ResNet(
  1.66 M, 111.789% Params, 91.5 MMac, 100.000% MACs, 
  (conv1): Conv2d(432, 0.029% Params, 442.37 KMac, 0.483% MACs, 3, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
  (bn1): BatchNorm2d(32, 0.002% Params, 32.77 KMac, 0.036% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (relu): ReLU(0, 0.000% Params, 16.38 KMac, 0.018% MACs, )
  (layer1): MakeLayer(
    15.17 k, 1.022% Params, 15.83 MMac, 17.298% MACs, 
    (downsample): Sequential(
      1.15 k, 0.078% Params, 1.18 MMac, 1.289% MACs, 
      (0): Conv2d(1.02 k, 0.069% Params, 1.05 MMac, 1.146% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (1): BatchNorm2d(128, 0.009% Params, 131.07 KMac, 0.143% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (blockdict): ModuleDict(
      14.02 k, 0.944% Params, 14.65 MMac, 16.009% MACs, 
      (block1): Bottleneck(
        4.93 k, 0.332% Params, 5.14 MMac, 5.623% MACs, 
        (conv1): Conv2d(256, 0.017% Params, 262.14 KMac, 0.287% MACs, 16, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(32, 0.002% Params, 32.77 KMac, 0.036% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(2.3 k, 0.155% Params, 2.36 MMac, 2.579% MACs, 16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(32, 0.002% Params, 32.77 KMac, 0.036% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(1.02 k, 0.069% Params, 1.05 MMac, 1.146% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(128, 0.009% Params, 131.07 KMac, 0.143% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 98.3 KMac, 0.107% MACs, )
        (downsample): Sequential(
          1.15 k, 0.078% Params, 1.18 MMac, 1.289% MACs, 
          (0): Conv2d(1.02 k, 0.069% Params, 1.05 MMac, 1.146% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (1): BatchNorm2d(128, 0.009% Params, 131.07 KMac, 0.143% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        )
      )
      (block2): Bottleneck(
        4.54 k, 0.306% Params, 4.75 MMac, 5.193% MACs, 
        (conv1): Conv2d(1.02 k, 0.069% Params, 1.05 MMac, 1.146% MACs, 64, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(32, 0.002% Params, 32.77 KMac, 0.036% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(2.3 k, 0.155% Params, 2.36 MMac, 2.579% MACs, 16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(32, 0.002% Params, 32.77 KMac, 0.036% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(1.02 k, 0.069% Params, 1.05 MMac, 1.146% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(128, 0.009% Params, 131.07 KMac, 0.143% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 98.3 KMac, 0.107% MACs, )
      )
      (block3): Bottleneck(
        4.54 k, 0.306% Params, 4.75 MMac, 5.193% MACs, 
        (conv1): Conv2d(1.02 k, 0.069% Params, 1.05 MMac, 1.146% MACs, 64, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(32, 0.002% Params, 32.77 KMac, 0.036% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(2.3 k, 0.155% Params, 2.36 MMac, 2.579% MACs, 16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(32, 0.002% Params, 32.77 KMac, 0.036% MACs, 16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(1.02 k, 0.069% Params, 1.05 MMac, 1.146% MACs, 16, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(128, 0.009% Params, 131.07 KMac, 0.143% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 98.3 KMac, 0.107% MACs, )
      )
    )
  )
  (layer2): MakeLayer(
    86.02 k, 5.796% Params, 23.86 MMac, 26.081% MACs, 
    (downsample): Sequential(
      8.45 k, 0.569% Params, 2.16 MMac, 2.364% MACs, 
      (0): Conv2d(8.19 k, 0.552% Params, 2.1 MMac, 2.292% MACs, 64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)
      (1): BatchNorm2d(256, 0.017% Params, 65.54 KMac, 0.072% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (blockdict): ModuleDict(
      77.57 k, 5.226% Params, 21.7 MMac, 23.718% MACs, 
      (block1): Bottleneck(
        24.19 k, 1.630% Params, 7.89 MMac, 8.622% MACs, 
        (conv1): Conv2d(2.05 k, 0.138% Params, 2.1 MMac, 2.292% MACs, 64, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.004% Params, 65.54 KMac, 0.072% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.621% Params, 2.36 MMac, 2.579% MACs, 32, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.004% Params, 16.38 KMac, 0.018% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.276% Params, 1.05 MMac, 1.146% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.017% Params, 65.54 KMac, 0.072% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 73.73 KMac, 0.081% MACs, )
        (downsample): Sequential(
          8.45 k, 0.569% Params, 2.16 MMac, 2.364% MACs, 
          (0): Conv2d(8.19 k, 0.552% Params, 2.1 MMac, 2.292% MACs, 64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)
          (1): BatchNorm2d(256, 0.017% Params, 65.54 KMac, 0.072% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        )
      )
      (block2): Bottleneck(
        17.79 k, 1.199% Params, 4.6 MMac, 5.032% MACs, 
        (conv1): Conv2d(4.1 k, 0.276% Params, 1.05 MMac, 1.146% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.004% Params, 16.38 KMac, 0.018% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.621% Params, 2.36 MMac, 2.579% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.004% Params, 16.38 KMac, 0.018% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.276% Params, 1.05 MMac, 1.146% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.017% Params, 65.54 KMac, 0.072% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.054% MACs, )
      )
      (block3): Bottleneck(
        17.79 k, 1.199% Params, 4.6 MMac, 5.032% MACs, 
        (conv1): Conv2d(4.1 k, 0.276% Params, 1.05 MMac, 1.146% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.004% Params, 16.38 KMac, 0.018% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.621% Params, 2.36 MMac, 2.579% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.004% Params, 16.38 KMac, 0.018% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.276% Params, 1.05 MMac, 1.146% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.017% Params, 65.54 KMac, 0.072% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.054% MACs, )
      )
      (block4): Bottleneck(
        17.79 k, 1.199% Params, 4.6 MMac, 5.032% MACs, 
        (conv1): Conv2d(4.1 k, 0.276% Params, 1.05 MMac, 1.146% MACs, 128, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(64, 0.004% Params, 16.38 KMac, 0.018% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(9.22 k, 0.621% Params, 2.36 MMac, 2.579% MACs, 32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(64, 0.004% Params, 16.38 KMac, 0.018% MACs, 32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(4.1 k, 0.276% Params, 1.05 MMac, 1.146% MACs, 32, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(256, 0.017% Params, 65.54 KMac, 0.072% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 49.15 KMac, 0.054% MACs, )
      )
    )
  )
  (layer3): MakeLayer(
    480.77 k, 32.393% Params, 32.53 MMac, 35.550% MACs, 
    (downsample): Sequential(
      33.28 k, 2.242% Params, 2.13 MMac, 2.328% MACs, 
      (0): Conv2d(32.77 k, 2.208% Params, 2.1 MMac, 2.292% MACs, 128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)
      (1): BatchNorm2d(512, 0.034% Params, 32.77 KMac, 0.036% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (blockdict): ModuleDict(
      447.49 k, 30.150% Params, 30.4 MMac, 33.222% MACs, 
      (block1): Bottleneck(
        95.49 k, 6.434% Params, 7.75 MMac, 8.465% MACs, 
        (conv1): Conv2d(8.19 k, 0.552% Params, 2.1 MMac, 2.292% MACs, 128, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.009% Params, 32.77 KMac, 0.036% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 2.484% Params, 2.36 MMac, 2.579% MACs, 64, 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.034% Params, 32.77 KMac, 0.036% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 36.86 KMac, 0.040% MACs, )
        (downsample): Sequential(
          33.28 k, 2.242% Params, 2.13 MMac, 2.328% MACs, 
          (0): Conv2d(32.77 k, 2.208% Params, 2.1 MMac, 2.292% MACs, 128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)
          (1): BatchNorm2d(512, 0.034% Params, 32.77 KMac, 0.036% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        )
      )
      (block2): Bottleneck(
        70.4 k, 4.743% Params, 4.53 MMac, 4.951% MACs, 
        (conv1): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 2.484% Params, 2.36 MMac, 2.579% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.034% Params, 32.77 KMac, 0.036% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.027% MACs, )
      )
      (block3): Bottleneck(
        70.4 k, 4.743% Params, 4.53 MMac, 4.951% MACs, 
        (conv1): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 2.484% Params, 2.36 MMac, 2.579% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.034% Params, 32.77 KMac, 0.036% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.027% MACs, )
      )
      (block4): Bottleneck(
        70.4 k, 4.743% Params, 4.53 MMac, 4.951% MACs, 
        (conv1): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 2.484% Params, 2.36 MMac, 2.579% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.034% Params, 32.77 KMac, 0.036% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.027% MACs, )
      )
      (block5): Bottleneck(
        70.4 k, 4.743% Params, 4.53 MMac, 4.951% MACs, 
        (conv1): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 2.484% Params, 2.36 MMac, 2.579% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.034% Params, 32.77 KMac, 0.036% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.027% MACs, )
      )
      (block6): Bottleneck(
        70.4 k, 4.743% Params, 4.53 MMac, 4.951% MACs, 
        (conv1): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(36.86 k, 2.484% Params, 2.36 MMac, 2.579% MACs, 64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(128, 0.009% Params, 8.19 KMac, 0.009% MACs, 64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(16.38 k, 1.104% Params, 1.05 MMac, 1.146% MACs, 64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(512, 0.034% Params, 32.77 KMac, 0.036% MACs, 256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 24.58 KMac, 0.027% MACs, )
      )
    )
  )
  (layer4): MakeLayer(
    1.07 M, 72.202% Params, 18.77 MMac, 20.519% MACs, 
    (downsample): Sequential(
      132.1 k, 8.900% Params, 2.11 MMac, 2.310% MACs, 
      (0): Conv2d(131.07 k, 8.831% Params, 2.1 MMac, 2.292% MACs, 256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
      (1): BatchNorm2d(1.02 k, 0.069% Params, 16.38 KMac, 0.018% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (blockdict): ModuleDict(
      939.52 k, 63.302% Params, 16.66 MMac, 18.209% MACs, 
      (block1): Bottleneck(
        379.39 k, 25.562% Params, 7.67 MMac, 8.387% MACs, 
        (conv1): Conv2d(32.77 k, 2.208% Params, 2.1 MMac, 2.292% MACs, 256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(256, 0.017% Params, 16.38 KMac, 0.018% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(147.46 k, 9.935% Params, 2.36 MMac, 2.579% MACs, 128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(256, 0.017% Params, 4.1 KMac, 0.004% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(65.54 k, 4.416% Params, 1.05 MMac, 1.146% MACs, 128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(1.02 k, 0.069% Params, 16.38 KMac, 0.018% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 18.43 KMac, 0.020% MACs, )
        (downsample): Sequential(
          132.1 k, 8.900% Params, 2.11 MMac, 2.310% MACs, 
          (0): Conv2d(131.07 k, 8.831% Params, 2.1 MMac, 2.292% MACs, 256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
          (1): BatchNorm2d(1.02 k, 0.069% Params, 16.38 KMac, 0.018% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        )
      )
      (block2): Bottleneck(
        280.06 k, 18.870% Params, 4.49 MMac, 4.911% MACs, 
        (conv1): Conv2d(65.54 k, 4.416% Params, 1.05 MMac, 1.146% MACs, 512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(256, 0.017% Params, 4.1 KMac, 0.004% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(147.46 k, 9.935% Params, 2.36 MMac, 2.579% MACs, 128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(256, 0.017% Params, 4.1 KMac, 0.004% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(65.54 k, 4.416% Params, 1.05 MMac, 1.146% MACs, 128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(1.02 k, 0.069% Params, 16.38 KMac, 0.018% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 12.29 KMac, 0.013% MACs, )
      )
      (block3): Bottleneck(
        280.06 k, 18.870% Params, 4.49 MMac, 4.911% MACs, 
        (conv1): Conv2d(65.54 k, 4.416% Params, 1.05 MMac, 1.146% MACs, 512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn1): BatchNorm2d(256, 0.017% Params, 4.1 KMac, 0.004% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv2): Conv2d(147.46 k, 9.935% Params, 2.36 MMac, 2.579% MACs, 128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn2): BatchNorm2d(256, 0.017% Params, 4.1 KMac, 0.004% MACs, 128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (conv3): Conv2d(65.54 k, 4.416% Params, 1.05 MMac, 1.146% MACs, 128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn3): BatchNorm2d(1.02 k, 0.069% Params, 16.38 KMac, 0.018% MACs, 512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(0, 0.000% Params, 12.29 KMac, 0.013% MACs, )
      )
    )
  )
  (avgpool): AdaptiveAvgPool2d(0, 0.000% Params, 8.19 KMac, 0.009% MACs, output_size=(1, 1))
  (fc): Linear(5.13 k, 0.346% Params, 5.13 KMac, 0.006% MACs, in_features=512, out_features=10, bias=True)
)