CirResNet(
  (conv1): Conv2d(3, 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
  (maxpool): Identity()
  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (relu): ReLU(inplace=True)
  (layer1): Sequential(
    (0): CirBasicBlock(
      (conv1): CirConv2d(in_features=64, out_features=64, kernel_size=3, stride=1, fix_block_size=1, search_space=[1, 2, 4, 8, 16])
      (bn1): CirBatchNorm2d(num_features=64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu1): ReLU(inplace=True)
      (conv2): CirConv2d(in_features=64, out_features=64, kernel_size=3, stride=1, fix_block_size=1, search_space=[1, 2, 4, 8, 16])
      (bn2): CirBatchNorm2d(num_features=64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu2): ReLU(inplace=True)
    )
    (1): CirBasicBlock(
      (conv1): CirConv2d(in_features=64, out_features=64, kernel_size=3, stride=1, fix_block_size=1, search_space=[1, 2, 4, 8, 16])
      (bn1): CirBatchNorm2d(num_features=64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu1): ReLU(inplace=True)
      (conv2): CirConv2d(in_features=64, out_features=64, kernel_size=3, stride=1, fix_block_size=2, search_space=[1, 2, 4, 8, 16])
      (bn2): CirBatchNorm2d(num_features=64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu2): ReLU(inplace=True)
    )
  )
  (layer2): Sequential(
    (0): CirBasicBlock(
      (conv1): CirConv2d(in_features=64, out_features=128, kernel_size=3, stride=2, fix_block_size=4, search_space=[1, 2, 4, 8, 16])
      (bn1): CirBatchNorm2d(num_features=128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu1): ReLU(inplace=True)
      (conv2): CirConv2d(in_features=128, out_features=128, kernel_size=3, stride=1, fix_block_size=1, search_space=[1, 2, 4, 8, 16])
      (bn2): CirBatchNorm2d(num_features=128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu2): ReLU(inplace=True)
      (downsample): Sequential(
        (0): CirConv2d(in_features=64, out_features=128, kernel_size=1, stride=2, fix_block_size=4, search_space=[1, 2, 4, 8, 16])
        (1): CirBatchNorm2d(num_features=128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      )
    )
    (1): CirBasicBlock(
      (conv1): CirConv2d(in_features=128, out_features=128, kernel_size=3, stride=1, fix_block_size=4, search_space=[1, 2, 4, 8, 16])
      (bn1): CirBatchNorm2d(num_features=128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu1): ReLU(inplace=True)
      (conv2): CirConv2d(in_features=128, out_features=128, kernel_size=3, stride=1, fix_block_size=4, search_space=[1, 2, 4, 8, 16])
      (bn2): CirBatchNorm2d(num_features=128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu2): ReLU(inplace=True)
    )
  )
  (layer3): Sequential(
    (0): CirBasicBlock(
      (conv1): CirConv2d(in_features=128, out_features=256, kernel_size=3, stride=2, fix_block_size=8, search_space=[1, 2, 4, 8, 16])
      (bn1): CirBatchNorm2d(num_features=256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu1): ReLU(inplace=True)
      (conv2): CirConv2d(in_features=256, out_features=256, kernel_size=3, stride=1, fix_block_size=4, search_space=[1, 2, 4, 8, 16])
      (bn2): CirBatchNorm2d(num_features=256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu2): ReLU(inplace=True)
      (downsample): Sequential(
        (0): CirConv2d(in_features=128, out_features=256, kernel_size=1, stride=2, fix_block_size=16, search_space=[1, 2, 4, 8, 16])
        (1): CirBatchNorm2d(num_features=256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      )
    )
    (1): CirBasicBlock(
      (conv1): CirConv2d(in_features=256, out_features=256, kernel_size=3, stride=1, fix_block_size=4, search_space=[1, 2, 4, 8, 16])
      (bn1): CirBatchNorm2d(num_features=256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu1): ReLU(inplace=True)
      (conv2): CirConv2d(in_features=256, out_features=256, kernel_size=3, stride=1, fix_block_size=4, search_space=[1, 2, 4, 8, 16])
      (bn2): CirBatchNorm2d(num_features=256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu2): ReLU(inplace=True)
    )
  )
  (layer4): Sequential(
    (0): CirBasicBlock(
      (conv1): CirConv2d(in_features=256, out_features=512, kernel_size=3, stride=2, fix_block_size=2, search_space=[1, 2, 4, 8, 16])
      (bn1): CirBatchNorm2d(num_features=512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu1): ReLU(inplace=True)
      (conv2): CirConv2d(in_features=512, out_features=512, kernel_size=3, stride=1, fix_block_size=2, search_space=[1, 2, 4, 8, 16])
      (bn2): CirBatchNorm2d(num_features=512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu2): ReLU(inplace=True)
      (downsample): Sequential(
        (0): CirConv2d(in_features=256, out_features=512, kernel_size=1, stride=2, fix_block_size=8, search_space=[1, 2, 4, 8, 16])
        (1): CirBatchNorm2d(num_features=512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      )
    )
    (1): CirBasicBlock(
      (conv1): CirConv2d(in_features=512, out_features=512, kernel_size=3, stride=1, fix_block_size=2, search_space=[1, 2, 4, 8, 16])
      (bn1): CirBatchNorm2d(num_features=512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu1): ReLU(inplace=True)
      (conv2): CirConv2d(in_features=512, out_features=512, kernel_size=3, stride=1, fix_block_size=8, search_space=[1, 2, 4, 8, 16])
      (bn2): CirBatchNorm2d(num_features=512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True, block_size=1)
      (relu2): ReLU(inplace=True)
    )
  )
  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))
  (fc): Linear(in_features=512, out_features=200, bias=True)
)