ResNet(
  3.84 M, 104.771% Params, 245.82 MMac, 100.000% MACs, 
  (conv1): Conv2d(432, 0.012% Params, 442.37 KMac, 0.180% MACs, 3, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (bn1): BatchNorm2d(32, 0.001% Params, 32.77 KMac, 0.013% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (relu): ReLU(0, 0.000% Params, 16.38 KMac, 0.007% MACs, )
  (layer1): MakeLayer(
    15.17 k, 0.414% Params, 15.83 MMac, 6.439% MACs, 
    (downsample): Sequential(
      1.15 k, 0.031% Params, 1.18 MMac, 0.480% MACs, 
      (0): Conv2d(1.02 k, 0.028% Params, 1.05 MMac, 0.427% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (1): BatchNorm2d(128, 0.003% Params, 131.07 KMac, 0.053% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (blockdict): ModuleDict(
      14.02 k, 0.382% Params, 14.65 MMac, 5.959% MACs, 
      (block1): Bottleneck(
        4.93 k, 0.134% Params, 5.14 MMac, 2.093% MACs, 
        (conv1): Conv2d(256, 0.007% Params, 262.14 KMac, 0.107% MACs, 16, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(32, 0.001% Params, 32.77 KMac, 0.013% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(2.3 k, 0.063% Params, 2.36 MMac, 0.960% MACs, 16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(32, 0.001% Params, 32.77 KMac, 0.013% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(1.02 k, 0.028% Params, 1.05 MMac, 0.427% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(128, 0.003% Params, 131.07 KMac, 0.053% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 98.3 KMac, 0.040% MACs, )
        (downsample): Sequential(
          1.15 k, 0.031% Params, 1.18 MMac, 0.480% MACs, 
          (0): Conv2d(1.02 k, 0.028% Params, 1.05 MMac, 0.427% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (1): BatchNorm2d(128, 0.003% Params, 131.07 KMac, 0.053% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        )
      )
      (block2): Bottleneck(
        4.54 k, 0.124% Params, 4.75 MMac, 1.933% MACs, 
        (conv1): Conv2d(1.02 k, 0.028% Params, 1.05 MMac, 0.427% MACs, 64, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(32, 0.001% Params, 32.77 KMac, 0.013% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(2.3 k, 0.063% Params, 2.36 MMac, 0.960% MACs, 16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(32, 0.001% Params, 32.77 KMac, 0.013% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(1.02 k, 0.028% Params, 1.05 MMac, 0.427% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(128, 0.003% Params, 131.07 KMac, 0.053% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 98.3 KMac, 0.040% MACs, )
      )
      (block3): Bottleneck(
        4.54 k, 0.124% Params, 4.75 MMac, 1.933% MACs, 
        (conv1): Conv2d(1.02 k, 0.028% Params, 1.05 MMac, 0.427% MACs, 64, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(32, 0.001% Params, 32.77 KMac, 0.013% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(2.3 k, 0.063% Params, 2.36 MMac, 0.960% MACs, 16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(32, 0.001% Params, 32.77 KMac, 0.013% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(1.02 k, 0.028% Params, 1.05 MMac, 0.427% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(128, 0.003% Params, 131.07 KMac, 0.053% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 98.3 KMac, 0.040% MACs, )
      )
    )
  )
  (layer2): MakeLayer(
    157.18 k, 4.286% Params, 42.28 MMac, 17.199% MACs, 
    (downsample): Sequential(
      8.45 k, 0.230% Params, 2.16 MMac, 0.880% MACs, 
      (0): Conv2d(8.19 k, 0.223% Params, 2.1 MMac, 0.853% MACs, 64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)
      (1): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (blockdict): ModuleDict(
      148.74 k, 4.056% Params, 40.12 MMac, 16.320% MACs, 
      (block1): Bottleneck(
        24.19 k, 0.660% Params, 7.89 MMac, 3.209% MACs, 
        (conv1): Conv2d(2.05 k, 0.056% Params, 2.1 MMac, 0.853% MACs, 64, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.002% Params, 65.54 KMac, 0.027% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.251% Params, 2.36 MMac, 0.960% MACs, 32, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 73.73 KMac, 0.030% MACs, )
        (downsample): Sequential(
          8.45 k, 0.230% Params, 2.16 MMac, 0.880% MACs, 
          (0): Conv2d(8.19 k, 0.223% Params, 2.1 MMac, 0.853% MACs, 64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)
          (1): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        )
      )
      (block2): Bottleneck(
        17.79 k, 0.485% Params, 4.6 MMac, 1.873% MACs, 
        (conv1): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.251% Params, 2.36 MMac, 0.960% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.020% MACs, )
      )
      (block3): Bottleneck(
        17.79 k, 0.485% Params, 4.6 MMac, 1.873% MACs, 
        (conv1): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.251% Params, 2.36 MMac, 0.960% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.020% MACs, )
      )
      (block4): Bottleneck(
        17.79 k, 0.485% Params, 4.6 MMac, 1.873% MACs, 
        (conv1): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.251% Params, 2.36 MMac, 0.960% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.020% MACs, )
      )
      (block5): Bottleneck(
        17.79 k, 0.485% Params, 4.6 MMac, 1.873% MACs, 
        (conv1): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.251% Params, 2.36 MMac, 0.960% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.020% MACs, )
      )
      (block6): Bottleneck(
        17.79 k, 0.485% Params, 4.6 MMac, 1.873% MACs, 
        (conv1): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.251% Params, 2.36 MMac, 0.960% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.020% MACs, )
      )
      (block7): Bottleneck(
        17.79 k, 0.485% Params, 4.6 MMac, 1.873% MACs, 
        (conv1): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.251% Params, 2.36 MMac, 0.960% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.020% MACs, )
      )
      (block8): Bottleneck(
        17.79 k, 0.485% Params, 4.6 MMac, 1.873% MACs, 
        (conv1): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.251% Params, 2.36 MMac, 0.960% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.002% Params, 16.38 KMac, 0.007% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.112% Params, 1.05 MMac, 0.427% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.007% Params, 65.54 KMac, 0.027% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.020% MACs, )
      )
    )
  )
  (layer3): MakeLayer(
    2.59 M, 70.699% Params, 168.43 MMac, 68.519% MACs, 
    (downsample): Sequential(
      33.28 k, 0.907% Params, 2.13 MMac, 0.866% MACs, 
      (0): Conv2d(32.77 k, 0.894% Params, 2.1 MMac, 0.853% MACs, 128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)
      (1): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (blockdict): ModuleDict(
      2.56 M, 69.791% Params, 166.3 MMac, 67.653% MACs, 
      (block1): Bottleneck(
        95.49 k, 2.604% Params, 7.75 MMac, 3.151% MACs, 
        (conv1): Conv2d(8.19 k, 0.223% Params, 2.1 MMac, 0.853% MACs, 128, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 32.77 KMac, 0.013% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 36.86 KMac, 0.015% MACs, )
        (downsample): Sequential(
          33.28 k, 0.907% Params, 2.13 MMac, 0.866% MACs, 
          (0): Conv2d(32.77 k, 0.894% Params, 2.1 MMac, 0.853% MACs, 128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)
          (1): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        )
      )
      (block2): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block3): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block4): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block5): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block6): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block7): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block8): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block9): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block10): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block11): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block12): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block13): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block14): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block15): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block16): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block17): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block18): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block19): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block20): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block21): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block22): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block23): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block24): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block25): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block26): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block27): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block28): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block29): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block30): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block31): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block32): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block33): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block34): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block35): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
      (block36): Bottleneck(
        70.4 k, 1.920% Params, 4.53 MMac, 1.843% MACs, 
        (conv1): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 1.005% Params, 2.36 MMac, 0.960% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.003% Params, 8.19 KMac, 0.003% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 0.447% Params, 1.05 MMac, 0.427% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.014% Params, 32.77 KMac, 0.013% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.010% MACs, )
      )
    )
  )
  (layer4): MakeLayer(
    1.07 M, 29.220% Params, 18.77 MMac, 7.637% MACs, 
    (downsample): Sequential(
      132.1 k, 3.602% Params, 2.11 MMac, 0.860% MACs, 
      (0): Conv2d(131.07 k, 3.574% Params, 2.1 MMac, 0.853% MACs, 256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
      (1): BatchNorm2d(1.02 k, 0.028% Params, 16.38 KMac, 0.007% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (blockdict): ModuleDict(
      939.52 k, 25.618% Params, 16.66 MMac, 6.778% MACs, 
      (block1): Bottleneck(
        379.39 k, 10.345% Params, 7.67 MMac, 3.122% MACs, 
        (conv1): Conv2d(32.77 k, 0.894% Params, 2.1 MMac, 0.853% MACs, 256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(256, 0.007% Params, 16.38 KMac, 0.007% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(147.46 k, 4.021% Params, 2.36 MMac, 0.960% MACs, 128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(256, 0.007% Params, 4.1 KMac, 0.002% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(65.54 k, 1.787% Params, 1.05 MMac, 0.427% MACs, 128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(1.02 k, 0.028% Params, 16.38 KMac, 0.007% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 18.43 KMac, 0.007% MACs, )
        (downsample): Sequential(
          132.1 k, 3.602% Params, 2.11 MMac, 0.860% MACs, 
          (0): Conv2d(131.07 k, 3.574% Params, 2.1 MMac, 0.853% MACs, 256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
          (1): BatchNorm2d(1.02 k, 0.028% Params, 16.38 KMac, 0.007% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        )
      )
      (block2): Bottleneck(
        280.06 k, 7.637% Params, 4.49 MMac, 1.828% MACs, 
        (conv1): Conv2d(65.54 k, 1.787% Params, 1.05 MMac, 0.427% MACs, 512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(256, 0.007% Params, 4.1 KMac, 0.002% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(147.46 k, 4.021% Params, 2.36 MMac, 0.960% MACs, 128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(256, 0.007% Params, 4.1 KMac, 0.002% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(65.54 k, 1.787% Params, 1.05 MMac, 0.427% MACs, 128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(1.02 k, 0.028% Params, 16.38 KMac, 0.007% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 12.29 KMac, 0.005% MACs, )
      )
      (block3): Bottleneck(
        280.06 k, 7.637% Params, 4.49 MMac, 1.828% MACs, 
        (conv1): Conv2d(65.54 k, 1.787% Params, 1.05 MMac, 0.427% MACs, 512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(256, 0.007% Params, 4.1 KMac, 0.002% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(147.46 k, 4.021% Params, 2.36 MMac, 0.960% MACs, 128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(256, 0.007% Params, 4.1 KMac, 0.002% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(65.54 k, 1.787% Params, 1.05 MMac, 0.427% MACs, 128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(1.02 k, 0.028% Params, 16.38 KMac, 0.007% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 12.29 KMac, 0.005% MACs, )
      )
    )
  )
  (avgpool): AdaptiveAvgPool2d(0, 0.000% Params, 8.19 KMac, 0.003% MACs, output_size=(1, 1))
  (fc): Linear(5.13 k, 0.140% Params, 5.13 KMac, 0.002% MACs, in_features=512, out_features=10, bias=True)
)