275e-02,
         3.3111e-03, -1.5583e-02,  3.4174e-03,  1.6904e-03, -2.2194e-03,
         8.1540e-03, -5.2879e-03,  5.7068e-02,  7.6725e-04, -1.5143e-03,
        -2.3235e-05,  3.0800e-03,  4.3283e-03,  5.5034e-03,  2.4037e-04,
         2.1546e-03, -4.1684e-03,  2.0192e-02,  3.0199e-03,  4.6915e-04,
        -1.7145e-02,  1.9288e-03, -8.4414e-03,  9.7783e-03, -2.9262e-03,
        -9.0175e-04, -4.0871e-03, -1.5984e-02,  4.7344e-03,  3.5965e-03,
         2.5861e-03, -4.4547e-03,  1.8441e-03, -4.0855e-03, -9.2126e-05,
         1.4735e-03,  1.4119e-02,  5.0377e-03, -1.6726e-03,  1.8336e-02,
         5.9918e-03, -5.2892e-04, -9.8826e-05,  7.4856e-03, -3.8063e-03,
        -1.3694e-02,  3.1452e-02], device='cuda:0', grad_fn=<DivBackward0>), tensor([-0.0605, -0.6843, -0.0268, -0.0088, -0.0022,  0.0209,  0.0073,  0.0043,
         0.0043,  0.0124, -0.0120,  0.0201,  0.0208, -0.0152, -0.0146, -0.0056,
        -0.0075,  0.0398,  0.0242, -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([-4.7288e-02,  7.0867e-01,  2.5521e-02,  2.0446e-02,  7.7277e-03,
         2.3375e-02,  8.0418e-05,  1.2601e-02, -1.0395e-02,  1.3551e-02,
         1.2084e-02, -7.9261e-03, -3.0283e-02,  8.8992e-04,  1.3703e-02,
         2.2715e-03,  1.3609e-03,  1.1631e-02, -5.7123e-04, -1.4077e-02,
        -1.4543e-03, -3.4093e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0598, -0.6542, -0.0416, -0.0146,  0.0108,  0.0108, -0.0048, -0.0467,
         0.0274, -0.0235, -0.0090, -0.0194, -0.0048, -0.0019,  0.0202,  0.0342,
        -0.0077,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 1.2183e-01,  6.3795e-01,  1.1773e-02,  8.7308e-03, -1.8217e-02,
        -1.0466e-02, -3.0512e-02,  2.3968e-03, -7.8300e-03,  3.5892e-04,
         1.0512e-02, -1.0897e-02, -2.3356e-03,  3.5460e-03, -9.4023e-05,
         2.2499e-03,  1.0079e-02, -4.8227e-03,  4.9598e-03,  9.8312e-03,
        -2.7584e-03, -1.4284e-03,  1.1660e-03, -1.9795e-02, -1.1951e-02,
        -5.3508e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-7.3470e-03, -6.2424e-01, -7.2522e-02, -1.1880e-02,  1.0653e-02,
        -4.2223e-03,  8.1360e-03,  1.3272e-02,  9.8565e-03, -1.1913e-02,
        -2.0353e-02, -2.7841e-04, -3.4990e-02,  4.0189e-03, -3.3400e-02,
        -5.4549e-03,  7.3485e-03, -2.6849e-02, -1.2590e-02, -1.3320e-02,
         6.7352e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([-3.9611e-02, -6.0425e-01, -6.8881e-02, -3.3139e-03, -5.1405e-04,
         1.5880e-02, -4.5191e-03,  5.3461e-04, -4.4311e-03,  3.4201e-03,
         3.0451e-03, -2.4319e-02,  6.1290e-03,  5.3828e-03, -5.6712e-03,
         5.9893e-03,  5.2786e-03, -7.3917e-03,  1.0360e-03,  3.1114e-02,
        -7.5343e-04, -2.1647e-03, -9.8960e-04,  5.1212e-04, -1.3911e-03,
         2.0172e-02, -2.0141e-04, -5.5089e-03,  3.9899e-04,  6.2865e-03,
         8.5257e-03,  2.8099e-03,  3.0621e-03,  2.0770e-03, -3.0671e-03,
         4.1376e-03,  3.4248e-03,  1.3647e-02,  1.5366e-02,  3.4629e-03,
         8.2886e-04,  2.3369e-03,  3.4192e-04,  3.2353e-04,  1.1846e-02,
        -3.4030e-02, -1.1620e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00], device='cuda:0', grad_fn=<DivBackward0>), tensor([ 0.0478,  0.4436, -0.1279, -0.0679, -0.0141, -0.0903, -0.0682,  0.0018,
         0.1041, -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,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #500: [tensor([-7.9884e-03, -7.3608e-01,  1.1884e-02, -1.1959e-02, -1.0497e-03,
        -1.6537e-03, -6.0127e-04,  2.1738e-03, -7.1368e-03, -5.7437e-04,
         1.5992e-03, -3.6419e-03, -1.3681e-03, -6.8856e-03, -5.3205e-03,
        -6.1265e-04, -1.6126e-04,  2.4606e-03, -9.1516e-04,  1.8421e-02,
         5.0484e-03,  2.2704e-03,  1.0170e-02,  3.7186e-03, -7.7830e-03,
        -3.5673e-03, -6.5574e-03, -1.5878e-04,  9.3121e-03, -2.0308e-03,
        -2.6330e-03,  1.1873e-03,  5.6604e-03, -7.9526e-03, -7.7246e-03,
         1.0203e-02,  2.0139e-03,  2.7686e-03, -4.7548e-03, -1.0745e-02,
        -4.8251e-03,  3.8396e-03,  1.1775e-02, -4.4213e-04,  8.8498e-03,
        -2.4217e-02, -1.7307e-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([ 4.9280e-02,  7.0257e-01,  5.5515e-02, -3.5892e-03, -2.0634e-02,
         2.4391e-03,  3.4233e-03,  1.6322e-02, -2.5323e-03,  3.6641e-03,
        -9.1243e-03, -1.9686e-02,  5.0233e-03, -6.0305e-03, -8.1708e-05,
        -3.6969e-03, -3.8891e-03, -1.6017e-02, -7.5532e-03,  2.5348e-03,
         8.1956e-03, -1.3383e-02, -6.7593e-03, -1.0537e-02,  2.8785e-03,
         2.4640e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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,  0.4522,  0.0494,  0.0136, -0.0250, -0.0339, -0.0362,  0.0277,
        -0.0103, -0.0239,  0.0050, -0.0054,  0.0070, -0.0106, -0.0056, -0.0087,
        -0.0026, -0.0098,  0.0014, -0.0026, -0.0170, -0.0060, -0.0177,  0.0059,
        -0.0209, -0.0141, -0.0131, -0.0381,  0.0033,  0.0061,  0.0071, -0.0041,
         0.0014,  0.0028,  0.0093,  0.0058,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0085,  0.6438, -0.0989, -0.0344,  0.0107, -0.0272,  0.0092,  0.0321,
         0.0053,  0.0216,  0.0103,  0.0019, -0.0154,  0.0535, -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,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0079, -0.7477, -0.0442, -0.0081, -0.0059, -0.0124, -0.0019, -0.0190,
        -0.0015, -0.0432, -0.0294, -0.0270,  0.0069,  0.0292,  0.0115,  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([ 7.8109e-03, -6.9730e-01,  2.7004e-02,  3.7441e-02, -4.3518e-03,
         1.0209e-02,  1.2866e-02,  2.9854e-02,  1.1698e-02,  1.6302e-03,
        -1.0898e-03,  1.9397e-02, -1.1946e-02,  1.5455e-02,  1.7347e-03,
         8.0259e-03,  4.2191e-03, -3.0590e-03,  7.8953e-04, -2.1726e-06,
         8.2280e-03, -1.5792e-02, -1.3982e-03,  5.0748e-03, -1.1496e-03,
        -4.3710e-03, -1.0181e-02, -6.3151e-03, -3.0710e-02, -1.0902e-02,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.2008e-02, -7.4250e-01, -7.9155e-02,  5.0796e-03, -8.1851e-03,
        -4.0372e-03, -6.2106e-03, -6.4617e-04,  1.4316e-02,  2.4313e-04,
         5.9446e-03,  2.6609e-03,  6.5053e-03,  2.3678e-03,  2.3077e-03,
        -1.1394e-03, -4.7827e-04, -3.3631e-03, -8.2738e-03, -4.7294e-03,
        -1.7393e-03,  1.2479e-03, -2.7478e-03,  3.6583e-03,  8.8177e-03,
        -2.8156e-03, -4.4747e-03, -8.0105e-04, -3.1373e-03, -6.3190e-03,
         7.5491e-03,  1.0188e-02, -3.2501e-03,  2.3341e-03, -9.3988e-04,
        -9.8280e-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([ 7.4966e-02, -6.4275e-01,  7.1155e-03, -1.5469e-02, -1.8906e-02,
         2.2736e-02,  4.1881e-03,  1.8290e-04,  1.5397e-02, -8.7951e-03,
         1.2347e-02, -1.1193e-02,  8.0374e-03,  1.1841e-03, -1.1203e-02,
        -4.8265e-03,  8.0549e-03,  5.5858e-03, -9.4077e-03,  5.1623e-03,
        -2.5292e-03, -1.4869e-02,  8.7608e-04, -7.0738e-03,  2.7415e-03,
        -5.8657e-03, -9.2033e-03, -8.1353e-04,  3.5689e-03,  8.1305e-04,
        -9.4965e-03,  4.2633e-03,  2.1974e-02,  1.9757e-02, -8.6509e-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([-0.1168, -0.5143, -0.0572, -0.1034, -0.0463,  0.0182, -0.0011,  0.0182,
        -0.0558,  0.0210, -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,  0.0000,  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, -0.5536,  0.0532,  0.0345,  0.0011,  0.0256,  0.0218,  0.0079,
        -0.0130,  0.0068,  0.0120, -0.0117, -0.0010, -0.0271, -0.0015,  0.0143,
         0.0151, -0.0052,  0.0053,  0.0013,  0.0228, -0.0076,  0.0031,  0.0483,
         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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0025,  0.5767, -0.0097, -0.0155,  0.0410,  0.0363,  0.0024, -0.0066,
        -0.0108,  0.0069,  0.0082, -0.0118, -0.0043,  0.0050,  0.0104,  0.0223,
        -0.0389, -0.0007,  0.0058, -0.0040, -0.0225,  0.0268,  0.0258, -0.0315,
         0.0737,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-3.9292e-02,  5.8925e-01,  2.3025e-02,  2.6057e-03,  2.0559e-02,
         1.1491e-02, -1.0343e-03,  9.3307e-03,  4.5385e-03,  8.6184e-03,
         2.8509e-03, -1.2245e-02, -7.3162e-04, -2.9676e-03,  3.5862e-03,
        -6.6167e-03, -9.0222e-03,  6.0814e-03,  1.6361e-03, -1.1134e-03,
         2.2414e-03,  2.1954e-03, -4.7971e-03,  1.5736e-03,  1.8713e-02,
         1.2426e-02, -5.8177e-03, -9.1228e-03, -2.5427e-04,  1.0504e-02,
         1.1661e-03, -2.2435e-02,  1.0601e-02,  2.1301e-03,  8.5388e-03,
         2.9232e-03, -1.7303e-03, -1.9195e-03,  1.0755e-02, -6.9220e-04,
        -5.8491e-03, -2.3440e-03, -1.1367e-02, -6.1750e-03,  5.8035e-03,
         3.5305e-03, -1.8103e-03, -2.0468e-03, -3.6601e-03, -8.2403e-03,
        -2.1553e-02, -2.8356e-03,  3.1583e-02,  6.0666e-03], device='cuda:0',
       grad_fn=<DivBackward0>)]
Iter #550: [tensor([-0.0217, -0.7096,  0.1190,  0.0097,  0.0175, -0.0113,  0.0265,  0.0144,
         0.0067, -0.0095,  0.0134,  0.0036,  0.0239,  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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0247,  0.6104,  0.0681,  0.0357,  0.0064,  0.0353,  0.0301,  0.0544,
         0.0178, -0.0266, -0.0053,  0.0050, -0.0048,  0.0120,  0.0062, -0.0085,
        -0.0078,  0.0100, -0.0036, -0.0256, -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-0.0079,  0.7789,  0.0473, -0.0160, -0.0088, -0.0560, -0.0509, -0.0185,
        -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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 5.7242e-02,  6.4872e-01,  7.6852e-02, -1.3501e-02, -5.2563e-04,
         1.1460e-02,  9.5588e-03, -3.3440e-03,  2.3792e-03,  1.4417e-02,
         8.6951e-03, -6.7790e-03, -1.3781e-03, -4.1460e-03,  1.6325e-03,
         1.3174e-02, -4.4610e-03,  7.9862e-03,  1.0446e-03,  1.5713e-02,
         4.2672e-04,  5.8837e-03, -8.0091e-04,  1.1580e-02,  2.1345e-03,
        -6.3434e-03,  2.6998e-02, -4.2825e-02,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-2.2304e-02, -6.1227e-01, -5.9191e-02,  1.8180e-02, -8.6450e-03,
        -2.9544e-03,  7.1887e-03,  3.6886e-02, -1.5644e-03, -1.4875e-02,
        -2.8422e-02,  2.1993e-02,  2.7889e-04,  3.4320e-03,  6.0808e-03,
        -2.6724e-02, -8.7685e-03, -1.4702e-02, -1.7273e-02,  1.9034e-02,
        -1.1247e-02,  5.7984e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-9.3268e-05,  6.4903e-01, -6.7321e-02,  8.3821e-03,  5.1459e-03,
         5.1633e-03,  1.2447e-02,  9.7641e-03, -2.0949e-02,  2.9246e-02,
         1.0083e-02, -7.8510e-03,  2.0054e-02, -6.1971e-03,  3.7052e-03,
        -1.9561e-03,  4.3743e-03,  5.0567e-03, -3.4763e-04,  9.8928e-03,
        -5.5126e-04, -2.7908e-03,  2.1427e-03, -1.4633e-03,  6.4394e-03,
         1.1242e-02, -4.8567e-04,  1.6302e-02,  1.3876e-02, -1.2047e-02,
         9.1108e-04,  5.4691e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+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, -0.5862, -0.0782,  0.0064, -0.0033, -0.0514, -0.0040,  0.0097,
         0.0049,  0.0229,  0.0033, -0.0014,  0.0068, -0.0046,  0.0160,  0.0046,
        -0.0033, -0.0014,  0.0034,  0.0030, -0.0036, -0.0069, -0.0022,  0.0161,
        -0.0029, -0.0088, -0.0202, -0.0052, -0.0017,  0.0027,  0.0104, -0.0121,
         0.0012, -0.0009, -0.0070, -0.0100, -0.0069, -0.0092, -0.0108, -0.0032,
         0.0061, -0.0073, -0.0157, -0.0044], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([-1.0865e-01, -6.2844e-01, -4.8740e-02,  1.1162e-02, -1.2954e-02,
        -2.3580e-02,  1.2173e-02,  2.7520e-03, -1.8945e-02,  2.3382e-04,
        -6.5814e-03,  2.0134e-02, -4.3377e-03, -2.1036e-02,  1.9629e-02,
         4.4861e-02,  1.5793e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1043,  0.6664,  0.0064,  0.0268, -0.0089, -0.0263, -0.0014,  0.0088,
        -0.0054,  0.0186,  0.0074,  0.0020, -0.0022,  0.0057,  0.0024, -0.0054,
         0.0036, -0.0111,  0.0044,  0.0034, -0.0031, -0.0072,  0.0123, -0.0153,
        -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([ 2.3381e-02, -6.7762e-01, -6.3013e-02, -1.3247e-02,  1.1332e-03,
         2.1928e-02,  7.1042e-03, -2.1168e-02,  1.4738e-02, -2.2828e-02,
        -1.6115e-02, -1.0750e-02,  1.4920e-03, -1.8423e-04,  7.8623e-04,
        -1.7554e-03, -1.0684e-02, -1.3465e-02,  8.5784e-03, -3.7168e-02,
         2.2382e-02,  1.0479e-02,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,
         0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.1144,  0.6361, -0.0088,  0.0789, -0.0189,  0.0447, -0.0065, -0.0190,
         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], device='cuda:0',
       grad_fn=<DivBackward0>), tensor([ 0.0064,  0.6668,  0.0182, -0.0345,  0.0030,  0.0044,  0.0139, -0.0117,
         0.0087,  0.0051,  0.0013, -0.0073, -0.0203, -0.0223,  0.0035, -0.0061,
        -0.0034,  0.0112,  0.0029, -0.0123,  0.0158,  0.0108, -0.0046, -0.0127,
        -0.0057,  0.0063,  0.0027,  0.0129, -0.0128, -0.0052,  0.0127,  0.0078,
         0.0050,  0.0217,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,  0.0000,
         0.0000,  0.0000,  0.0000,  0.0000], device='cuda:0',
       grad_fn=<DivBackward0>)]
