Iter #50: tensor([[0.5923, 0.4077],
        [0.5441, 0.4559],
        [0.6170, 0.3830],
        [0.6312, 0.3688],
        [0.6492, 0.3508],
        [0.5800, 0.4200],
        [0.5355, 0.4645],
        [0.5783, 0.4217],
        [0.6111, 0.3889],
        [0.5568, 0.4432],
        [0.5993, 0.4007],
        [0.5841, 0.4159]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.4503, 0.5497],
        [0.5675, 0.4325],
        [0.4255, 0.5745],
        [0.4930, 0.5070],
        [0.4895, 0.5105],
        [0.5971, 0.4029],
        [0.5416, 0.4584],
        [0.3923, 0.6077],
        [0.6095, 0.3905],
        [0.5749, 0.4251],
        [0.5434, 0.4566],
        [0.5555, 0.4445]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.4494, 0.5506],
        [0.5811, 0.4189],
        [0.4344, 0.5656],
        [0.6793, 0.3207],
        [0.6542, 0.3458],
        [0.4021, 0.5979],
        [0.5278, 0.4722],
        [0.5844, 0.4156],
        [0.3800, 0.6200],
        [0.5218, 0.4782],
        [0.4874, 0.5126],
        [0.4556, 0.5444]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.6299, 0.3701],
        [0.4453, 0.5547],
        [0.4538, 0.5462],
        [0.4296, 0.5704],
        [0.4186, 0.5814],
        [0.5938, 0.4062],
        [0.4597, 0.5403],
        [0.6275, 0.3725],
        [0.5302, 0.4698],
        [0.4675, 0.5325],
        [0.6152, 0.3848],
        [0.4810, 0.5190]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.5692, 0.4308],
        [0.4958, 0.5042],
        [0.4648, 0.5352],
        [0.4798, 0.5202],
        [0.4800, 0.5200],
        [0.6388, 0.3612],
        [0.3977, 0.6023],
        [0.5016, 0.4984],
        [0.4660, 0.5340],
        [0.4702, 0.5298],
        [0.5061, 0.4939],
        [0.5577, 0.4423]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.5204, 0.4796],
        [0.5412, 0.4588],
        [0.6543, 0.3457],
        [0.5128, 0.4872],
        [0.4537, 0.5463],
        [0.4841, 0.5159],
        [0.5895, 0.4105],
        [0.6261, 0.3739],
        [0.2811, 0.7189],
        [0.4799, 0.5201],
        [0.4985, 0.5015],
        [0.5506, 0.4494]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.4568, 0.5432],
        [0.6872, 0.3128],
        [0.5671, 0.4329],
        [0.4088, 0.5912],
        [0.6217, 0.3783],
        [0.4235, 0.5765],
        [0.6040, 0.3960],
        [0.4386, 0.5614],
        [0.4447, 0.5553],
        [0.5173, 0.4827],
        [0.4951, 0.5049],
        [0.5600, 0.4400]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.4121, 0.5879],
        [0.6509, 0.3491],
        [0.5590, 0.4410],
        [0.4074, 0.5926],
        [0.3972, 0.6028],
        [0.4191, 0.5809],
        [0.5332, 0.4668],
        [0.5232, 0.4768],
        [0.4916, 0.5084],
        [0.4190, 0.5810],
        [0.5988, 0.4012],
        [0.4958, 0.5042]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.3745, 0.6255],
        [0.4598, 0.5402],
        [0.6285, 0.3715],
        [0.4930, 0.5070],
        [0.6548, 0.3452],
        [0.5135, 0.4865],
        [0.4830, 0.5170],
        [0.6643, 0.3357],
        [0.4406, 0.5594],
        [0.4923, 0.5077],
        [0.6261, 0.3739],
        [0.5703, 0.4297]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.4179, 0.5821],
        [0.6139, 0.3861],
        [0.5218, 0.4782],
        [0.3730, 0.6270],
        [0.3416, 0.6584],
        [0.4161, 0.5839],
        [0.4428, 0.5572],
        [0.4688, 0.5312],
        [0.5976, 0.4024],
        [0.4660, 0.5340],
        [0.4878, 0.5122],
        [0.4529, 0.5471]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.5388, 0.4612],
        [0.6113, 0.3887],
        [0.5847, 0.4153],
        [0.4001, 0.5999],
        [0.5151, 0.4849],
        [0.3937, 0.6063],
        [0.2972, 0.7028],
        [0.5857, 0.4143],
        [0.6363, 0.3637],
        [0.4220, 0.5780],
        [0.5357, 0.4643],
        [0.5149, 0.4851]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #50: tensor([[0.3922, 0.6078],
        [0.4226, 0.5774],
        [0.4236, 0.5764],
        [0.4759, 0.5241],
        [0.4725, 0.5275],
        [0.5675, 0.4325],
        [0.4570, 0.5430],
        [0.6429, 0.3571],
        [0.6625, 0.3375],
        [0.6739, 0.3261],
        [0.4411, 0.5589],
        [0.7030, 0.2970]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.3942, 0.6058],
        [0.5852, 0.4148],
        [0.3777, 0.6223],
        [0.4611, 0.5389],
        [0.4917, 0.5083],
        [0.6195, 0.3805],
        [0.5959, 0.4041],
        [0.3427, 0.6573],
        [0.4580, 0.5420],
        [0.4704, 0.5296],
        [0.6716, 0.3284],
        [0.6289, 0.3711]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.4224, 0.5776],
        [0.5136, 0.4864],
        [0.3522, 0.6478],
        [0.6351, 0.3649],
        [0.7189, 0.2811],
        [0.2982, 0.7018],
        [0.6370, 0.3630],
        [0.5796, 0.4204],
        [0.2907, 0.7093],
        [0.5964, 0.4036],
        [0.4960, 0.5040],
        [0.5282, 0.4718]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.6240, 0.3760],
        [0.6864, 0.3136],
        [0.3502, 0.6498],
        [0.4912, 0.5088],
        [0.4798, 0.5202],
        [0.6145, 0.3855],
        [0.5834, 0.4166],
        [0.5726, 0.4274],
        [0.3775, 0.6225],
        [0.3852, 0.6148],
        [0.6666, 0.3334],
        [0.4514, 0.5486]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.7088, 0.2912],
        [0.4732, 0.5268],
        [0.5800, 0.4200],
        [0.5653, 0.4347],
        [0.5283, 0.4717],
        [0.5896, 0.4104],
        [0.5022, 0.4978],
        [0.3301, 0.6699],
        [0.5688, 0.4312],
        [0.6394, 0.3606],
        [0.5471, 0.4529],
        [0.5122, 0.4878]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.4483, 0.5517],
        [0.5311, 0.4689],
        [0.4385, 0.5615],
        [0.6825, 0.3175],
        [0.4620, 0.5380],
        [0.5012, 0.4988],
        [0.5535, 0.4465],
        [0.5566, 0.4434],
        [0.2795, 0.7205],
        [0.4353, 0.5647],
        [0.4357, 0.5643],
        [0.5299, 0.4701]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.2635, 0.7365],
        [0.6865, 0.3135],
        [0.4044, 0.5956],
        [0.4336, 0.5664],
        [0.7434, 0.2566],
        [0.3873, 0.6127],
        [0.5664, 0.4336],
        [0.3954, 0.6046],
        [0.3569, 0.6431],
        [0.5163, 0.4837],
        [0.4093, 0.5907],
        [0.5002, 0.4998]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.5501, 0.4499],
        [0.5284, 0.4716],
        [0.6133, 0.3867],
        [0.4141, 0.5859],
        [0.4763, 0.5237],
        [0.5766, 0.4234],
        [0.3626, 0.6374],
        [0.3933, 0.6067],
        [0.3854, 0.6146],
        [0.3034, 0.6966],
        [0.6222, 0.3778],
        [0.5923, 0.4077]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.4909, 0.5091],
        [0.3682, 0.6318],
        [0.4387, 0.5613],
        [0.4795, 0.5205],
        [0.6407, 0.3593],
        [0.5447, 0.4553],
        [0.3261, 0.6739],
        [0.3747, 0.6253],
        [0.5430, 0.4570],
        [0.6922, 0.3078],
        [0.6532, 0.3468],
        [0.6470, 0.3530]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.5323, 0.4677],
        [0.4452, 0.5548],
        [0.5297, 0.4703],
        [0.3527, 0.6473],
        [0.5229, 0.4771],
        [0.3860, 0.6140],
        [0.5012, 0.4988],
        [0.3849, 0.6151],
        [0.4980, 0.5020],
        [0.5722, 0.4278],
        [0.5980, 0.4020],
        [0.4400, 0.5600]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.6266, 0.3734],
        [0.7254, 0.2746],
        [0.4911, 0.5089],
        [0.3994, 0.6006],
        [0.4886, 0.5114],
        [0.4118, 0.5882],
        [0.3114, 0.6886],
        [0.5989, 0.4011],
        [0.6253, 0.3747],
        [0.4904, 0.5096],
        [0.5765, 0.4235],
        [0.6006, 0.3994]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #50: tensor([[0.5762, 0.4238],
        [0.4481, 0.5519],
        [0.5708, 0.4292],
        [0.5099, 0.4901],
        [0.4457, 0.5543],
        [0.7001, 0.2999],
        [0.3891, 0.6109],
        [0.4866, 0.5134],
        [0.5368, 0.4632],
        [0.7094, 0.2906],
        [0.3404, 0.6596],
        [0.5093, 0.4907]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.5806, 0.4194],
        [0.5953, 0.4047],
        [0.3906, 0.6094],
        [0.4412, 0.5588],
        [0.3834, 0.6166],
        [0.4525, 0.5475],
        [0.6015, 0.3985],
        [0.3456, 0.6544],
        [0.5415, 0.4585],
        [0.5658, 0.4342],
        [0.3492, 0.6508],
        [0.4632, 0.5368]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.4368, 0.5632],
        [0.5731, 0.4269],
        [0.3250, 0.6750],
        [0.7068, 0.2932],
        [0.6198, 0.3802],
        [0.4286, 0.5714],
        [0.6550, 0.3450],
        [0.6668, 0.3332],
        [0.2041, 0.7959],
        [0.6945, 0.3055],
        [0.5397, 0.4603],
        [0.5753, 0.4247]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.5968, 0.4032],
        [0.5018, 0.4982],
        [0.4215, 0.5785],
        [0.3197, 0.6803],
        [0.5108, 0.4892],
        [0.5284, 0.4716],
        [0.3898, 0.6102],
        [0.5184, 0.4816],
        [0.3650, 0.6350],
        [0.4737, 0.5263],
        [0.6056, 0.3944],
        [0.4451, 0.5549]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.6413, 0.3587],
        [0.6029, 0.3971],
        [0.5913, 0.4087],
        [0.4184, 0.5816],
        [0.4435, 0.5565],
        [0.6207, 0.3793],
        [0.4778, 0.5222],
        [0.4100, 0.5900],
        [0.5731, 0.4269],
        [0.4980, 0.5020],
        [0.4392, 0.5608],
        [0.7006, 0.2994]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.5748, 0.4252],
        [0.4746, 0.5254],
        [0.4630, 0.5370],
        [0.4818, 0.5182],
        [0.5926, 0.4074],
        [0.5351, 0.4649],
        [0.6094, 0.3906],
        [0.6471, 0.3529],
        [0.2262, 0.7738],
        [0.4287, 0.5713],
        [0.3584, 0.6416],
        [0.5016, 0.4984]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.2652, 0.7348],
        [0.6044, 0.3956],
        [0.6392, 0.3608],
        [0.3412, 0.6588],
        [0.5080, 0.4920],
        [0.3820, 0.6180],
        [0.5106, 0.4894],
        [0.4503, 0.5497],
        [0.4786, 0.5214],
        [0.4599, 0.5401],
        [0.4786, 0.5214],
        [0.5120, 0.4880]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.5349, 0.4651],
        [0.7068, 0.2932],
        [0.7347, 0.2653],
        [0.4387, 0.5613],
        [0.4391, 0.5609],
        [0.3732, 0.6268],
        [0.4029, 0.5971],
        [0.4814, 0.5186],
        [0.5188, 0.4812],
        [0.4813, 0.5187],
        [0.6120, 0.3880],
        [0.5076, 0.4924]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.5279, 0.4721],
        [0.4175, 0.5825],
        [0.4558, 0.5442],
        [0.5451, 0.4549],
        [0.5841, 0.4159],
        [0.5649, 0.4351],
        [0.3922, 0.6078],
        [0.6104, 0.3896],
        [0.4702, 0.5298],
        [0.6628, 0.3372],
        [0.7012, 0.2988],
        [0.4415, 0.5585]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.3999, 0.6001],
        [0.6063, 0.3937],
        [0.3954, 0.6046],
        [0.3886, 0.6114],
        [0.3732, 0.6268],
        [0.3814, 0.6186],
        [0.6168, 0.3832],
        [0.4430, 0.5570],
        [0.4399, 0.5601],
        [0.4373, 0.5627],
        [0.5453, 0.4547],
        [0.5522, 0.4478]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.5470, 0.4530],
        [0.7120, 0.2880],
        [0.4886, 0.5114],
        [0.4530, 0.5470],
        [0.4416, 0.5584],
        [0.5425, 0.4575],
        [0.3150, 0.6850],
        [0.5877, 0.4123],
        [0.5019, 0.4981],
        [0.3546, 0.6454],
        [0.5815, 0.4185],
        [0.3926, 0.6074]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #50: tensor([[0.4769, 0.5231],
        [0.5432, 0.4568],
        [0.6263, 0.3737],
        [0.5332, 0.4668],
        [0.4069, 0.5931],
        [0.5352, 0.4648],
        [0.5178, 0.4822],
        [0.6148, 0.3852],
        [0.4374, 0.5626],
        [0.6428, 0.3572],
        [0.5220, 0.4780],
        [0.6258, 0.3742]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.4191, 0.5809],
        [0.6765, 0.3235],
        [0.3329, 0.6671],
        [0.5427, 0.4573],
        [0.5885, 0.4115],
        [0.5087, 0.4913],
        [0.4825, 0.5175],
        [0.4732, 0.5268],
        [0.5023, 0.4977],
        [0.5470, 0.4530],
        [0.4221, 0.5779],
        [0.6248, 0.3752]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.3659, 0.6341],
        [0.5259, 0.4741],
        [0.3251, 0.6749],
        [0.6380, 0.3620],
        [0.5862, 0.4138],
        [0.3410, 0.6590],
        [0.5460, 0.4540],
        [0.4154, 0.5846],
        [0.4276, 0.5724],
        [0.5826, 0.4174],
        [0.6752, 0.3248],
        [0.4054, 0.5946]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.5891, 0.4109],
        [0.6133, 0.3867],
        [0.4066, 0.5934],
        [0.4835, 0.5165],
        [0.5340, 0.4660],
        [0.5135, 0.4865],
        [0.4646, 0.5354],
        [0.5200, 0.4800],
        [0.5100, 0.4900],
        [0.5550, 0.4450],
        [0.6814, 0.3186],
        [0.3658, 0.6342]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.4984, 0.5016],
        [0.4495, 0.5505],
        [0.5246, 0.4754],
        [0.5041, 0.4959],
        [0.5509, 0.4491],
        [0.6328, 0.3672],
        [0.5247, 0.4753],
        [0.5496, 0.4504],
        [0.4762, 0.5238],
        [0.4825, 0.5175],
        [0.4857, 0.5143],
        [0.6069, 0.3931]], device='cuda:0', grad_fn=<SoftmaxBackward>)
