Iter #50: tensor([[0.4930, 0.5070],
        [0.5345, 0.4655],
        [0.5527, 0.4473],
        [0.5023, 0.4977],
        [0.5543, 0.4457],
        [0.5529, 0.4471],
        [0.4597, 0.5403],
        [0.5636, 0.4364],
        [0.5420, 0.4580],
        [0.5530, 0.4470],
        [0.5168, 0.4832],
        [0.5325, 0.4675]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.6165, 0.3835],
        [0.5223, 0.4777],
        [0.5274, 0.4726],
        [0.5514, 0.4486],
        [0.4381, 0.5619],
        [0.6424, 0.3576],
        [0.5555, 0.4445],
        [0.5280, 0.4720],
        [0.4310, 0.5690],
        [0.4953, 0.5047],
        [0.4012, 0.5988],
        [0.5923, 0.4077]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.3527, 0.6473],
        [0.5609, 0.4391],
        [0.4057, 0.5943],
        [0.4377, 0.5623],
        [0.4795, 0.5205],
        [0.3879, 0.6121],
        [0.5035, 0.4965],
        [0.5386, 0.4614],
        [0.5661, 0.4339],
        [0.3683, 0.6317],
        [0.5524, 0.4476],
        [0.5453, 0.4547]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.4491, 0.5509],
        [0.5506, 0.4494],
        [0.4407, 0.5593],
        [0.3197, 0.6803],
        [0.3501, 0.6499],
        [0.3458, 0.6542],
        [0.5218, 0.4782],
        [0.5433, 0.4567],
        [0.5472, 0.4528],
        [0.4469, 0.5531],
        [0.5063, 0.4937],
        [0.6073, 0.3927]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.4911, 0.5089],
        [0.3688, 0.6312],
        [0.3043, 0.6957],
        [0.5418, 0.4582],
        [0.4680, 0.5320],
        [0.4823, 0.5177],
        [0.4297, 0.5703],
        [0.3486, 0.6514],
        [0.3671, 0.6329],
        [0.4030, 0.5970],
        [0.6237, 0.3763],
        [0.4313, 0.5687]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.6980, 0.3020],
        [0.3580, 0.6420],
        [0.4733, 0.5267],
        [0.4909, 0.5091],
        [0.4781, 0.5219],
        [0.3625, 0.6375],
        [0.4577, 0.5423],
        [0.5989, 0.4011],
        [0.5516, 0.4484],
        [0.5622, 0.4378],
        [0.6424, 0.3576],
        [0.3803, 0.6197]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.6403, 0.3597],
        [0.6495, 0.3505],
        [0.5053, 0.4947],
        [0.4972, 0.5028],
        [0.4204, 0.5796],
        [0.3111, 0.6889],
        [0.2972, 0.7028],
        [0.3888, 0.6112],
        [0.4661, 0.5339],
        [0.5103, 0.4897],
        [0.6018, 0.3982],
        [0.6820, 0.3180]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.4990, 0.5010],
        [0.3273, 0.6727],
        [0.4817, 0.5183],
        [0.4851, 0.5149],
        [0.4415, 0.5585],
        [0.4605, 0.5395],
        [0.5521, 0.4479],
        [0.4245, 0.5755],
        [0.5832, 0.4168],
        [0.6727, 0.3273],
        [0.6064, 0.3936],
        [0.4965, 0.5035]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.5562, 0.4438],
        [0.5750, 0.4250],
        [0.3569, 0.6431],
        [0.4517, 0.5483],
        [0.4511, 0.5489],
        [0.4728, 0.5272],
        [0.3954, 0.6046],
        [0.5885, 0.4115],
        [0.3957, 0.6043],
        [0.4166, 0.5834],
        [0.2817, 0.7183],
        [0.4019, 0.5981]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.5933, 0.4067],
        [0.5355, 0.4645],
        [0.4972, 0.5028],
        [0.2971, 0.7029],
        [0.3370, 0.6630],
        [0.6057, 0.3943],
        [0.3623, 0.6377],
        [0.4232, 0.5768],
        [0.3595, 0.6405],
        [0.3107, 0.6893],
        [0.3555, 0.6445],
        [0.4234, 0.5766]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.6121, 0.3879],
        [0.4168, 0.5832],
        [0.4548, 0.5452],
        [0.5428, 0.4572],
        [0.5358, 0.4642],
        [0.5303, 0.4697],
        [0.3827, 0.6173],
        [0.5777, 0.4223],
        [0.5899, 0.4101],
        [0.3796, 0.6204],
        [0.5688, 0.4312],
        [0.3262, 0.6738]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #50: tensor([[0.5914, 0.4086],
        [0.7405, 0.2595],
        [0.6815, 0.3185],
        [0.2593, 0.7407],
        [0.3972, 0.6028],
        [0.5562, 0.4438],
        [0.4105, 0.5895],
        [0.5439, 0.4561],
        [0.5577, 0.4423],
        [0.6134, 0.3866],
        [0.5616, 0.4384],
        [0.2700, 0.7300]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.6479, 0.3521],
        [0.5622, 0.4378],
        [0.5230, 0.4770],
        [0.5055, 0.4945],
        [0.4993, 0.5007],
        [0.4270, 0.5730],
        [0.7231, 0.2769],
        [0.6083, 0.3917],
        [0.5927, 0.4073],
        [0.3418, 0.6582],
        [0.2335, 0.7665],
        [0.6104, 0.3896]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.2532, 0.7468],
        [0.4748, 0.5252],
        [0.4063, 0.5937],
        [0.5416, 0.4584],
        [0.4775, 0.5225],
        [0.3968, 0.6032],
        [0.4714, 0.5286],
        [0.4159, 0.5841],
        [0.6361, 0.3639],
        [0.4266, 0.5734],
        [0.5340, 0.4660],
        [0.4946, 0.5054]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.4255, 0.5745],
        [0.5642, 0.4358],
        [0.4078, 0.5922],
        [0.3275, 0.6725],
        [0.2507, 0.7493],
        [0.4544, 0.5456],
        [0.6775, 0.3225],
        [0.5191, 0.4809],
        [0.4906, 0.5094],
        [0.4483, 0.5517],
        [0.6143, 0.3857],
        [0.7302, 0.2698]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.4235, 0.5765],
        [0.5308, 0.4692],
        [0.4217, 0.5783],
        [0.6030, 0.3970],
        [0.4770, 0.5230],
        [0.6769, 0.3231],
        [0.4746, 0.5254],
        [0.2522, 0.7478],
        [0.4580, 0.5420],
        [0.4631, 0.5369],
        [0.5448, 0.4552],
        [0.4567, 0.5433]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.7112, 0.2888],
        [0.3122, 0.6878],
        [0.3454, 0.6546],
        [0.3607, 0.6393],
        [0.5819, 0.4181],
        [0.5016, 0.4984],
        [0.5972, 0.4028],
        [0.5018, 0.4982],
        [0.4175, 0.5825],
        [0.5751, 0.4249],
        [0.3734, 0.6266],
        [0.4218, 0.5782]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.7576, 0.2424],
        [0.6401, 0.3599],
        [0.4884, 0.5116],
        [0.4208, 0.5792],
        [0.5823, 0.4177],
        [0.4424, 0.5576],
        [0.2516, 0.7484],
        [0.4604, 0.5396],
        [0.4077, 0.5923],
        [0.5521, 0.4479],
        [0.6907, 0.3093],
        [0.6345, 0.3655]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.5528, 0.4472],
        [0.3296, 0.6704],
        [0.6879, 0.3121],
        [0.4628, 0.5372],
        [0.6425, 0.3575],
        [0.5657, 0.4343],
        [0.6842, 0.3158],
        [0.3655, 0.6345],
        [0.5196, 0.4804],
        [0.4733, 0.5267],
        [0.4724, 0.5276],
        [0.3688, 0.6312]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.5609, 0.4391],
        [0.5892, 0.4108],
        [0.3895, 0.6105],
        [0.4063, 0.5937],
        [0.5771, 0.4229],
        [0.4896, 0.5104],
        [0.2818, 0.7182],
        [0.7291, 0.2709],
        [0.5649, 0.4351],
        [0.5471, 0.4529],
        [0.2831, 0.7169],
        [0.3399, 0.6601]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.5387, 0.4613],
        [0.5862, 0.4138],
        [0.6423, 0.3577],
        [0.3996, 0.6004],
        [0.3127, 0.6873],
        [0.5138, 0.4862],
        [0.4257, 0.5743],
        [0.3577, 0.6423],
        [0.4593, 0.5407],
        [0.4569, 0.5431],
        [0.4524, 0.5476],
        [0.5234, 0.4766]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.5284, 0.4716],
        [0.4937, 0.5063],
        [0.5239, 0.4761],
        [0.6321, 0.3679],
        [0.6414, 0.3586],
        [0.6284, 0.3716],
        [0.4310, 0.5690],
        [0.4249, 0.5751],
        [0.5583, 0.4417],
        [0.3805, 0.6195],
        [0.6000, 0.4000],
        [0.3379, 0.6621]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #50: tensor([[0.5960, 0.4040],
        [0.6062, 0.3938],
        [0.6805, 0.3195],
        [0.4877, 0.5123],
        [0.5794, 0.4206],
        [0.5925, 0.4075],
        [0.4154, 0.5846],
        [0.5834, 0.4166],
        [0.4885, 0.5115],
        [0.5689, 0.4311],
        [0.5958, 0.4042],
        [0.3247, 0.6753]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.5539, 0.4461],
        [0.4502, 0.5498],
        [0.4776, 0.5224],
        [0.5267, 0.4733],
        [0.4505, 0.5495],
        [0.4180, 0.5820],
        [0.5799, 0.4201],
        [0.3957, 0.6043],
        [0.4402, 0.5598],
        [0.5652, 0.4348],
        [0.2206, 0.7794],
        [0.6997, 0.3003]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.3982, 0.6018],
        [0.5259, 0.4741],
        [0.5771, 0.4229],
        [0.4053, 0.5947],
        [0.4829, 0.5171],
        [0.3482, 0.6518],
        [0.5464, 0.4536],
        [0.4319, 0.5681],
        [0.5189, 0.4811],
        [0.5255, 0.4745],
        [0.5761, 0.4239],
        [0.5762, 0.4238]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.4317, 0.5683],
        [0.5464, 0.4536],
        [0.4046, 0.5954],
        [0.4277, 0.5723],
        [0.3839, 0.6161],
        [0.3963, 0.6037],
        [0.8141, 0.1859],
        [0.6146, 0.3854],
        [0.6461, 0.3539],
        [0.4528, 0.5472],
        [0.5321, 0.4679],
        [0.6622, 0.3378]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.4968, 0.5032],
        [0.4238, 0.5762],
        [0.3516, 0.6484],
        [0.5859, 0.4141],
        [0.5533, 0.4467],
        [0.6362, 0.3638],
        [0.5069, 0.4931],
        [0.4637, 0.5363],
        [0.5683, 0.4317],
        [0.3842, 0.6158],
        [0.5025, 0.4975],
        [0.3573, 0.6427]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.7357, 0.2643],
        [0.4309, 0.5691],
        [0.4906, 0.5094],
        [0.3747, 0.6253],
        [0.6182, 0.3818],
        [0.4354, 0.5646],
        [0.5315, 0.4685],
        [0.6822, 0.3178],
        [0.3799, 0.6201],
        [0.4147, 0.5853],
        [0.5019, 0.4981],
        [0.3622, 0.6378]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.4754, 0.5246],
        [0.5789, 0.4211],
        [0.4042, 0.5958],
        [0.4617, 0.5383],
        [0.5535, 0.4465],
        [0.3735, 0.6265],
        [0.3228, 0.6772],
        [0.4275, 0.5725],
        [0.4168, 0.5832],
        [0.5960, 0.4040],
        [0.5304, 0.4696],
        [0.5283, 0.4717]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.3989, 0.6011],
        [0.4931, 0.5069],
        [0.3990, 0.6010],
        [0.5488, 0.4512],
        [0.6473, 0.3527],
        [0.4753, 0.5247],
        [0.6422, 0.3578],
        [0.4497, 0.5503],
        [0.5765, 0.4235],
        [0.5771, 0.4229],
        [0.5993, 0.4007],
        [0.7043, 0.2957]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.5822, 0.4178],
        [0.5145, 0.4855],
        [0.5837, 0.4163],
        [0.5375, 0.4625],
        [0.6823, 0.3177],
        [0.3991, 0.6009],
        [0.4674, 0.5326],
        [0.6801, 0.3199],
        [0.4596, 0.5404],
        [0.5096, 0.4904],
        [0.2736, 0.7264],
        [0.2758, 0.7242]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.4617, 0.5383],
        [0.4771, 0.5229],
        [0.6715, 0.3285],
        [0.4751, 0.5249],
        [0.4162, 0.5838],
        [0.5621, 0.4379],
        [0.4283, 0.5717],
        [0.3497, 0.6503],
        [0.5321, 0.4679],
        [0.4908, 0.5092],
        [0.5404, 0.4596],
        [0.5428, 0.4572]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.6764, 0.3236],
        [0.5263, 0.4737],
        [0.4172, 0.5828],
        [0.3729, 0.6271],
        [0.6102, 0.3898],
        [0.5177, 0.4823],
        [0.2936, 0.7064],
        [0.5879, 0.4121],
        [0.5096, 0.4904],
        [0.3970, 0.6030],
        [0.6250, 0.3750],
        [0.3228, 0.6772]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #50: tensor([[0.6002, 0.3998],
        [0.7675, 0.2325],
        [0.6283, 0.3717],
        [0.3773, 0.6227],
        [0.4153, 0.5847],
        [0.4626, 0.5374],
        [0.4209, 0.5791],
        [0.5085, 0.4915],
        [0.4830, 0.5170],
        [0.4754, 0.5246],
        [0.5277, 0.4723],
        [0.4970, 0.5030]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.6212, 0.3788],
        [0.4081, 0.5919],
        [0.4496, 0.5504],
        [0.4294, 0.5706],
        [0.6622, 0.3378],
        [0.6668, 0.3332],
        [0.6043, 0.3957],
        [0.5722, 0.4278],
        [0.4716, 0.5284],
        [0.4963, 0.5037],
        [0.2669, 0.7331],
        [0.4695, 0.5305]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.4877, 0.5123],
        [0.4454, 0.5546],
        [0.4495, 0.5505],
        [0.4539, 0.5461],
        [0.5959, 0.4041],
        [0.4220, 0.5780],
        [0.4155, 0.5845],
        [0.3714, 0.6286],
        [0.5208, 0.4792],
        [0.5933, 0.4067],
        [0.4844, 0.5156],
        [0.4767, 0.5233]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.4548, 0.5452],
        [0.4136, 0.5864],
        [0.4960, 0.5040],
        [0.3991, 0.6009],
        [0.3430, 0.6569],
        [0.4234, 0.5766],
        [0.5917, 0.4083],
        [0.5956, 0.4044],
        [0.5275, 0.4725],
        [0.5546, 0.4454],
        [0.4540, 0.5460],
        [0.6325, 0.3675]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.6422, 0.3578],
        [0.4078, 0.5922],
        [0.4730, 0.5270],
        [0.5124, 0.4876],
        [0.5284, 0.4716],
        [0.5265, 0.4735],
        [0.5966, 0.4034],
        [0.3600, 0.6400],
        [0.4718, 0.5282],
        [0.5153, 0.4847],
        [0.5904, 0.4096],
        [0.3670, 0.6330]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.4574, 0.5426],
        [0.4705, 0.5295],
        [0.3260, 0.6740],
        [0.2907, 0.7093],
        [0.3255, 0.6745],
        [0.5536, 0.4464],
        [0.5607, 0.4393],
        [0.5489, 0.4511],
        [0.4901, 0.5099],
        [0.4357, 0.5643],
        [0.5381, 0.4619],
        [0.4025, 0.5975]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.4734, 0.5266],
        [0.7153, 0.2847],
        [0.3883, 0.6117],
        [0.4077, 0.5923],
        [0.5877, 0.4123],
        [0.3946, 0.6054],
        [0.3516, 0.6484],
        [0.4941, 0.5059],
        [0.4022, 0.5978],
        [0.5162, 0.4838],
        [0.6519, 0.3481],
        [0.6569, 0.3431]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.4938, 0.5062],
        [0.4534, 0.5466],
        [0.5878, 0.4122],
        [0.4198, 0.5802],
        [0.6083, 0.3917],
        [0.6034, 0.3966],
        [0.4578, 0.5422],
        [0.5203, 0.4797],
        [0.5620, 0.4380],
        [0.6529, 0.3471],
        [0.6105, 0.3895],
        [0.5191, 0.4809]], device='cuda:0', grad_fn=<SoftmaxBackward>)
