Iter #50: tensor([[0.5898, 0.4102],
        [0.6089, 0.3911],
        [0.5478, 0.4522],
        [0.5757, 0.4243],
        [0.5608, 0.4392],
        [0.5378, 0.4622],
        [0.6253, 0.3747],
        [0.6068, 0.3932],
        [0.5345, 0.4655],
        [0.6256, 0.3744],
        [0.5599, 0.4401],
        [0.5425, 0.4575]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.5343, 0.4657],
        [0.6046, 0.3954],
        [0.6169, 0.3831],
        [0.6526, 0.3474],
        [0.6544, 0.3456],
        [0.6214, 0.3786],
        [0.5604, 0.4396],
        [0.6149, 0.3851],
        [0.5579, 0.4421],
        [0.6295, 0.3705],
        [0.5711, 0.4289],
        [0.5827, 0.4173]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.5256, 0.4744],
        [0.5540, 0.4460],
        [0.5855, 0.4145],
        [0.5585, 0.4415],
        [0.5431, 0.4569],
        [0.5430, 0.4570],
        [0.5964, 0.4036],
        [0.6485, 0.3515],
        [0.4201, 0.5799],
        [0.5509, 0.4491],
        [0.4543, 0.5457],
        [0.6169, 0.3831]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.6443, 0.3557],
        [0.5641, 0.4359],
        [0.5989, 0.4011],
        [0.5924, 0.4076],
        [0.5562, 0.4438],
        [0.5492, 0.4508],
        [0.6019, 0.3981],
        [0.4086, 0.5914],
        [0.6436, 0.3564],
        [0.5844, 0.4156],
        [0.5951, 0.4049],
        [0.4789, 0.5211]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.5411, 0.4589],
        [0.4948, 0.5052],
        [0.6093, 0.3907],
        [0.6189, 0.3811],
        [0.6014, 0.3986],
        [0.6408, 0.3592],
        [0.6253, 0.3747],
        [0.6040, 0.3960],
        [0.6333, 0.3667],
        [0.5751, 0.4249],
        [0.6121, 0.3879],
        [0.6355, 0.3645]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.5800, 0.4200],
        [0.5990, 0.4010],
        [0.5424, 0.4576],
        [0.5873, 0.4127],
        [0.5129, 0.4871],
        [0.5228, 0.4772],
        [0.6122, 0.3878],
        [0.5909, 0.4091],
        [0.4166, 0.5834],
        [0.5098, 0.4902],
        [0.6029, 0.3971],
        [0.5486, 0.4514]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.5307, 0.4693],
        [0.5271, 0.4729],
        [0.4114, 0.5886],
        [0.4814, 0.5186],
        [0.5500, 0.4500],
        [0.4962, 0.5038],
        [0.4054, 0.5946],
        [0.3451, 0.6549],
        [0.4744, 0.5256],
        [0.5647, 0.4353],
        [0.5788, 0.4212],
        [0.5142, 0.4858]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.4435, 0.5565],
        [0.5591, 0.4409],
        [0.4809, 0.5191],
        [0.5006, 0.4994],
        [0.5356, 0.4644],
        [0.5567, 0.4433],
        [0.5512, 0.4488],
        [0.5421, 0.4579],
        [0.4994, 0.5006],
        [0.4265, 0.5735],
        [0.5562, 0.4438],
        [0.5738, 0.4262]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.5359, 0.4641],
        [0.6131, 0.3869],
        [0.5233, 0.4767],
        [0.6093, 0.3907],
        [0.5664, 0.4336],
        [0.5326, 0.4674],
        [0.5068, 0.4932],
        [0.5361, 0.4639],
        [0.5692, 0.4308],
        [0.5380, 0.4620],
        [0.6050, 0.3950],
        [0.5379, 0.4621]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.4221, 0.5779],
        [0.5174, 0.4826],
        [0.5257, 0.4743],
        [0.5642, 0.4358],
        [0.5028, 0.4972],
        [0.6566, 0.3434],
        [0.5363, 0.4637],
        [0.6109, 0.3891],
        [0.4807, 0.5193],
        [0.4020, 0.5980],
        [0.5877, 0.4123],
        [0.5928, 0.4072]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.5028, 0.4972],
        [0.6282, 0.3718],
        [0.5165, 0.4835],
        [0.5285, 0.4715],
        [0.5323, 0.4677],
        [0.4350, 0.5650],
        [0.4952, 0.5048],
        [0.5732, 0.4268],
        [0.5400, 0.4600],
        [0.4632, 0.5368],
        [0.5706, 0.4294],
        [0.6244, 0.3756]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #50: tensor([[0.5859, 0.4141],
        [0.4521, 0.5479],
        [0.5630, 0.4370],
        [0.6094, 0.3906],
        [0.4992, 0.5008],
        [0.4200, 0.5800],
        [0.5592, 0.4408],
        [0.5801, 0.4199],
        [0.5955, 0.4045],
        [0.6080, 0.3920],
        [0.3864, 0.6136],
        [0.4245, 0.5755]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.5128, 0.4872],
        [0.5532, 0.4468],
        [0.4587, 0.5413],
        [0.4069, 0.5931],
        [0.5649, 0.4351],
        [0.4495, 0.5505],
        [0.4761, 0.5239],
        [0.5857, 0.4143],
        [0.4691, 0.5309],
        [0.4451, 0.5549],
        [0.6569, 0.3431],
        [0.5498, 0.4502]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.5462, 0.4538],
        [0.6222, 0.3778],
        [0.4286, 0.5714],
        [0.6747, 0.3253],
        [0.4972, 0.5028],
        [0.4539, 0.5461],
        [0.5957, 0.4043],
        [0.5271, 0.4729],
        [0.5095, 0.4905],
        [0.5149, 0.4851],
        [0.4810, 0.5190],
        [0.4569, 0.5431]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.4528, 0.5472],
        [0.3160, 0.6840],
        [0.4091, 0.5909],
        [0.3508, 0.6492],
        [0.5570, 0.4430],
        [0.4029, 0.5971],
        [0.4940, 0.5060],
        [0.4794, 0.5206],
        [0.4927, 0.5073],
        [0.5081, 0.4919],
        [0.5351, 0.4649],
        [0.3848, 0.6152]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.5690, 0.4310],
        [0.4741, 0.5259],
        [0.5284, 0.4716],
        [0.5666, 0.4334],
        [0.6200, 0.3800],
        [0.5589, 0.4411],
        [0.5949, 0.4051],
        [0.3746, 0.6254],
        [0.4094, 0.5906],
        [0.5374, 0.4626],
        [0.4742, 0.5258],
        [0.5224, 0.4776]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.5094, 0.4906],
        [0.5409, 0.4591],
        [0.4706, 0.5294],
        [0.4127, 0.5873],
        [0.3692, 0.6308],
        [0.4840, 0.5160],
        [0.5185, 0.4815],
        [0.5477, 0.4523],
        [0.5729, 0.4271],
        [0.5486, 0.4514],
        [0.4922, 0.5078],
        [0.4772, 0.5228]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.5322, 0.4678],
        [0.3054, 0.6946],
        [0.2995, 0.7005],
        [0.4901, 0.5099],
        [0.5602, 0.4398],
        [0.5407, 0.4593],
        [0.4734, 0.5266],
        [0.6248, 0.3752],
        [0.5008, 0.4992],
        [0.5692, 0.4308],
        [0.5423, 0.4577],
        [0.4712, 0.5288]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.4863, 0.5137],
        [0.5236, 0.4764],
        [0.6306, 0.3694],
        [0.5565, 0.4435],
        [0.5203, 0.4797],
        [0.5830, 0.4170],
        [0.4946, 0.5054],
        [0.4858, 0.5142],
        [0.4794, 0.5206],
        [0.5238, 0.4762],
        [0.5212, 0.4788],
        [0.7804, 0.2196]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.5764, 0.4236],
        [0.4962, 0.5038],
        [0.5411, 0.4589],
        [0.3862, 0.6138],
        [0.3441, 0.6559],
        [0.4226, 0.5774],
        [0.4932, 0.5068],
        [0.5494, 0.4506],
        [0.3995, 0.6005],
        [0.4966, 0.5034],
        [0.4267, 0.5733],
        [0.5109, 0.4891]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.4633, 0.5367],
        [0.5070, 0.4930],
        [0.3258, 0.6742],
        [0.5872, 0.4128],
        [0.4668, 0.5332],
        [0.4199, 0.5801],
        [0.5744, 0.4256],
        [0.5331, 0.4669],
        [0.5124, 0.4876],
        [0.3950, 0.6050],
        [0.6075, 0.3925],
        [0.5140, 0.4860]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.3705, 0.6295],
        [0.4973, 0.5027],
        [0.3752, 0.6248],
        [0.4333, 0.5667],
        [0.4094, 0.5906],
        [0.3271, 0.6729],
        [0.3771, 0.6229],
        [0.3692, 0.6308],
        [0.5422, 0.4578],
        [0.3893, 0.6107],
        [0.4724, 0.5276],
        [0.3203, 0.6797]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #50: tensor([[0.5292, 0.4708],
        [0.5050, 0.4950],
        [0.2700, 0.7300],
        [0.5561, 0.4439],
        [0.5707, 0.4293],
        [0.4439, 0.5561],
        [0.4732, 0.5268],
        [0.5501, 0.4499],
        [0.4393, 0.5607],
        [0.4413, 0.5587],
        [0.2548, 0.7452],
        [0.4940, 0.5060]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.5140, 0.4860],
        [0.6137, 0.3863],
        [0.4986, 0.5014],
        [0.3738, 0.6262],
        [0.4590, 0.5410],
        [0.5016, 0.4984],
        [0.4062, 0.5938],
        [0.3519, 0.6481],
        [0.5370, 0.4630],
        [0.3276, 0.6724],
        [0.3911, 0.6089],
        [0.5941, 0.4059]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.5761, 0.4239],
        [0.5001, 0.4999],
        [0.5315, 0.4685],
        [0.5096, 0.4904],
        [0.5138, 0.4862],
        [0.3817, 0.6183],
        [0.4755, 0.5245],
        [0.3968, 0.6032],
        [0.4885, 0.5115],
        [0.4860, 0.5140],
        [0.6238, 0.3762],
        [0.3294, 0.6706]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.4944, 0.5056],
        [0.4172, 0.5828],
        [0.6643, 0.3357],
        [0.3479, 0.6521],
        [0.6218, 0.3782],
        [0.4980, 0.5020],
        [0.3773, 0.6227],
        [0.2648, 0.7352],
        [0.3817, 0.6183],
        [0.5784, 0.4216],
        [0.6217, 0.3783],
        [0.3642, 0.6358]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.4670, 0.5330],
        [0.5529, 0.4471],
        [0.4937, 0.5063],
        [0.5142, 0.4858],
        [0.7340, 0.2660],
        [0.4291, 0.5709],
        [0.4553, 0.5447],
        [0.4415, 0.5585],
        [0.3901, 0.6099],
        [0.6319, 0.3681],
        [0.4685, 0.5315],
        [0.6158, 0.3842]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.5934, 0.4066],
        [0.4893, 0.5107],
        [0.3834, 0.6166],
        [0.2762, 0.7238],
        [0.3534, 0.6466],
        [0.3977, 0.6023],
        [0.5993, 0.4007],
        [0.5765, 0.4235],
        [0.4516, 0.5484],
        [0.7062, 0.2938],
        [0.5380, 0.4620],
        [0.6255, 0.3745]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.4042, 0.5958],
        [0.5412, 0.4588],
        [0.4165, 0.5835],
        [0.5691, 0.4309],
        [0.6265, 0.3735],
        [0.3838, 0.6162],
        [0.4399, 0.5601],
        [0.3520, 0.6480],
        [0.4799, 0.5201],
        [0.4954, 0.5046],
        [0.5006, 0.4994],
        [0.5745, 0.4255]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.3811, 0.6189],
        [0.4871, 0.5129],
        [0.5393, 0.4607],
        [0.4689, 0.5311],
        [0.4237, 0.5763],
        [0.4891, 0.5109],
        [0.5965, 0.4035],
        [0.5080, 0.4920],
        [0.6098, 0.3902],
        [0.5249, 0.4751],
        [0.4521, 0.5479],
        [0.5328, 0.4672]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.4541, 0.5459],
        [0.4803, 0.5197],
        [0.5007, 0.4993],
        [0.4222, 0.5778],
        [0.3429, 0.6571],
        [0.4509, 0.5491],
        [0.5547, 0.4453],
        [0.5066, 0.4934],
        [0.4477, 0.5523],
        [0.3865, 0.6135],
        [0.4582, 0.5418],
        [0.4844, 0.5156]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.2165, 0.7835],
        [0.3690, 0.6310],
        [0.4208, 0.5792],
        [0.6027, 0.3973],
        [0.3552, 0.6448],
        [0.4739, 0.5261],
        [0.4743, 0.5257],
        [0.4078, 0.5922],
        [0.4051, 0.5949],
        [0.2734, 0.7266],
        [0.5111, 0.4889],
        [0.5269, 0.4731]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.3956, 0.6044],
        [0.4939, 0.5061],
        [0.3831, 0.6169],
        [0.4906, 0.5094],
        [0.4341, 0.5659],
        [0.5117, 0.4883],
        [0.5057, 0.4943],
        [0.4009, 0.5991],
        [0.5033, 0.4967],
        [0.4782, 0.5218],
        [0.4061, 0.5939],
        [0.3849, 0.6151]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #50: tensor([[0.5355, 0.4645],
        [0.4701, 0.5299],
        [0.6220, 0.3780],
        [0.4915, 0.5085],
        [0.3954, 0.6046],
        [0.3680, 0.6320],
        [0.4610, 0.5390],
        [0.4882, 0.5118],
        [0.5950, 0.4050],
        [0.6637, 0.3363],
        [0.2781, 0.7219],
        [0.3932, 0.6068]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #100: tensor([[0.4382, 0.5618],
        [0.5601, 0.4399],
        [0.5493, 0.4507],
        [0.2451, 0.7549],
        [0.4279, 0.5721],
        [0.4590, 0.5410],
        [0.4935, 0.5065],
        [0.5050, 0.4950],
        [0.5469, 0.4531],
        [0.3617, 0.6383],
        [0.5622, 0.4378],
        [0.5847, 0.4153]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #150: tensor([[0.5188, 0.4812],
        [0.6715, 0.3285],
        [0.4321, 0.5679],
        [0.5253, 0.4747],
        [0.5479, 0.4521],
        [0.3958, 0.6042],
        [0.3880, 0.6120],
        [0.3806, 0.6194],
        [0.4039, 0.5961],
        [0.5109, 0.4891],
        [0.5317, 0.4683],
        [0.2923, 0.7077]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #200: tensor([[0.3726, 0.6274],
        [0.3203, 0.6797],
        [0.2345, 0.7655],
        [0.4125, 0.5875],
        [0.4434, 0.5566],
        [0.3528, 0.6472],
        [0.5096, 0.4904],
        [0.3580, 0.6420],
        [0.4086, 0.5914],
        [0.7453, 0.2547],
        [0.4334, 0.5666],
        [0.4412, 0.5588]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #250: tensor([[0.3247, 0.6753],
        [0.3980, 0.6020],
        [0.4865, 0.5135],
        [0.4237, 0.5763],
        [0.5794, 0.4206],
        [0.4307, 0.5693],
        [0.5695, 0.4305],
        [0.5008, 0.4992],
        [0.3874, 0.6126],
        [0.5576, 0.4424],
        [0.4780, 0.5220],
        [0.4806, 0.5194]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #300: tensor([[0.4810, 0.5190],
        [0.6245, 0.3755],
        [0.6880, 0.3120],
        [0.3611, 0.6389],
        [0.5226, 0.4774],
        [0.2977, 0.7023],
        [0.5505, 0.4495],
        [0.5541, 0.4459],
        [0.2559, 0.7441],
        [0.5145, 0.4855],
        [0.5681, 0.4319],
        [0.6895, 0.3105]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #350: tensor([[0.3880, 0.6120],
        [0.3559, 0.6441],
        [0.4334, 0.5666],
        [0.5457, 0.4543],
        [0.7414, 0.2586],
        [0.5291, 0.4709],
        [0.5070, 0.4930],
        [0.6866, 0.3134],
        [0.4517, 0.5483],
        [0.4987, 0.5013],
        [0.7563, 0.2437],
        [0.6651, 0.3349]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #400: tensor([[0.5843, 0.4157],
        [0.4328, 0.5672],
        [0.4465, 0.5535],
        [0.4151, 0.5849],
        [0.5158, 0.4842],
        [0.5862, 0.4138],
        [0.5194, 0.4806],
        [0.4337, 0.5663],
        [0.5747, 0.4253],
        [0.6299, 0.3701],
        [0.5643, 0.4357],
        [0.6839, 0.3161]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #450: tensor([[0.7332, 0.2668],
        [0.5732, 0.4268],
        [0.6965, 0.3035],
        [0.3950, 0.6050],
        [0.5738, 0.4262],
        [0.4793, 0.5207],
        [0.6055, 0.3945],
        [0.4540, 0.5460],
        [0.5282, 0.4718],
        [0.4392, 0.5608],
        [0.4100, 0.5900],
        [0.6619, 0.3381]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #500: tensor([[0.3987, 0.6013],
        [0.5027, 0.4973],
        [0.5065, 0.4935],
        [0.5389, 0.4611],
        [0.4792, 0.5208],
        [0.4162, 0.5838],
        [0.4646, 0.5354],
        [0.4389, 0.5611],
        [0.3653, 0.6347],
        [0.1457, 0.8543],
        [0.6240, 0.3760],
        [0.5774, 0.4226]], device='cuda:0', grad_fn=<SoftmaxBackward>)
Iter #550: tensor([[0.3659, 0.6341],
        [0.6222, 0.3778],
        [0.5105, 0.4895],
        [0.5956, 0.4044],
        [0.4761, 0.5239],
        [0.5192, 0.4808],
        [0.4933, 0.5067],
        [0.4857, 0.5143],
        [0.5036, 0.4964],
        [0.4963, 0.5037],
        [0.5001, 0.4999],
        [0.6178, 0.3822]], device='cuda:0', grad_fn=<SoftmaxBackward>)
