['features.0.1.weight:torch.Size([24, 3, 3, 3])', 'features.0.2.weight:torch.Size([24])', 'features.0.2.bias:torch.Size([24])', 'features.0.2.running_mean:torch.Size([24])', 'features.0.2.running_var:torch.Size([24])', 'features.0.2.num_batches_tracked:torch.Size([])', 'features.1.conv.0.1.weight:torch.Size([24, 1, 3, 3])', 'features.1.conv.0.2.weight:torch.Size([24])', 'features.1.conv.0.2.bias:torch.Size([24])', 'features.1.conv.0.2.running_mean:torch.Size([24])', 'features.1.conv.0.2.running_var:torch.Size([24])', 'features.1.conv.0.2.num_batches_tracked:torch.Size([])', 'features.1.conv.1.se.1.weight:torch.Size([6, 24, 1, 1])', 'features.1.conv.1.se.1.bias:torch.Size([6])', 'features.1.conv.1.se.3.weight:torch.Size([24, 6, 1, 1])', 'features.1.conv.1.se.3.bias:torch.Size([24])', 'features.1.conv.2.weight:torch.Size([16, 24, 1, 1])', 'features.1.conv.3.weight:torch.Size([16])', 'features.1.conv.3.bias:torch.Size([16])', 'features.1.conv.3.running_mean:torch.Size([16])', 'features.1.conv.3.running_var:torch.Size([16])', 'features.1.conv.3.num_batches_tracked:torch.Size([])', 'features.2.conv.0.1.weight:torch.Size([96, 16, 1, 1])', 'features.2.conv.0.2.weight:torch.Size([96])', 'features.2.conv.0.2.bias:torch.Size([96])', 'features.2.conv.0.2.running_mean:torch.Size([96])', 'features.2.conv.0.2.running_var:torch.Size([96])', 'features.2.conv.0.2.num_batches_tracked:torch.Size([])', 'features.2.conv.1.1.weight:torch.Size([96, 1, 3, 3])', 'features.2.conv.1.2.weight:torch.Size([96])', 'features.2.conv.1.2.bias:torch.Size([96])', 'features.2.conv.1.2.running_mean:torch.Size([96])', 'features.2.conv.1.2.running_var:torch.Size([96])', 'features.2.conv.1.2.num_batches_tracked:torch.Size([])', 'features.2.conv.2.se.1.weight:torch.Size([4, 96, 1, 1])', 'features.2.conv.2.se.1.bias:torch.Size([4])', 'features.2.conv.2.se.3.weight:torch.Size([96, 4, 1, 1])', 'features.2.conv.2.se.3.bias:torch.Size([96])', 'features.2.conv.3.weight:torch.Size([24, 96, 1, 1])', 'features.2.conv.4.weight:torch.Size([24])', 'features.2.conv.4.bias:torch.Size([24])', 'features.2.conv.4.running_mean:torch.Size([24])', 'features.2.conv.4.running_var:torch.Size([24])', 'features.2.conv.4.num_batches_tracked:torch.Size([])', 'features.3.conv.0.1.weight:torch.Size([144, 24, 1, 1])', 'features.3.conv.0.2.weight:torch.Size([144])', 'features.3.conv.0.2.bias:torch.Size([144])', 'features.3.conv.0.2.running_mean:torch.Size([144])', 'features.3.conv.0.2.running_var:torch.Size([144])', 'features.3.conv.0.2.num_batches_tracked:torch.Size([])', 'features.3.conv.1.1.weight:torch.Size([144, 1, 3, 3])', 'features.3.conv.1.2.weight:torch.Size([144])', 'features.3.conv.1.2.bias:torch.Size([144])', 'features.3.conv.1.2.running_mean:torch.Size([144])', 'features.3.conv.1.2.running_var:torch.Size([144])', 'features.3.conv.1.2.num_batches_tracked:torch.Size([])', 'features.3.conv.2.se.1.weight:torch.Size([6, 144, 1, 1])', 'features.3.conv.2.se.1.bias:torch.Size([6])', 'features.3.conv.2.se.3.weight:torch.Size([144, 6, 1, 1])', 'features.3.conv.2.se.3.bias:torch.Size([144])', 'features.3.conv.3.weight:torch.Size([24, 144, 1, 1])', 'features.3.conv.4.weight:torch.Size([24])', 'features.3.conv.4.bias:torch.Size([24])', 'features.3.conv.4.running_mean:torch.Size([24])', 'features.3.conv.4.running_var:torch.Size([24])', 'features.3.conv.4.num_batches_tracked:torch.Size([])', 'features.4.conv.0.1.weight:torch.Size([144, 24, 1, 1])', 'features.4.conv.0.2.weight:torch.Size([144])', 'features.4.conv.0.2.bias:torch.Size([144])', 'features.4.conv.0.2.running_mean:torch.Size([144])', 'features.4.conv.0.2.running_var:torch.Size([144])', 'features.4.conv.0.2.num_batches_tracked:torch.Size([])', 'features.4.conv.1.1.weight:torch.Size([144, 1, 5, 5])', 'features.4.conv.1.2.weight:torch.Size([144])', 'features.4.conv.1.2.bias:torch.Size([144])', 'features.4.conv.1.2.running_mean:torch.Size([144])', 'features.4.conv.1.2.running_var:torch.Size([144])', 'features.4.conv.1.2.num_batches_tracked:torch.Size([])', 'features.4.conv.2.se.1.weight:torch.Size([6, 144, 1, 1])', 'features.4.conv.2.se.1.bias:torch.Size([6])', 'features.4.conv.2.se.3.weight:torch.Size([144, 6, 1, 1])', 'features.4.conv.2.se.3.bias:torch.Size([144])', 'features.4.conv.3.weight:torch.Size([32, 144, 1, 1])', 'features.4.conv.4.weight:torch.Size([32])', 'features.4.conv.4.bias:torch.Size([32])', 'features.4.conv.4.running_mean:torch.Size([32])', 'features.4.conv.4.running_var:torch.Size([32])', 'features.4.conv.4.num_batches_tracked:torch.Size([])', 'features.5.conv.0.1.weight:torch.Size([192, 32, 1, 1])', 'features.5.conv.0.2.weight:torch.Size([192])', 'features.5.conv.0.2.bias:torch.Size([192])', 'features.5.conv.0.2.running_mean:torch.Size([192])', 'features.5.conv.0.2.running_var:torch.Size([192])', 'features.5.conv.0.2.num_batches_tracked:torch.Size([])', 'features.5.conv.1.1.weight:torch.Size([192, 1, 5, 5])', 'features.5.conv.1.2.weight:torch.Size([192])', 'features.5.conv.1.2.bias:torch.Size([192])', 'features.5.conv.1.2.running_mean:torch.Size([192])', 'features.5.conv.1.2.running_var:torch.Size([192])', 'features.5.conv.1.2.num_batches_tracked:torch.Size([])', 'features.5.conv.2.se.1.weight:torch.Size([8, 192, 1, 1])', 'features.5.conv.2.se.1.bias:torch.Size([8])', 'features.5.conv.2.se.3.weight:torch.Size([192, 8, 1, 1])', 'features.5.conv.2.se.3.bias:torch.Size([192])', 'features.5.conv.3.weight:torch.Size([32, 192, 1, 1])', 'features.5.conv.4.weight:torch.Size([32])', 'features.5.conv.4.bias:torch.Size([32])', 'features.5.conv.4.running_mean:torch.Size([32])', 'features.5.conv.4.running_var:torch.Size([32])', 'features.5.conv.4.num_batches_tracked:torch.Size([])', 'features.6.conv.0.1.weight:torch.Size([192, 32, 1, 1])', 'features.6.conv.0.2.weight:torch.Size([192])', 'features.6.conv.0.2.bias:torch.Size([192])', 'features.6.conv.0.2.running_mean:torch.Size([192])', 'features.6.conv.0.2.running_var:torch.Size([192])', 'features.6.conv.0.2.num_batches_tracked:torch.Size([])', 'features.6.conv.1.1.weight:torch.Size([192, 1, 3, 3])', 'features.6.conv.1.2.weight:torch.Size([192])', 'features.6.conv.1.2.bias:torch.Size([192])', 'features.6.conv.1.2.running_mean:torch.Size([192])', 'features.6.conv.1.2.running_var:torch.Size([192])', 'features.6.conv.1.2.num_batches_tracked:torch.Size([])', 'features.6.conv.2.se.1.weight:torch.Size([8, 192, 1, 1])', 'features.6.conv.2.se.1.bias:torch.Size([8])', 'features.6.conv.2.se.3.weight:torch.Size([192, 8, 1, 1])', 'features.6.conv.2.se.3.bias:torch.Size([192])', 'features.6.conv.3.weight:torch.Size([64, 192, 1, 1])', 'features.6.conv.4.weight:torch.Size([64])', 'features.6.conv.4.bias:torch.Size([64])', 'features.6.conv.4.running_mean:torch.Size([64])', 'features.6.conv.4.running_var:torch.Size([64])', 'features.6.conv.4.num_batches_tracked:torch.Size([])', 'features.7.conv.0.1.weight:torch.Size([384, 64, 1, 1])', 'features.7.conv.0.2.weight:torch.Size([384])', 'features.7.conv.0.2.bias:torch.Size([384])', 'features.7.conv.0.2.running_mean:torch.Size([384])', 'features.7.conv.0.2.running_var:torch.Size([384])', 'features.7.conv.0.2.num_batches_tracked:torch.Size([])', 'features.7.conv.1.1.weight:torch.Size([384, 1, 3, 3])', 'features.7.conv.1.2.weight:torch.Size([384])', 'features.7.conv.1.2.bias:torch.Size([384])', 'features.7.conv.1.2.running_mean:torch.Size([384])', 'features.7.conv.1.2.running_var:torch.Size([384])', 'features.7.conv.1.2.num_batches_tracked:torch.Size([])', 'features.7.conv.2.se.1.weight:torch.Size([16, 384, 1, 1])', 'features.7.conv.2.se.1.bias:torch.Size([16])', 'features.7.conv.2.se.3.weight:torch.Size([384, 16, 1, 1])', 'features.7.conv.2.se.3.bias:torch.Size([384])', 'features.7.conv.3.weight:torch.Size([64, 384, 1, 1])', 'features.7.conv.4.weight:torch.Size([64])', 'features.7.conv.4.bias:torch.Size([64])', 'features.7.conv.4.running_mean:torch.Size([64])', 'features.7.conv.4.running_var:torch.Size([64])', 'features.7.conv.4.num_batches_tracked:torch.Size([])', 'features.8.conv.0.1.weight:torch.Size([384, 64, 1, 1])', 'features.8.conv.0.2.weight:torch.Size([384])', 'features.8.conv.0.2.bias:torch.Size([384])', 'features.8.conv.0.2.running_mean:torch.Size([384])', 'features.8.conv.0.2.running_var:torch.Size([384])', 'features.8.conv.0.2.num_batches_tracked:torch.Size([])', 'features.8.conv.1.1.weight:torch.Size([384, 1, 3, 3])', 'features.8.conv.1.2.weight:torch.Size([384])', 'features.8.conv.1.2.bias:torch.Size([384])', 'features.8.conv.1.2.running_mean:torch.Size([384])', 'features.8.conv.1.2.running_var:torch.Size([384])', 'features.8.conv.1.2.num_batches_tracked:torch.Size([])', 'features.8.conv.2.se.1.weight:torch.Size([16, 384, 1, 1])', 'features.8.conv.2.se.1.bias:torch.Size([16])', 'features.8.conv.2.se.3.weight:torch.Size([384, 16, 1, 1])', 'features.8.conv.2.se.3.bias:torch.Size([384])', 'features.8.conv.3.weight:torch.Size([64, 384, 1, 1])', 'features.8.conv.4.weight:torch.Size([64])', 'features.8.conv.4.bias:torch.Size([64])', 'features.8.conv.4.running_mean:torch.Size([64])', 'features.8.conv.4.running_var:torch.Size([64])', 'features.8.conv.4.num_batches_tracked:torch.Size([])', 'features.9.conv.0.1.weight:torch.Size([384, 64, 1, 1])', 'features.9.conv.0.2.weight:torch.Size([384])', 'features.9.conv.0.2.bias:torch.Size([384])', 'features.9.conv.0.2.running_mean:torch.Size([384])', 'features.9.conv.0.2.running_var:torch.Size([384])', 'features.9.conv.0.2.num_batches_tracked:torch.Size([])', 'features.9.conv.1.1.weight:torch.Size([384, 1, 5, 5])', 'features.9.conv.1.2.weight:torch.Size([384])', 'features.9.conv.1.2.bias:torch.Size([384])', 'features.9.conv.1.2.running_mean:torch.Size([384])', 'features.9.conv.1.2.running_var:torch.Size([384])', 'features.9.conv.1.2.num_batches_tracked:torch.Size([])', 'features.9.conv.2.se.1.weight:torch.Size([16, 384, 1, 1])', 'features.9.conv.2.se.1.bias:torch.Size([16])', 'features.9.conv.2.se.3.weight:torch.Size([384, 16, 1, 1])', 'features.9.conv.2.se.3.bias:torch.Size([384])', 'features.9.conv.3.weight:torch.Size([96, 384, 1, 1])', 'features.9.conv.4.weight:torch.Size([96])', 'features.9.conv.4.bias:torch.Size([96])', 'features.9.conv.4.running_mean:torch.Size([96])', 'features.9.conv.4.running_var:torch.Size([96])', 'features.9.conv.4.num_batches_tracked:torch.Size([])', 'features.10.conv.0.1.weight:torch.Size([576, 96, 1, 1])', 'features.10.conv.0.2.weight:torch.Size([576])', 'features.10.conv.0.2.bias:torch.Size([576])', 'features.10.conv.0.2.running_mean:torch.Size([576])', 'features.10.conv.0.2.running_var:torch.Size([576])', 'features.10.conv.0.2.num_batches_tracked:torch.Size([])', 'features.10.conv.1.1.weight:torch.Size([576, 1, 5, 5])', 'features.10.conv.1.2.weight:torch.Size([576])', 'features.10.conv.1.2.bias:torch.Size([576])', 'features.10.conv.1.2.running_mean:torch.Size([576])', 'features.10.conv.1.2.running_var:torch.Size([576])', 'features.10.conv.1.2.num_batches_tracked:torch.Size([])', 'features.10.conv.2.se.1.weight:torch.Size([24, 576, 1, 1])', 'features.10.conv.2.se.1.bias:torch.Size([24])', 'features.10.conv.2.se.3.weight:torch.Size([576, 24, 1, 1])', 'features.10.conv.2.se.3.bias:torch.Size([576])', 'features.10.conv.3.weight:torch.Size([96, 576, 1, 1])', 'features.10.conv.4.weight:torch.Size([96])', 'features.10.conv.4.bias:torch.Size([96])', 'features.10.conv.4.running_mean:torch.Size([96])', 'features.10.conv.4.running_var:torch.Size([96])', 'features.10.conv.4.num_batches_tracked:torch.Size([])', 'features.11.conv.0.1.weight:torch.Size([576, 96, 1, 1])', 'features.11.conv.0.2.weight:torch.Size([576])', 'features.11.conv.0.2.bias:torch.Size([576])', 'features.11.conv.0.2.running_mean:torch.Size([576])', 'features.11.conv.0.2.running_var:torch.Size([576])', 'features.11.conv.0.2.num_batches_tracked:torch.Size([])', 'features.11.conv.1.1.weight:torch.Size([576, 1, 5, 5])', 'features.11.conv.1.2.weight:torch.Size([576])', 'features.11.conv.1.2.bias:torch.Size([576])', 'features.11.conv.1.2.running_mean:torch.Size([576])', 'features.11.conv.1.2.running_var:torch.Size([576])', 'features.11.conv.1.2.num_batches_tracked:torch.Size([])', 'features.11.conv.2.se.1.weight:torch.Size([24, 576, 1, 1])', 'features.11.conv.2.se.1.bias:torch.Size([24])', 'features.11.conv.2.se.3.weight:torch.Size([576, 24, 1, 1])', 'features.11.conv.2.se.3.bias:torch.Size([576])', 'features.11.conv.3.weight:torch.Size([96, 576, 1, 1])', 'features.11.conv.4.weight:torch.Size([96])', 'features.11.conv.4.bias:torch.Size([96])', 'features.11.conv.4.running_mean:torch.Size([96])', 'features.11.conv.4.running_var:torch.Size([96])', 'features.11.conv.4.num_batches_tracked:torch.Size([])', 'features.12.conv.0.1.weight:torch.Size([576, 96, 1, 1])', 'features.12.conv.0.2.weight:torch.Size([576])', 'features.12.conv.0.2.bias:torch.Size([576])', 'features.12.conv.0.2.running_mean:torch.Size([576])', 'features.12.conv.0.2.running_var:torch.Size([576])', 'features.12.conv.0.2.num_batches_tracked:torch.Size([])', 'features.12.conv.1.1.weight:torch.Size([576, 1, 5, 5])', 'features.12.conv.1.2.weight:torch.Size([576])', 'features.12.conv.1.2.bias:torch.Size([576])', 'features.12.conv.1.2.running_mean:torch.Size([576])', 'features.12.conv.1.2.running_var:torch.Size([576])', 'features.12.conv.1.2.num_batches_tracked:torch.Size([])', 'features.12.conv.2.se.1.weight:torch.Size([24, 576, 1, 1])', 'features.12.conv.2.se.1.bias:torch.Size([24])', 'features.12.conv.2.se.3.weight:torch.Size([576, 24, 1, 1])', 'features.12.conv.2.se.3.bias:torch.Size([576])', 'features.12.conv.3.weight:torch.Size([160, 576, 1, 1])', 'features.12.conv.4.weight:torch.Size([160])', 'features.12.conv.4.bias:torch.Size([160])', 'features.12.conv.4.running_mean:torch.Size([160])', 'features.12.conv.4.running_var:torch.Size([160])', 'features.12.conv.4.num_batches_tracked:torch.Size([])', 'features.13.conv.0.1.weight:torch.Size([960, 160, 1, 1])', 'features.13.conv.0.2.weight:torch.Size([960])', 'features.13.conv.0.2.bias:torch.Size([960])', 'features.13.conv.0.2.running_mean:torch.Size([960])', 'features.13.conv.0.2.running_var:torch.Size([960])', 'features.13.conv.0.2.num_batches_tracked:torch.Size([])', 'features.13.conv.1.1.weight:torch.Size([960, 1, 5, 5])', 'features.13.conv.1.2.weight:torch.Size([960])', 'features.13.conv.1.2.bias:torch.Size([960])', 'features.13.conv.1.2.running_mean:torch.Size([960])', 'features.13.conv.1.2.running_var:torch.Size([960])', 'features.13.conv.1.2.num_batches_tracked:torch.Size([])', 'features.13.conv.2.se.1.weight:torch.Size([40, 960, 1, 1])', 'features.13.conv.2.se.1.bias:torch.Size([40])', 'features.13.conv.2.se.3.weight:torch.Size([960, 40, 1, 1])', 'features.13.conv.2.se.3.bias:torch.Size([960])', 'features.13.conv.3.weight:torch.Size([160, 960, 1, 1])', 'features.13.conv.4.weight:torch.Size([160])', 'features.13.conv.4.bias:torch.Size([160])', 'features.13.conv.4.running_mean:torch.Size([160])', 'features.13.conv.4.running_var:torch.Size([160])', 'features.13.conv.4.num_batches_tracked:torch.Size([])', 'features.14.conv.0.1.weight:torch.Size([960, 160, 1, 1])', 'features.14.conv.0.2.weight:torch.Size([960])', 'features.14.conv.0.2.bias:torch.Size([960])', 'features.14.conv.0.2.running_mean:torch.Size([960])', 'features.14.conv.0.2.running_var:torch.Size([960])', 'features.14.conv.0.2.num_batches_tracked:torch.Size([])', 'features.14.conv.1.1.weight:torch.Size([960, 1, 5, 5])', 'features.14.conv.1.2.weight:torch.Size([960])', 'features.14.conv.1.2.bias:torch.Size([960])', 'features.14.conv.1.2.running_mean:torch.Size([960])', 'features.14.conv.1.2.running_var:torch.Size([960])', 'features.14.conv.1.2.num_batches_tracked:torch.Size([])', 'features.14.conv.2.se.1.weight:torch.Size([40, 960, 1, 1])', 'features.14.conv.2.se.1.bias:torch.Size([40])', 'features.14.conv.2.se.3.weight:torch.Size([960, 40, 1, 1])', 'features.14.conv.2.se.3.bias:torch.Size([960])', 'features.14.conv.3.weight:torch.Size([160, 960, 1, 1])', 'features.14.conv.4.weight:torch.Size([160])', 'features.14.conv.4.bias:torch.Size([160])', 'features.14.conv.4.running_mean:torch.Size([160])', 'features.14.conv.4.running_var:torch.Size([160])', 'features.14.conv.4.num_batches_tracked:torch.Size([])', 'features.15.conv.0.1.weight:torch.Size([960, 160, 1, 1])', 'features.15.conv.0.2.weight:torch.Size([960])', 'features.15.conv.0.2.bias:torch.Size([960])', 'features.15.conv.0.2.running_mean:torch.Size([960])', 'features.15.conv.0.2.running_var:torch.Size([960])', 'features.15.conv.0.2.num_batches_tracked:torch.Size([])', 'features.15.conv.1.1.weight:torch.Size([960, 1, 3, 3])', 'features.15.conv.1.2.weight:torch.Size([960])', 'features.15.conv.1.2.bias:torch.Size([960])', 'features.15.conv.1.2.running_mean:torch.Size([960])', 'features.15.conv.1.2.running_var:torch.Size([960])', 'features.15.conv.1.2.num_batches_tracked:torch.Size([])', 'features.15.conv.2.se.1.weight:torch.Size([40, 960, 1, 1])', 'features.15.conv.2.se.1.bias:torch.Size([40])', 'features.15.conv.2.se.3.weight:torch.Size([960, 40, 1, 1])', 'features.15.conv.2.se.3.bias:torch.Size([960])', 'features.15.conv.3.weight:torch.Size([264, 960, 1, 1])', 'features.15.conv.4.weight:torch.Size([264])', 'features.15.conv.4.bias:torch.Size([264])', 'features.15.conv.4.running_mean:torch.Size([264])', 'features.15.conv.4.running_var:torch.Size([264])', 'features.15.conv.4.num_batches_tracked:torch.Size([])', 'trans1_branch1.0.weight:torch.Size([21, 24, 3, 3])', 'trans1_branch1.0.bias:torch.Size([21])', 'trans1_branch1.1.weight:torch.Size([21])', 'trans1_branch1.1.bias:torch.Size([21])', 'trans1_branch1.1.running_mean:torch.Size([21])', 'trans1_branch1.1.running_var:torch.Size([21])', 'trans1_branch1.1.num_batches_tracked:torch.Size([])', 'trans1_branch2.0.weight:torch.Size([42, 32, 3, 3])', 'trans1_branch2.0.bias:torch.Size([42])', 'trans1_branch2.1.weight:torch.Size([42])', 'trans1_branch2.1.bias:torch.Size([42])', 'trans1_branch2.1.running_mean:torch.Size([42])', 'trans1_branch2.1.running_var:torch.Size([42])', 'trans1_branch2.1.num_batches_tracked:torch.Size([])', 'stage2.0.branches.0.0.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage2.0.branches.0.0.bn1.weight:torch.Size([21])', 'stage2.0.branches.0.0.bn1.bias:torch.Size([21])', 'stage2.0.branches.0.0.bn1.running_mean:torch.Size([21])', 'stage2.0.branches.0.0.bn1.running_var:torch.Size([21])', 'stage2.0.branches.0.0.bn1.num_batches_tracked:torch.Size([])', 'stage2.0.branches.0.0.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage2.0.branches.0.0.bn2.weight:torch.Size([21])', 'stage2.0.branches.0.0.bn2.bias:torch.Size([21])', 'stage2.0.branches.0.0.bn2.running_mean:torch.Size([21])', 'stage2.0.branches.0.0.bn2.running_var:torch.Size([21])', 'stage2.0.branches.0.0.bn2.num_batches_tracked:torch.Size([])', 'stage2.0.branches.0.1.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage2.0.branches.0.1.bn1.weight:torch.Size([21])', 'stage2.0.branches.0.1.bn1.bias:torch.Size([21])', 'stage2.0.branches.0.1.bn1.running_mean:torch.Size([21])', 'stage2.0.branches.0.1.bn1.running_var:torch.Size([21])', 'stage2.0.branches.0.1.bn1.num_batches_tracked:torch.Size([])', 'stage2.0.branches.0.1.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage2.0.branches.0.1.bn2.weight:torch.Size([21])', 'stage2.0.branches.0.1.bn2.bias:torch.Size([21])', 'stage2.0.branches.0.1.bn2.running_mean:torch.Size([21])', 'stage2.0.branches.0.1.bn2.running_var:torch.Size([21])', 'stage2.0.branches.0.1.bn2.num_batches_tracked:torch.Size([])', 'stage2.0.branches.1.0.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage2.0.branches.1.0.bn1.weight:torch.Size([42])', 'stage2.0.branches.1.0.bn1.bias:torch.Size([42])', 'stage2.0.branches.1.0.bn1.running_mean:torch.Size([42])', 'stage2.0.branches.1.0.bn1.running_var:torch.Size([42])', 'stage2.0.branches.1.0.bn1.num_batches_tracked:torch.Size([])', 'stage2.0.branches.1.0.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage2.0.branches.1.0.bn2.weight:torch.Size([42])', 'stage2.0.branches.1.0.bn2.bias:torch.Size([42])', 'stage2.0.branches.1.0.bn2.running_mean:torch.Size([42])', 'stage2.0.branches.1.0.bn2.running_var:torch.Size([42])', 'stage2.0.branches.1.0.bn2.num_batches_tracked:torch.Size([])', 'stage2.0.branches.1.1.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage2.0.branches.1.1.bn1.weight:torch.Size([42])', 'stage2.0.branches.1.1.bn1.bias:torch.Size([42])', 'stage2.0.branches.1.1.bn1.running_mean:torch.Size([42])', 'stage2.0.branches.1.1.bn1.running_var:torch.Size([42])', 'stage2.0.branches.1.1.bn1.num_batches_tracked:torch.Size([])', 'stage2.0.branches.1.1.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage2.0.branches.1.1.bn2.weight:torch.Size([42])', 'stage2.0.branches.1.1.bn2.bias:torch.Size([42])', 'stage2.0.branches.1.1.bn2.running_mean:torch.Size([42])', 'stage2.0.branches.1.1.bn2.running_var:torch.Size([42])', 'stage2.0.branches.1.1.bn2.num_batches_tracked:torch.Size([])', 'stage2.0.fuse_layers.0.1.0.weight:torch.Size([21, 42, 1, 1])', 'stage2.0.fuse_layers.0.1.1.weight:torch.Size([21])', 'stage2.0.fuse_layers.0.1.1.bias:torch.Size([21])', 'stage2.0.fuse_layers.0.1.1.running_mean:torch.Size([21])', 'stage2.0.fuse_layers.0.1.1.running_var:torch.Size([21])', 'stage2.0.fuse_layers.0.1.1.num_batches_tracked:torch.Size([])', 'stage2.0.fuse_layers.1.0.0.0.weight:torch.Size([42, 21, 3, 3])', 'stage2.0.fuse_layers.1.0.0.1.weight:torch.Size([42])', 'stage2.0.fuse_layers.1.0.0.1.bias:torch.Size([42])', 'stage2.0.fuse_layers.1.0.0.1.running_mean:torch.Size([42])', 'stage2.0.fuse_layers.1.0.0.1.running_var:torch.Size([42])', 'stage2.0.fuse_layers.1.0.0.1.num_batches_tracked:torch.Size([])', 'trans2_branch1.0.weight:torch.Size([21, 21, 3, 3])', 'trans2_branch1.0.bias:torch.Size([21])', 'trans2_branch1.1.weight:torch.Size([21])', 'trans2_branch1.1.bias:torch.Size([21])', 'trans2_branch1.1.running_mean:torch.Size([21])', 'trans2_branch1.1.running_var:torch.Size([21])', 'trans2_branch1.1.num_batches_tracked:torch.Size([])', 'trans2_branch2.0.weight:torch.Size([42, 42, 3, 3])', 'trans2_branch2.0.bias:torch.Size([42])', 'trans2_branch2.1.weight:torch.Size([42])', 'trans2_branch2.1.bias:torch.Size([42])', 'trans2_branch2.1.running_mean:torch.Size([42])', 'trans2_branch2.1.running_var:torch.Size([42])', 'trans2_branch2.1.num_batches_tracked:torch.Size([])', 'trans2_branch3.0.weight:torch.Size([83, 96, 3, 3])', 'trans2_branch3.0.bias:torch.Size([83])', 'trans2_branch3.1.weight:torch.Size([83])', 'trans2_branch3.1.bias:torch.Size([83])', 'trans2_branch3.1.running_mean:torch.Size([83])', 'trans2_branch3.1.running_var:torch.Size([83])', 'trans2_branch3.1.num_batches_tracked:torch.Size([])', 'stage3.0.branches.0.0.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage3.0.branches.0.0.bn1.weight:torch.Size([21])', 'stage3.0.branches.0.0.bn1.bias:torch.Size([21])', 'stage3.0.branches.0.0.bn1.running_mean:torch.Size([21])', 'stage3.0.branches.0.0.bn1.running_var:torch.Size([21])', 'stage3.0.branches.0.0.bn1.num_batches_tracked:torch.Size([])', 'stage3.0.branches.0.0.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage3.0.branches.0.0.bn2.weight:torch.Size([21])', 'stage3.0.branches.0.0.bn2.bias:torch.Size([21])', 'stage3.0.branches.0.0.bn2.running_mean:torch.Size([21])', 'stage3.0.branches.0.0.bn2.running_var:torch.Size([21])', 'stage3.0.branches.0.0.bn2.num_batches_tracked:torch.Size([])', 'stage3.0.branches.0.1.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage3.0.branches.0.1.bn1.weight:torch.Size([21])', 'stage3.0.branches.0.1.bn1.bias:torch.Size([21])', 'stage3.0.branches.0.1.bn1.running_mean:torch.Size([21])', 'stage3.0.branches.0.1.bn1.running_var:torch.Size([21])', 'stage3.0.branches.0.1.bn1.num_batches_tracked:torch.Size([])', 'stage3.0.branches.0.1.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage3.0.branches.0.1.bn2.weight:torch.Size([21])', 'stage3.0.branches.0.1.bn2.bias:torch.Size([21])', 'stage3.0.branches.0.1.bn2.running_mean:torch.Size([21])', 'stage3.0.branches.0.1.bn2.running_var:torch.Size([21])', 'stage3.0.branches.0.1.bn2.num_batches_tracked:torch.Size([])', 'stage3.0.branches.1.0.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage3.0.branches.1.0.bn1.weight:torch.Size([42])', 'stage3.0.branches.1.0.bn1.bias:torch.Size([42])', 'stage3.0.branches.1.0.bn1.running_mean:torch.Size([42])', 'stage3.0.branches.1.0.bn1.running_var:torch.Size([42])', 'stage3.0.branches.1.0.bn1.num_batches_tracked:torch.Size([])', 'stage3.0.branches.1.0.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage3.0.branches.1.0.bn2.weight:torch.Size([42])', 'stage3.0.branches.1.0.bn2.bias:torch.Size([42])', 'stage3.0.branches.1.0.bn2.running_mean:torch.Size([42])', 'stage3.0.branches.1.0.bn2.running_var:torch.Size([42])', 'stage3.0.branches.1.0.bn2.num_batches_tracked:torch.Size([])', 'stage3.0.branches.1.1.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage3.0.branches.1.1.bn1.weight:torch.Size([42])', 'stage3.0.branches.1.1.bn1.bias:torch.Size([42])', 'stage3.0.branches.1.1.bn1.running_mean:torch.Size([42])', 'stage3.0.branches.1.1.bn1.running_var:torch.Size([42])', 'stage3.0.branches.1.1.bn1.num_batches_tracked:torch.Size([])', 'stage3.0.branches.1.1.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage3.0.branches.1.1.bn2.weight:torch.Size([42])', 'stage3.0.branches.1.1.bn2.bias:torch.Size([42])', 'stage3.0.branches.1.1.bn2.running_mean:torch.Size([42])', 'stage3.0.branches.1.1.bn2.running_var:torch.Size([42])', 'stage3.0.branches.1.1.bn2.num_batches_tracked:torch.Size([])', 'stage3.0.branches.2.0.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage3.0.branches.2.0.bn1.weight:torch.Size([83])', 'stage3.0.branches.2.0.bn1.bias:torch.Size([83])', 'stage3.0.branches.2.0.bn1.running_mean:torch.Size([83])', 'stage3.0.branches.2.0.bn1.running_var:torch.Size([83])', 'stage3.0.branches.2.0.bn1.num_batches_tracked:torch.Size([])', 'stage3.0.branches.2.0.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage3.0.branches.2.0.bn2.weight:torch.Size([83])', 'stage3.0.branches.2.0.bn2.bias:torch.Size([83])', 'stage3.0.branches.2.0.bn2.running_mean:torch.Size([83])', 'stage3.0.branches.2.0.bn2.running_var:torch.Size([83])', 'stage3.0.branches.2.0.bn2.num_batches_tracked:torch.Size([])', 'stage3.0.branches.2.1.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage3.0.branches.2.1.bn1.weight:torch.Size([83])', 'stage3.0.branches.2.1.bn1.bias:torch.Size([83])', 'stage3.0.branches.2.1.bn1.running_mean:torch.Size([83])', 'stage3.0.branches.2.1.bn1.running_var:torch.Size([83])', 'stage3.0.branches.2.1.bn1.num_batches_tracked:torch.Size([])', 'stage3.0.branches.2.1.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage3.0.branches.2.1.bn2.weight:torch.Size([83])', 'stage3.0.branches.2.1.bn2.bias:torch.Size([83])', 'stage3.0.branches.2.1.bn2.running_mean:torch.Size([83])', 'stage3.0.branches.2.1.bn2.running_var:torch.Size([83])', 'stage3.0.branches.2.1.bn2.num_batches_tracked:torch.Size([])', 'stage3.0.fuse_layers.0.1.0.weight:torch.Size([21, 42, 1, 1])', 'stage3.0.fuse_layers.0.1.1.weight:torch.Size([21])', 'stage3.0.fuse_layers.0.1.1.bias:torch.Size([21])', 'stage3.0.fuse_layers.0.1.1.running_mean:torch.Size([21])', 'stage3.0.fuse_layers.0.1.1.running_var:torch.Size([21])', 'stage3.0.fuse_layers.0.1.1.num_batches_tracked:torch.Size([])', 'stage3.0.fuse_layers.0.2.0.weight:torch.Size([21, 83, 1, 1])', 'stage3.0.fuse_layers.0.2.1.weight:torch.Size([21])', 'stage3.0.fuse_layers.0.2.1.bias:torch.Size([21])', 'stage3.0.fuse_layers.0.2.1.running_mean:torch.Size([21])', 'stage3.0.fuse_layers.0.2.1.running_var:torch.Size([21])', 'stage3.0.fuse_layers.0.2.1.num_batches_tracked:torch.Size([])', 'stage3.0.fuse_layers.1.0.0.0.weight:torch.Size([42, 21, 3, 3])', 'stage3.0.fuse_layers.1.0.0.1.weight:torch.Size([42])', 'stage3.0.fuse_layers.1.0.0.1.bias:torch.Size([42])', 'stage3.0.fuse_layers.1.0.0.1.running_mean:torch.Size([42])', 'stage3.0.fuse_layers.1.0.0.1.running_var:torch.Size([42])', 'stage3.0.fuse_layers.1.0.0.1.num_batches_tracked:torch.Size([])', 'stage3.0.fuse_layers.1.2.0.weight:torch.Size([42, 83, 1, 1])', 'stage3.0.fuse_layers.1.2.1.weight:torch.Size([42])', 'stage3.0.fuse_layers.1.2.1.bias:torch.Size([42])', 'stage3.0.fuse_layers.1.2.1.running_mean:torch.Size([42])', 'stage3.0.fuse_layers.1.2.1.running_var:torch.Size([42])', 'stage3.0.fuse_layers.1.2.1.num_batches_tracked:torch.Size([])', 'stage3.0.fuse_layers.2.0.0.0.weight:torch.Size([21, 21, 3, 3])', 'stage3.0.fuse_layers.2.0.0.1.weight:torch.Size([21])', 'stage3.0.fuse_layers.2.0.0.1.bias:torch.Size([21])', 'stage3.0.fuse_layers.2.0.0.1.running_mean:torch.Size([21])', 'stage3.0.fuse_layers.2.0.0.1.running_var:torch.Size([21])', 'stage3.0.fuse_layers.2.0.0.1.num_batches_tracked:torch.Size([])', 'stage3.0.fuse_layers.2.0.1.0.weight:torch.Size([83, 21, 3, 3])', 'stage3.0.fuse_layers.2.0.1.1.weight:torch.Size([83])', 'stage3.0.fuse_layers.2.0.1.1.bias:torch.Size([83])', 'stage3.0.fuse_layers.2.0.1.1.running_mean:torch.Size([83])', 'stage3.0.fuse_layers.2.0.1.1.running_var:torch.Size([83])', 'stage3.0.fuse_layers.2.0.1.1.num_batches_tracked:torch.Size([])', 'stage3.0.fuse_layers.2.1.0.0.weight:torch.Size([83, 42, 3, 3])', 'stage3.0.fuse_layers.2.1.0.1.weight:torch.Size([83])', 'stage3.0.fuse_layers.2.1.0.1.bias:torch.Size([83])', 'stage3.0.fuse_layers.2.1.0.1.running_mean:torch.Size([83])', 'stage3.0.fuse_layers.2.1.0.1.running_var:torch.Size([83])', 'stage3.0.fuse_layers.2.1.0.1.num_batches_tracked:torch.Size([])', 'stage3.1.branches.0.0.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage3.1.branches.0.0.bn1.weight:torch.Size([21])', 'stage3.1.branches.0.0.bn1.bias:torch.Size([21])', 'stage3.1.branches.0.0.bn1.running_mean:torch.Size([21])', 'stage3.1.branches.0.0.bn1.running_var:torch.Size([21])', 'stage3.1.branches.0.0.bn1.num_batches_tracked:torch.Size([])', 'stage3.1.branches.0.0.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage3.1.branches.0.0.bn2.weight:torch.Size([21])', 'stage3.1.branches.0.0.bn2.bias:torch.Size([21])', 'stage3.1.branches.0.0.bn2.running_mean:torch.Size([21])', 'stage3.1.branches.0.0.bn2.running_var:torch.Size([21])', 'stage3.1.branches.0.0.bn2.num_batches_tracked:torch.Size([])', 'stage3.1.branches.0.1.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage3.1.branches.0.1.bn1.weight:torch.Size([21])', 'stage3.1.branches.0.1.bn1.bias:torch.Size([21])', 'stage3.1.branches.0.1.bn1.running_mean:torch.Size([21])', 'stage3.1.branches.0.1.bn1.running_var:torch.Size([21])', 'stage3.1.branches.0.1.bn1.num_batches_tracked:torch.Size([])', 'stage3.1.branches.0.1.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage3.1.branches.0.1.bn2.weight:torch.Size([21])', 'stage3.1.branches.0.1.bn2.bias:torch.Size([21])', 'stage3.1.branches.0.1.bn2.running_mean:torch.Size([21])', 'stage3.1.branches.0.1.bn2.running_var:torch.Size([21])', 'stage3.1.branches.0.1.bn2.num_batches_tracked:torch.Size([])', 'stage3.1.branches.1.0.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage3.1.branches.1.0.bn1.weight:torch.Size([42])', 'stage3.1.branches.1.0.bn1.bias:torch.Size([42])', 'stage3.1.branches.1.0.bn1.running_mean:torch.Size([42])', 'stage3.1.branches.1.0.bn1.running_var:torch.Size([42])', 'stage3.1.branches.1.0.bn1.num_batches_tracked:torch.Size([])', 'stage3.1.branches.1.0.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage3.1.branches.1.0.bn2.weight:torch.Size([42])', 'stage3.1.branches.1.0.bn2.bias:torch.Size([42])', 'stage3.1.branches.1.0.bn2.running_mean:torch.Size([42])', 'stage3.1.branches.1.0.bn2.running_var:torch.Size([42])', 'stage3.1.branches.1.0.bn2.num_batches_tracked:torch.Size([])', 'stage3.1.branches.1.1.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage3.1.branches.1.1.bn1.weight:torch.Size([42])', 'stage3.1.branches.1.1.bn1.bias:torch.Size([42])', 'stage3.1.branches.1.1.bn1.running_mean:torch.Size([42])', 'stage3.1.branches.1.1.bn1.running_var:torch.Size([42])', 'stage3.1.branches.1.1.bn1.num_batches_tracked:torch.Size([])', 'stage3.1.branches.1.1.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage3.1.branches.1.1.bn2.weight:torch.Size([42])', 'stage3.1.branches.1.1.bn2.bias:torch.Size([42])', 'stage3.1.branches.1.1.bn2.running_mean:torch.Size([42])', 'stage3.1.branches.1.1.bn2.running_var:torch.Size([42])', 'stage3.1.branches.1.1.bn2.num_batches_tracked:torch.Size([])', 'stage3.1.branches.2.0.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage3.1.branches.2.0.bn1.weight:torch.Size([83])', 'stage3.1.branches.2.0.bn1.bias:torch.Size([83])', 'stage3.1.branches.2.0.bn1.running_mean:torch.Size([83])', 'stage3.1.branches.2.0.bn1.running_var:torch.Size([83])', 'stage3.1.branches.2.0.bn1.num_batches_tracked:torch.Size([])', 'stage3.1.branches.2.0.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage3.1.branches.2.0.bn2.weight:torch.Size([83])', 'stage3.1.branches.2.0.bn2.bias:torch.Size([83])', 'stage3.1.branches.2.0.bn2.running_mean:torch.Size([83])', 'stage3.1.branches.2.0.bn2.running_var:torch.Size([83])', 'stage3.1.branches.2.0.bn2.num_batches_tracked:torch.Size([])', 'stage3.1.branches.2.1.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage3.1.branches.2.1.bn1.weight:torch.Size([83])', 'stage3.1.branches.2.1.bn1.bias:torch.Size([83])', 'stage3.1.branches.2.1.bn1.running_mean:torch.Size([83])', 'stage3.1.branches.2.1.bn1.running_var:torch.Size([83])', 'stage3.1.branches.2.1.bn1.num_batches_tracked:torch.Size([])', 'stage3.1.branches.2.1.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage3.1.branches.2.1.bn2.weight:torch.Size([83])', 'stage3.1.branches.2.1.bn2.bias:torch.Size([83])', 'stage3.1.branches.2.1.bn2.running_mean:torch.Size([83])', 'stage3.1.branches.2.1.bn2.running_var:torch.Size([83])', 'stage3.1.branches.2.1.bn2.num_batches_tracked:torch.Size([])', 'stage3.1.fuse_layers.0.1.0.weight:torch.Size([21, 42, 1, 1])', 'stage3.1.fuse_layers.0.1.1.weight:torch.Size([21])', 'stage3.1.fuse_layers.0.1.1.bias:torch.Size([21])', 'stage3.1.fuse_layers.0.1.1.running_mean:torch.Size([21])', 'stage3.1.fuse_layers.0.1.1.running_var:torch.Size([21])', 'stage3.1.fuse_layers.0.1.1.num_batches_tracked:torch.Size([])', 'stage3.1.fuse_layers.0.2.0.weight:torch.Size([21, 83, 1, 1])', 'stage3.1.fuse_layers.0.2.1.weight:torch.Size([21])', 'stage3.1.fuse_layers.0.2.1.bias:torch.Size([21])', 'stage3.1.fuse_layers.0.2.1.running_mean:torch.Size([21])', 'stage3.1.fuse_layers.0.2.1.running_var:torch.Size([21])', 'stage3.1.fuse_layers.0.2.1.num_batches_tracked:torch.Size([])', 'stage3.1.fuse_layers.1.0.0.0.weight:torch.Size([42, 21, 3, 3])', 'stage3.1.fuse_layers.1.0.0.1.weight:torch.Size([42])', 'stage3.1.fuse_layers.1.0.0.1.bias:torch.Size([42])', 'stage3.1.fuse_layers.1.0.0.1.running_mean:torch.Size([42])', 'stage3.1.fuse_layers.1.0.0.1.running_var:torch.Size([42])', 'stage3.1.fuse_layers.1.0.0.1.num_batches_tracked:torch.Size([])', 'stage3.1.fuse_layers.1.2.0.weight:torch.Size([42, 83, 1, 1])', 'stage3.1.fuse_layers.1.2.1.weight:torch.Size([42])', 'stage3.1.fuse_layers.1.2.1.bias:torch.Size([42])', 'stage3.1.fuse_layers.1.2.1.running_mean:torch.Size([42])', 'stage3.1.fuse_layers.1.2.1.running_var:torch.Size([42])', 'stage3.1.fuse_layers.1.2.1.num_batches_tracked:torch.Size([])', 'stage3.1.fuse_layers.2.0.0.0.weight:torch.Size([21, 21, 3, 3])', 'stage3.1.fuse_layers.2.0.0.1.weight:torch.Size([21])', 'stage3.1.fuse_layers.2.0.0.1.bias:torch.Size([21])', 'stage3.1.fuse_layers.2.0.0.1.running_mean:torch.Size([21])', 'stage3.1.fuse_layers.2.0.0.1.running_var:torch.Size([21])', 'stage3.1.fuse_layers.2.0.0.1.num_batches_tracked:torch.Size([])', 'stage3.1.fuse_layers.2.0.1.0.weight:torch.Size([83, 21, 3, 3])', 'stage3.1.fuse_layers.2.0.1.1.weight:torch.Size([83])', 'stage3.1.fuse_layers.2.0.1.1.bias:torch.Size([83])', 'stage3.1.fuse_layers.2.0.1.1.running_mean:torch.Size([83])', 'stage3.1.fuse_layers.2.0.1.1.running_var:torch.Size([83])', 'stage3.1.fuse_layers.2.0.1.1.num_batches_tracked:torch.Size([])', 'stage3.1.fuse_layers.2.1.0.0.weight:torch.Size([83, 42, 3, 3])', 'stage3.1.fuse_layers.2.1.0.1.weight:torch.Size([83])', 'stage3.1.fuse_layers.2.1.0.1.bias:torch.Size([83])', 'stage3.1.fuse_layers.2.1.0.1.running_mean:torch.Size([83])', 'stage3.1.fuse_layers.2.1.0.1.running_var:torch.Size([83])', 'stage3.1.fuse_layers.2.1.0.1.num_batches_tracked:torch.Size([])', 'trans3_branch1.0.weight:torch.Size([21, 21, 3, 3])', 'trans3_branch1.0.bias:torch.Size([21])', 'trans3_branch1.1.weight:torch.Size([21])', 'trans3_branch1.1.bias:torch.Size([21])', 'trans3_branch1.1.running_mean:torch.Size([21])', 'trans3_branch1.1.running_var:torch.Size([21])', 'trans3_branch1.1.num_batches_tracked:torch.Size([])', 'trans3_branch2.0.weight:torch.Size([42, 42, 3, 3])', 'trans3_branch2.0.bias:torch.Size([42])', 'trans3_branch2.1.weight:torch.Size([42])', 'trans3_branch2.1.bias:torch.Size([42])', 'trans3_branch2.1.running_mean:torch.Size([42])', 'trans3_branch2.1.running_var:torch.Size([42])', 'trans3_branch2.1.num_batches_tracked:torch.Size([])', 'trans3_branch3.0.weight:torch.Size([83, 83, 3, 3])', 'trans3_branch3.0.bias:torch.Size([83])', 'trans3_branch3.1.weight:torch.Size([83])', 'trans3_branch3.1.bias:torch.Size([83])', 'trans3_branch3.1.running_mean:torch.Size([83])', 'trans3_branch3.1.running_var:torch.Size([83])', 'trans3_branch3.1.num_batches_tracked:torch.Size([])', 'trans3_branch4.0.weight:torch.Size([166, 264, 3, 3])', 'trans3_branch4.0.bias:torch.Size([166])', 'trans3_branch4.1.weight:torch.Size([166])', 'trans3_branch4.1.bias:torch.Size([166])', 'trans3_branch4.1.running_mean:torch.Size([166])', 'trans3_branch4.1.running_var:torch.Size([166])', 'trans3_branch4.1.num_batches_tracked:torch.Size([])', 'stage4.0.branches.0.0.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage4.0.branches.0.0.bn1.weight:torch.Size([21])', 'stage4.0.branches.0.0.bn1.bias:torch.Size([21])', 'stage4.0.branches.0.0.bn1.running_mean:torch.Size([21])', 'stage4.0.branches.0.0.bn1.running_var:torch.Size([21])', 'stage4.0.branches.0.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.0.branches.0.0.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage4.0.branches.0.0.bn2.weight:torch.Size([21])', 'stage4.0.branches.0.0.bn2.bias:torch.Size([21])', 'stage4.0.branches.0.0.bn2.running_mean:torch.Size([21])', 'stage4.0.branches.0.0.bn2.running_var:torch.Size([21])', 'stage4.0.branches.0.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.0.branches.0.1.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage4.0.branches.0.1.bn1.weight:torch.Size([21])', 'stage4.0.branches.0.1.bn1.bias:torch.Size([21])', 'stage4.0.branches.0.1.bn1.running_mean:torch.Size([21])', 'stage4.0.branches.0.1.bn1.running_var:torch.Size([21])', 'stage4.0.branches.0.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.0.branches.0.1.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage4.0.branches.0.1.bn2.weight:torch.Size([21])', 'stage4.0.branches.0.1.bn2.bias:torch.Size([21])', 'stage4.0.branches.0.1.bn2.running_mean:torch.Size([21])', 'stage4.0.branches.0.1.bn2.running_var:torch.Size([21])', 'stage4.0.branches.0.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.0.branches.1.0.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage4.0.branches.1.0.bn1.weight:torch.Size([42])', 'stage4.0.branches.1.0.bn1.bias:torch.Size([42])', 'stage4.0.branches.1.0.bn1.running_mean:torch.Size([42])', 'stage4.0.branches.1.0.bn1.running_var:torch.Size([42])', 'stage4.0.branches.1.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.0.branches.1.0.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage4.0.branches.1.0.bn2.weight:torch.Size([42])', 'stage4.0.branches.1.0.bn2.bias:torch.Size([42])', 'stage4.0.branches.1.0.bn2.running_mean:torch.Size([42])', 'stage4.0.branches.1.0.bn2.running_var:torch.Size([42])', 'stage4.0.branches.1.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.0.branches.1.1.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage4.0.branches.1.1.bn1.weight:torch.Size([42])', 'stage4.0.branches.1.1.bn1.bias:torch.Size([42])', 'stage4.0.branches.1.1.bn1.running_mean:torch.Size([42])', 'stage4.0.branches.1.1.bn1.running_var:torch.Size([42])', 'stage4.0.branches.1.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.0.branches.1.1.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage4.0.branches.1.1.bn2.weight:torch.Size([42])', 'stage4.0.branches.1.1.bn2.bias:torch.Size([42])', 'stage4.0.branches.1.1.bn2.running_mean:torch.Size([42])', 'stage4.0.branches.1.1.bn2.running_var:torch.Size([42])', 'stage4.0.branches.1.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.0.branches.2.0.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage4.0.branches.2.0.bn1.weight:torch.Size([83])', 'stage4.0.branches.2.0.bn1.bias:torch.Size([83])', 'stage4.0.branches.2.0.bn1.running_mean:torch.Size([83])', 'stage4.0.branches.2.0.bn1.running_var:torch.Size([83])', 'stage4.0.branches.2.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.0.branches.2.0.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage4.0.branches.2.0.bn2.weight:torch.Size([83])', 'stage4.0.branches.2.0.bn2.bias:torch.Size([83])', 'stage4.0.branches.2.0.bn2.running_mean:torch.Size([83])', 'stage4.0.branches.2.0.bn2.running_var:torch.Size([83])', 'stage4.0.branches.2.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.0.branches.2.1.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage4.0.branches.2.1.bn1.weight:torch.Size([83])', 'stage4.0.branches.2.1.bn1.bias:torch.Size([83])', 'stage4.0.branches.2.1.bn1.running_mean:torch.Size([83])', 'stage4.0.branches.2.1.bn1.running_var:torch.Size([83])', 'stage4.0.branches.2.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.0.branches.2.1.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage4.0.branches.2.1.bn2.weight:torch.Size([83])', 'stage4.0.branches.2.1.bn2.bias:torch.Size([83])', 'stage4.0.branches.2.1.bn2.running_mean:torch.Size([83])', 'stage4.0.branches.2.1.bn2.running_var:torch.Size([83])', 'stage4.0.branches.2.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.0.branches.3.0.conv1.weight:torch.Size([166, 166, 3, 3])', 'stage4.0.branches.3.0.bn1.weight:torch.Size([166])', 'stage4.0.branches.3.0.bn1.bias:torch.Size([166])', 'stage4.0.branches.3.0.bn1.running_mean:torch.Size([166])', 'stage4.0.branches.3.0.bn1.running_var:torch.Size([166])', 'stage4.0.branches.3.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.0.branches.3.0.conv2.weight:torch.Size([166, 166, 3, 3])', 'stage4.0.branches.3.0.bn2.weight:torch.Size([166])', 'stage4.0.branches.3.0.bn2.bias:torch.Size([166])', 'stage4.0.branches.3.0.bn2.running_mean:torch.Size([166])', 'stage4.0.branches.3.0.bn2.running_var:torch.Size([166])', 'stage4.0.branches.3.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.0.branches.3.1.conv1.weight:torch.Size([166, 166, 3, 3])', 'stage4.0.branches.3.1.bn1.weight:torch.Size([166])', 'stage4.0.branches.3.1.bn1.bias:torch.Size([166])', 'stage4.0.branches.3.1.bn1.running_mean:torch.Size([166])', 'stage4.0.branches.3.1.bn1.running_var:torch.Size([166])', 'stage4.0.branches.3.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.0.branches.3.1.conv2.weight:torch.Size([166, 166, 3, 3])', 'stage4.0.branches.3.1.bn2.weight:torch.Size([166])', 'stage4.0.branches.3.1.bn2.bias:torch.Size([166])', 'stage4.0.branches.3.1.bn2.running_mean:torch.Size([166])', 'stage4.0.branches.3.1.bn2.running_var:torch.Size([166])', 'stage4.0.branches.3.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.0.1.0.weight:torch.Size([21, 42, 1, 1])', 'stage4.0.fuse_layers.0.1.1.weight:torch.Size([21])', 'stage4.0.fuse_layers.0.1.1.bias:torch.Size([21])', 'stage4.0.fuse_layers.0.1.1.running_mean:torch.Size([21])', 'stage4.0.fuse_layers.0.1.1.running_var:torch.Size([21])', 'stage4.0.fuse_layers.0.1.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.0.2.0.weight:torch.Size([21, 83, 1, 1])', 'stage4.0.fuse_layers.0.2.1.weight:torch.Size([21])', 'stage4.0.fuse_layers.0.2.1.bias:torch.Size([21])', 'stage4.0.fuse_layers.0.2.1.running_mean:torch.Size([21])', 'stage4.0.fuse_layers.0.2.1.running_var:torch.Size([21])', 'stage4.0.fuse_layers.0.2.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.0.3.0.weight:torch.Size([21, 166, 1, 1])', 'stage4.0.fuse_layers.0.3.1.weight:torch.Size([21])', 'stage4.0.fuse_layers.0.3.1.bias:torch.Size([21])', 'stage4.0.fuse_layers.0.3.1.running_mean:torch.Size([21])', 'stage4.0.fuse_layers.0.3.1.running_var:torch.Size([21])', 'stage4.0.fuse_layers.0.3.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.1.0.0.0.weight:torch.Size([42, 21, 3, 3])', 'stage4.0.fuse_layers.1.0.0.1.weight:torch.Size([42])', 'stage4.0.fuse_layers.1.0.0.1.bias:torch.Size([42])', 'stage4.0.fuse_layers.1.0.0.1.running_mean:torch.Size([42])', 'stage4.0.fuse_layers.1.0.0.1.running_var:torch.Size([42])', 'stage4.0.fuse_layers.1.0.0.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.1.2.0.weight:torch.Size([42, 83, 1, 1])', 'stage4.0.fuse_layers.1.2.1.weight:torch.Size([42])', 'stage4.0.fuse_layers.1.2.1.bias:torch.Size([42])', 'stage4.0.fuse_layers.1.2.1.running_mean:torch.Size([42])', 'stage4.0.fuse_layers.1.2.1.running_var:torch.Size([42])', 'stage4.0.fuse_layers.1.2.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.1.3.0.weight:torch.Size([42, 166, 1, 1])', 'stage4.0.fuse_layers.1.3.1.weight:torch.Size([42])', 'stage4.0.fuse_layers.1.3.1.bias:torch.Size([42])', 'stage4.0.fuse_layers.1.3.1.running_mean:torch.Size([42])', 'stage4.0.fuse_layers.1.3.1.running_var:torch.Size([42])', 'stage4.0.fuse_layers.1.3.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.2.0.0.0.weight:torch.Size([21, 21, 3, 3])', 'stage4.0.fuse_layers.2.0.0.1.weight:torch.Size([21])', 'stage4.0.fuse_layers.2.0.0.1.bias:torch.Size([21])', 'stage4.0.fuse_layers.2.0.0.1.running_mean:torch.Size([21])', 'stage4.0.fuse_layers.2.0.0.1.running_var:torch.Size([21])', 'stage4.0.fuse_layers.2.0.0.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.2.0.1.0.weight:torch.Size([83, 21, 3, 3])', 'stage4.0.fuse_layers.2.0.1.1.weight:torch.Size([83])', 'stage4.0.fuse_layers.2.0.1.1.bias:torch.Size([83])', 'stage4.0.fuse_layers.2.0.1.1.running_mean:torch.Size([83])', 'stage4.0.fuse_layers.2.0.1.1.running_var:torch.Size([83])', 'stage4.0.fuse_layers.2.0.1.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.2.1.0.0.weight:torch.Size([83, 42, 3, 3])', 'stage4.0.fuse_layers.2.1.0.1.weight:torch.Size([83])', 'stage4.0.fuse_layers.2.1.0.1.bias:torch.Size([83])', 'stage4.0.fuse_layers.2.1.0.1.running_mean:torch.Size([83])', 'stage4.0.fuse_layers.2.1.0.1.running_var:torch.Size([83])', 'stage4.0.fuse_layers.2.1.0.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.2.3.0.weight:torch.Size([83, 166, 1, 1])', 'stage4.0.fuse_layers.2.3.1.weight:torch.Size([83])', 'stage4.0.fuse_layers.2.3.1.bias:torch.Size([83])', 'stage4.0.fuse_layers.2.3.1.running_mean:torch.Size([83])', 'stage4.0.fuse_layers.2.3.1.running_var:torch.Size([83])', 'stage4.0.fuse_layers.2.3.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.3.0.0.0.weight:torch.Size([21, 21, 3, 3])', 'stage4.0.fuse_layers.3.0.0.1.weight:torch.Size([21])', 'stage4.0.fuse_layers.3.0.0.1.bias:torch.Size([21])', 'stage4.0.fuse_layers.3.0.0.1.running_mean:torch.Size([21])', 'stage4.0.fuse_layers.3.0.0.1.running_var:torch.Size([21])', 'stage4.0.fuse_layers.3.0.0.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.3.0.1.0.weight:torch.Size([21, 21, 3, 3])', 'stage4.0.fuse_layers.3.0.1.1.weight:torch.Size([21])', 'stage4.0.fuse_layers.3.0.1.1.bias:torch.Size([21])', 'stage4.0.fuse_layers.3.0.1.1.running_mean:torch.Size([21])', 'stage4.0.fuse_layers.3.0.1.1.running_var:torch.Size([21])', 'stage4.0.fuse_layers.3.0.1.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.3.0.2.0.weight:torch.Size([166, 21, 3, 3])', 'stage4.0.fuse_layers.3.0.2.1.weight:torch.Size([166])', 'stage4.0.fuse_layers.3.0.2.1.bias:torch.Size([166])', 'stage4.0.fuse_layers.3.0.2.1.running_mean:torch.Size([166])', 'stage4.0.fuse_layers.3.0.2.1.running_var:torch.Size([166])', 'stage4.0.fuse_layers.3.0.2.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.3.1.0.0.weight:torch.Size([42, 42, 3, 3])', 'stage4.0.fuse_layers.3.1.0.1.weight:torch.Size([42])', 'stage4.0.fuse_layers.3.1.0.1.bias:torch.Size([42])', 'stage4.0.fuse_layers.3.1.0.1.running_mean:torch.Size([42])', 'stage4.0.fuse_layers.3.1.0.1.running_var:torch.Size([42])', 'stage4.0.fuse_layers.3.1.0.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.3.1.1.0.weight:torch.Size([166, 42, 3, 3])', 'stage4.0.fuse_layers.3.1.1.1.weight:torch.Size([166])', 'stage4.0.fuse_layers.3.1.1.1.bias:torch.Size([166])', 'stage4.0.fuse_layers.3.1.1.1.running_mean:torch.Size([166])', 'stage4.0.fuse_layers.3.1.1.1.running_var:torch.Size([166])', 'stage4.0.fuse_layers.3.1.1.1.num_batches_tracked:torch.Size([])', 'stage4.0.fuse_layers.3.2.0.0.weight:torch.Size([166, 83, 3, 3])', 'stage4.0.fuse_layers.3.2.0.1.weight:torch.Size([166])', 'stage4.0.fuse_layers.3.2.0.1.bias:torch.Size([166])', 'stage4.0.fuse_layers.3.2.0.1.running_mean:torch.Size([166])', 'stage4.0.fuse_layers.3.2.0.1.running_var:torch.Size([166])', 'stage4.0.fuse_layers.3.2.0.1.num_batches_tracked:torch.Size([])', 'stage4.1.branches.0.0.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage4.1.branches.0.0.bn1.weight:torch.Size([21])', 'stage4.1.branches.0.0.bn1.bias:torch.Size([21])', 'stage4.1.branches.0.0.bn1.running_mean:torch.Size([21])', 'stage4.1.branches.0.0.bn1.running_var:torch.Size([21])', 'stage4.1.branches.0.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.1.branches.0.0.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage4.1.branches.0.0.bn2.weight:torch.Size([21])', 'stage4.1.branches.0.0.bn2.bias:torch.Size([21])', 'stage4.1.branches.0.0.bn2.running_mean:torch.Size([21])', 'stage4.1.branches.0.0.bn2.running_var:torch.Size([21])', 'stage4.1.branches.0.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.1.branches.0.1.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage4.1.branches.0.1.bn1.weight:torch.Size([21])', 'stage4.1.branches.0.1.bn1.bias:torch.Size([21])', 'stage4.1.branches.0.1.bn1.running_mean:torch.Size([21])', 'stage4.1.branches.0.1.bn1.running_var:torch.Size([21])', 'stage4.1.branches.0.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.1.branches.0.1.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage4.1.branches.0.1.bn2.weight:torch.Size([21])', 'stage4.1.branches.0.1.bn2.bias:torch.Size([21])', 'stage4.1.branches.0.1.bn2.running_mean:torch.Size([21])', 'stage4.1.branches.0.1.bn2.running_var:torch.Size([21])', 'stage4.1.branches.0.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.1.branches.1.0.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage4.1.branches.1.0.bn1.weight:torch.Size([42])', 'stage4.1.branches.1.0.bn1.bias:torch.Size([42])', 'stage4.1.branches.1.0.bn1.running_mean:torch.Size([42])', 'stage4.1.branches.1.0.bn1.running_var:torch.Size([42])', 'stage4.1.branches.1.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.1.branches.1.0.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage4.1.branches.1.0.bn2.weight:torch.Size([42])', 'stage4.1.branches.1.0.bn2.bias:torch.Size([42])', 'stage4.1.branches.1.0.bn2.running_mean:torch.Size([42])', 'stage4.1.branches.1.0.bn2.running_var:torch.Size([42])', 'stage4.1.branches.1.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.1.branches.1.1.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage4.1.branches.1.1.bn1.weight:torch.Size([42])', 'stage4.1.branches.1.1.bn1.bias:torch.Size([42])', 'stage4.1.branches.1.1.bn1.running_mean:torch.Size([42])', 'stage4.1.branches.1.1.bn1.running_var:torch.Size([42])', 'stage4.1.branches.1.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.1.branches.1.1.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage4.1.branches.1.1.bn2.weight:torch.Size([42])', 'stage4.1.branches.1.1.bn2.bias:torch.Size([42])', 'stage4.1.branches.1.1.bn2.running_mean:torch.Size([42])', 'stage4.1.branches.1.1.bn2.running_var:torch.Size([42])', 'stage4.1.branches.1.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.1.branches.2.0.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage4.1.branches.2.0.bn1.weight:torch.Size([83])', 'stage4.1.branches.2.0.bn1.bias:torch.Size([83])', 'stage4.1.branches.2.0.bn1.running_mean:torch.Size([83])', 'stage4.1.branches.2.0.bn1.running_var:torch.Size([83])', 'stage4.1.branches.2.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.1.branches.2.0.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage4.1.branches.2.0.bn2.weight:torch.Size([83])', 'stage4.1.branches.2.0.bn2.bias:torch.Size([83])', 'stage4.1.branches.2.0.bn2.running_mean:torch.Size([83])', 'stage4.1.branches.2.0.bn2.running_var:torch.Size([83])', 'stage4.1.branches.2.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.1.branches.2.1.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage4.1.branches.2.1.bn1.weight:torch.Size([83])', 'stage4.1.branches.2.1.bn1.bias:torch.Size([83])', 'stage4.1.branches.2.1.bn1.running_mean:torch.Size([83])', 'stage4.1.branches.2.1.bn1.running_var:torch.Size([83])', 'stage4.1.branches.2.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.1.branches.2.1.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage4.1.branches.2.1.bn2.weight:torch.Size([83])', 'stage4.1.branches.2.1.bn2.bias:torch.Size([83])', 'stage4.1.branches.2.1.bn2.running_mean:torch.Size([83])', 'stage4.1.branches.2.1.bn2.running_var:torch.Size([83])', 'stage4.1.branches.2.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.1.branches.3.0.conv1.weight:torch.Size([166, 166, 3, 3])', 'stage4.1.branches.3.0.bn1.weight:torch.Size([166])', 'stage4.1.branches.3.0.bn1.bias:torch.Size([166])', 'stage4.1.branches.3.0.bn1.running_mean:torch.Size([166])', 'stage4.1.branches.3.0.bn1.running_var:torch.Size([166])', 'stage4.1.branches.3.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.1.branches.3.0.conv2.weight:torch.Size([166, 166, 3, 3])', 'stage4.1.branches.3.0.bn2.weight:torch.Size([166])', 'stage4.1.branches.3.0.bn2.bias:torch.Size([166])', 'stage4.1.branches.3.0.bn2.running_mean:torch.Size([166])', 'stage4.1.branches.3.0.bn2.running_var:torch.Size([166])', 'stage4.1.branches.3.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.1.branches.3.1.conv1.weight:torch.Size([166, 166, 3, 3])', 'stage4.1.branches.3.1.bn1.weight:torch.Size([166])', 'stage4.1.branches.3.1.bn1.bias:torch.Size([166])', 'stage4.1.branches.3.1.bn1.running_mean:torch.Size([166])', 'stage4.1.branches.3.1.bn1.running_var:torch.Size([166])', 'stage4.1.branches.3.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.1.branches.3.1.conv2.weight:torch.Size([166, 166, 3, 3])', 'stage4.1.branches.3.1.bn2.weight:torch.Size([166])', 'stage4.1.branches.3.1.bn2.bias:torch.Size([166])', 'stage4.1.branches.3.1.bn2.running_mean:torch.Size([166])', 'stage4.1.branches.3.1.bn2.running_var:torch.Size([166])', 'stage4.1.branches.3.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.0.1.0.weight:torch.Size([21, 42, 1, 1])', 'stage4.1.fuse_layers.0.1.1.weight:torch.Size([21])', 'stage4.1.fuse_layers.0.1.1.bias:torch.Size([21])', 'stage4.1.fuse_layers.0.1.1.running_mean:torch.Size([21])', 'stage4.1.fuse_layers.0.1.1.running_var:torch.Size([21])', 'stage4.1.fuse_layers.0.1.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.0.2.0.weight:torch.Size([21, 83, 1, 1])', 'stage4.1.fuse_layers.0.2.1.weight:torch.Size([21])', 'stage4.1.fuse_layers.0.2.1.bias:torch.Size([21])', 'stage4.1.fuse_layers.0.2.1.running_mean:torch.Size([21])', 'stage4.1.fuse_layers.0.2.1.running_var:torch.Size([21])', 'stage4.1.fuse_layers.0.2.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.0.3.0.weight:torch.Size([21, 166, 1, 1])', 'stage4.1.fuse_layers.0.3.1.weight:torch.Size([21])', 'stage4.1.fuse_layers.0.3.1.bias:torch.Size([21])', 'stage4.1.fuse_layers.0.3.1.running_mean:torch.Size([21])', 'stage4.1.fuse_layers.0.3.1.running_var:torch.Size([21])', 'stage4.1.fuse_layers.0.3.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.1.0.0.0.weight:torch.Size([42, 21, 3, 3])', 'stage4.1.fuse_layers.1.0.0.1.weight:torch.Size([42])', 'stage4.1.fuse_layers.1.0.0.1.bias:torch.Size([42])', 'stage4.1.fuse_layers.1.0.0.1.running_mean:torch.Size([42])', 'stage4.1.fuse_layers.1.0.0.1.running_var:torch.Size([42])', 'stage4.1.fuse_layers.1.0.0.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.1.2.0.weight:torch.Size([42, 83, 1, 1])', 'stage4.1.fuse_layers.1.2.1.weight:torch.Size([42])', 'stage4.1.fuse_layers.1.2.1.bias:torch.Size([42])', 'stage4.1.fuse_layers.1.2.1.running_mean:torch.Size([42])', 'stage4.1.fuse_layers.1.2.1.running_var:torch.Size([42])', 'stage4.1.fuse_layers.1.2.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.1.3.0.weight:torch.Size([42, 166, 1, 1])', 'stage4.1.fuse_layers.1.3.1.weight:torch.Size([42])', 'stage4.1.fuse_layers.1.3.1.bias:torch.Size([42])', 'stage4.1.fuse_layers.1.3.1.running_mean:torch.Size([42])', 'stage4.1.fuse_layers.1.3.1.running_var:torch.Size([42])', 'stage4.1.fuse_layers.1.3.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.2.0.0.0.weight:torch.Size([21, 21, 3, 3])', 'stage4.1.fuse_layers.2.0.0.1.weight:torch.Size([21])', 'stage4.1.fuse_layers.2.0.0.1.bias:torch.Size([21])', 'stage4.1.fuse_layers.2.0.0.1.running_mean:torch.Size([21])', 'stage4.1.fuse_layers.2.0.0.1.running_var:torch.Size([21])', 'stage4.1.fuse_layers.2.0.0.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.2.0.1.0.weight:torch.Size([83, 21, 3, 3])', 'stage4.1.fuse_layers.2.0.1.1.weight:torch.Size([83])', 'stage4.1.fuse_layers.2.0.1.1.bias:torch.Size([83])', 'stage4.1.fuse_layers.2.0.1.1.running_mean:torch.Size([83])', 'stage4.1.fuse_layers.2.0.1.1.running_var:torch.Size([83])', 'stage4.1.fuse_layers.2.0.1.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.2.1.0.0.weight:torch.Size([83, 42, 3, 3])', 'stage4.1.fuse_layers.2.1.0.1.weight:torch.Size([83])', 'stage4.1.fuse_layers.2.1.0.1.bias:torch.Size([83])', 'stage4.1.fuse_layers.2.1.0.1.running_mean:torch.Size([83])', 'stage4.1.fuse_layers.2.1.0.1.running_var:torch.Size([83])', 'stage4.1.fuse_layers.2.1.0.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.2.3.0.weight:torch.Size([83, 166, 1, 1])', 'stage4.1.fuse_layers.2.3.1.weight:torch.Size([83])', 'stage4.1.fuse_layers.2.3.1.bias:torch.Size([83])', 'stage4.1.fuse_layers.2.3.1.running_mean:torch.Size([83])', 'stage4.1.fuse_layers.2.3.1.running_var:torch.Size([83])', 'stage4.1.fuse_layers.2.3.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.3.0.0.0.weight:torch.Size([21, 21, 3, 3])', 'stage4.1.fuse_layers.3.0.0.1.weight:torch.Size([21])', 'stage4.1.fuse_layers.3.0.0.1.bias:torch.Size([21])', 'stage4.1.fuse_layers.3.0.0.1.running_mean:torch.Size([21])', 'stage4.1.fuse_layers.3.0.0.1.running_var:torch.Size([21])', 'stage4.1.fuse_layers.3.0.0.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.3.0.1.0.weight:torch.Size([21, 21, 3, 3])', 'stage4.1.fuse_layers.3.0.1.1.weight:torch.Size([21])', 'stage4.1.fuse_layers.3.0.1.1.bias:torch.Size([21])', 'stage4.1.fuse_layers.3.0.1.1.running_mean:torch.Size([21])', 'stage4.1.fuse_layers.3.0.1.1.running_var:torch.Size([21])', 'stage4.1.fuse_layers.3.0.1.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.3.0.2.0.weight:torch.Size([166, 21, 3, 3])', 'stage4.1.fuse_layers.3.0.2.1.weight:torch.Size([166])', 'stage4.1.fuse_layers.3.0.2.1.bias:torch.Size([166])', 'stage4.1.fuse_layers.3.0.2.1.running_mean:torch.Size([166])', 'stage4.1.fuse_layers.3.0.2.1.running_var:torch.Size([166])', 'stage4.1.fuse_layers.3.0.2.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.3.1.0.0.weight:torch.Size([42, 42, 3, 3])', 'stage4.1.fuse_layers.3.1.0.1.weight:torch.Size([42])', 'stage4.1.fuse_layers.3.1.0.1.bias:torch.Size([42])', 'stage4.1.fuse_layers.3.1.0.1.running_mean:torch.Size([42])', 'stage4.1.fuse_layers.3.1.0.1.running_var:torch.Size([42])', 'stage4.1.fuse_layers.3.1.0.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.3.1.1.0.weight:torch.Size([166, 42, 3, 3])', 'stage4.1.fuse_layers.3.1.1.1.weight:torch.Size([166])', 'stage4.1.fuse_layers.3.1.1.1.bias:torch.Size([166])', 'stage4.1.fuse_layers.3.1.1.1.running_mean:torch.Size([166])', 'stage4.1.fuse_layers.3.1.1.1.running_var:torch.Size([166])', 'stage4.1.fuse_layers.3.1.1.1.num_batches_tracked:torch.Size([])', 'stage4.1.fuse_layers.3.2.0.0.weight:torch.Size([166, 83, 3, 3])', 'stage4.1.fuse_layers.3.2.0.1.weight:torch.Size([166])', 'stage4.1.fuse_layers.3.2.0.1.bias:torch.Size([166])', 'stage4.1.fuse_layers.3.2.0.1.running_mean:torch.Size([166])', 'stage4.1.fuse_layers.3.2.0.1.running_var:torch.Size([166])', 'stage4.1.fuse_layers.3.2.0.1.num_batches_tracked:torch.Size([])', 'stage4.2.branches.0.0.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage4.2.branches.0.0.bn1.weight:torch.Size([21])', 'stage4.2.branches.0.0.bn1.bias:torch.Size([21])', 'stage4.2.branches.0.0.bn1.running_mean:torch.Size([21])', 'stage4.2.branches.0.0.bn1.running_var:torch.Size([21])', 'stage4.2.branches.0.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.2.branches.0.0.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage4.2.branches.0.0.bn2.weight:torch.Size([21])', 'stage4.2.branches.0.0.bn2.bias:torch.Size([21])', 'stage4.2.branches.0.0.bn2.running_mean:torch.Size([21])', 'stage4.2.branches.0.0.bn2.running_var:torch.Size([21])', 'stage4.2.branches.0.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.2.branches.0.1.conv1.weight:torch.Size([21, 21, 3, 3])', 'stage4.2.branches.0.1.bn1.weight:torch.Size([21])', 'stage4.2.branches.0.1.bn1.bias:torch.Size([21])', 'stage4.2.branches.0.1.bn1.running_mean:torch.Size([21])', 'stage4.2.branches.0.1.bn1.running_var:torch.Size([21])', 'stage4.2.branches.0.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.2.branches.0.1.conv2.weight:torch.Size([21, 21, 3, 3])', 'stage4.2.branches.0.1.bn2.weight:torch.Size([21])', 'stage4.2.branches.0.1.bn2.bias:torch.Size([21])', 'stage4.2.branches.0.1.bn2.running_mean:torch.Size([21])', 'stage4.2.branches.0.1.bn2.running_var:torch.Size([21])', 'stage4.2.branches.0.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.2.branches.1.0.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage4.2.branches.1.0.bn1.weight:torch.Size([42])', 'stage4.2.branches.1.0.bn1.bias:torch.Size([42])', 'stage4.2.branches.1.0.bn1.running_mean:torch.Size([42])', 'stage4.2.branches.1.0.bn1.running_var:torch.Size([42])', 'stage4.2.branches.1.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.2.branches.1.0.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage4.2.branches.1.0.bn2.weight:torch.Size([42])', 'stage4.2.branches.1.0.bn2.bias:torch.Size([42])', 'stage4.2.branches.1.0.bn2.running_mean:torch.Size([42])', 'stage4.2.branches.1.0.bn2.running_var:torch.Size([42])', 'stage4.2.branches.1.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.2.branches.1.1.conv1.weight:torch.Size([42, 42, 3, 3])', 'stage4.2.branches.1.1.bn1.weight:torch.Size([42])', 'stage4.2.branches.1.1.bn1.bias:torch.Size([42])', 'stage4.2.branches.1.1.bn1.running_mean:torch.Size([42])', 'stage4.2.branches.1.1.bn1.running_var:torch.Size([42])', 'stage4.2.branches.1.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.2.branches.1.1.conv2.weight:torch.Size([42, 42, 3, 3])', 'stage4.2.branches.1.1.bn2.weight:torch.Size([42])', 'stage4.2.branches.1.1.bn2.bias:torch.Size([42])', 'stage4.2.branches.1.1.bn2.running_mean:torch.Size([42])', 'stage4.2.branches.1.1.bn2.running_var:torch.Size([42])', 'stage4.2.branches.1.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.2.branches.2.0.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage4.2.branches.2.0.bn1.weight:torch.Size([83])', 'stage4.2.branches.2.0.bn1.bias:torch.Size([83])', 'stage4.2.branches.2.0.bn1.running_mean:torch.Size([83])', 'stage4.2.branches.2.0.bn1.running_var:torch.Size([83])', 'stage4.2.branches.2.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.2.branches.2.0.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage4.2.branches.2.0.bn2.weight:torch.Size([83])', 'stage4.2.branches.2.0.bn2.bias:torch.Size([83])', 'stage4.2.branches.2.0.bn2.running_mean:torch.Size([83])', 'stage4.2.branches.2.0.bn2.running_var:torch.Size([83])', 'stage4.2.branches.2.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.2.branches.2.1.conv1.weight:torch.Size([83, 83, 3, 3])', 'stage4.2.branches.2.1.bn1.weight:torch.Size([83])', 'stage4.2.branches.2.1.bn1.bias:torch.Size([83])', 'stage4.2.branches.2.1.bn1.running_mean:torch.Size([83])', 'stage4.2.branches.2.1.bn1.running_var:torch.Size([83])', 'stage4.2.branches.2.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.2.branches.2.1.conv2.weight:torch.Size([83, 83, 3, 3])', 'stage4.2.branches.2.1.bn2.weight:torch.Size([83])', 'stage4.2.branches.2.1.bn2.bias:torch.Size([83])', 'stage4.2.branches.2.1.bn2.running_mean:torch.Size([83])', 'stage4.2.branches.2.1.bn2.running_var:torch.Size([83])', 'stage4.2.branches.2.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.2.branches.3.0.conv1.weight:torch.Size([166, 166, 3, 3])', 'stage4.2.branches.3.0.bn1.weight:torch.Size([166])', 'stage4.2.branches.3.0.bn1.bias:torch.Size([166])', 'stage4.2.branches.3.0.bn1.running_mean:torch.Size([166])', 'stage4.2.branches.3.0.bn1.running_var:torch.Size([166])', 'stage4.2.branches.3.0.bn1.num_batches_tracked:torch.Size([])', 'stage4.2.branches.3.0.conv2.weight:torch.Size([166, 166, 3, 3])', 'stage4.2.branches.3.0.bn2.weight:torch.Size([166])', 'stage4.2.branches.3.0.bn2.bias:torch.Size([166])', 'stage4.2.branches.3.0.bn2.running_mean:torch.Size([166])', 'stage4.2.branches.3.0.bn2.running_var:torch.Size([166])', 'stage4.2.branches.3.0.bn2.num_batches_tracked:torch.Size([])', 'stage4.2.branches.3.1.conv1.weight:torch.Size([166, 166, 3, 3])', 'stage4.2.branches.3.1.bn1.weight:torch.Size([166])', 'stage4.2.branches.3.1.bn1.bias:torch.Size([166])', 'stage4.2.branches.3.1.bn1.running_mean:torch.Size([166])', 'stage4.2.branches.3.1.bn1.running_var:torch.Size([166])', 'stage4.2.branches.3.1.bn1.num_batches_tracked:torch.Size([])', 'stage4.2.branches.3.1.conv2.weight:torch.Size([166, 166, 3, 3])', 'stage4.2.branches.3.1.bn2.weight:torch.Size([166])', 'stage4.2.branches.3.1.bn2.bias:torch.Size([166])', 'stage4.2.branches.3.1.bn2.running_mean:torch.Size([166])', 'stage4.2.branches.3.1.bn2.running_var:torch.Size([166])', 'stage4.2.branches.3.1.bn2.num_batches_tracked:torch.Size([])', 'stage4.2.fuse_layers.0.1.0.weight:torch.Size([21, 42, 1, 1])', 'stage4.2.fuse_layers.0.1.1.weight:torch.Size([21])', 'stage4.2.fuse_layers.0.1.1.bias:torch.Size([21])', 'stage4.2.fuse_layers.0.1.1.running_mean:torch.Size([21])', 'stage4.2.fuse_layers.0.1.1.running_var:torch.Size([21])', 'stage4.2.fuse_layers.0.1.1.num_batches_tracked:torch.Size([])', 'stage4.2.fuse_layers.0.2.0.weight:torch.Size([21, 83, 1, 1])', 'stage4.2.fuse_layers.0.2.1.weight:torch.Size([21])', 'stage4.2.fuse_layers.0.2.1.bias:torch.Size([21])', 'stage4.2.fuse_layers.0.2.1.running_mean:torch.Size([21])', 'stage4.2.fuse_layers.0.2.1.running_var:torch.Size([21])', 'stage4.2.fuse_layers.0.2.1.num_batches_tracked:torch.Size([])', 'stage4.2.fuse_layers.0.3.0.weight:torch.Size([21, 166, 1, 1])', 'stage4.2.fuse_layers.0.3.1.weight:torch.Size([21])', 'stage4.2.fuse_layers.0.3.1.bias:torch.Size([21])', 'stage4.2.fuse_layers.0.3.1.running_mean:torch.Size([21])', 'stage4.2.fuse_layers.0.3.1.running_var:torch.Size([21])', 'stage4.2.fuse_layers.0.3.1.num_batches_tracked:torch.Size([])', 'final_layers.0.weight:torch.Size([34, 21, 1, 1])', 'final_layers.0.bias:torch.Size([34])', 'final_layers.1.weight:torch.Size([17, 21, 1, 1])', 'final_layers.1.bias:torch.Size([17])', 'deconv_layers.0.0.0.weight:torch.Size([55, 21, 4, 4])', 'deconv_layers.0.0.1.weight:torch.Size([21])', 'deconv_layers.0.0.1.bias:torch.Size([21])', 'deconv_layers.0.0.1.running_mean:torch.Size([21])', 'deconv_layers.0.0.1.running_var:torch.Size([21])', 'deconv_layers.0.0.1.num_batches_tracked:torch.Size([])', 'deconv_layers.0.1.0.conv1.weight:torch.Size([21, 21, 3, 3])', 'deconv_layers.0.1.0.bn1.weight:torch.Size([21])', 'deconv_layers.0.1.0.bn1.bias:torch.Size([21])', 'deconv_layers.0.1.0.bn1.running_mean:torch.Size([21])', 'deconv_layers.0.1.0.bn1.running_var:torch.Size([21])', 'deconv_layers.0.1.0.bn1.num_batches_tracked:torch.Size([])', 'deconv_layers.0.1.0.conv2.weight:torch.Size([21, 21, 3, 3])', 'deconv_layers.0.1.0.bn2.weight:torch.Size([21])', 'deconv_layers.0.1.0.bn2.bias:torch.Size([21])', 'deconv_layers.0.1.0.bn2.running_mean:torch.Size([21])', 'deconv_layers.0.1.0.bn2.running_var:torch.Size([21])', 'deconv_layers.0.1.0.bn2.num_batches_tracked:torch.Size([])', 'deconv_layers.0.2.0.conv1.weight:torch.Size([21, 21, 3, 3])', 'deconv_layers.0.2.0.bn1.weight:torch.Size([21])', 'deconv_layers.0.2.0.bn1.bias:torch.Size([21])', 'deconv_layers.0.2.0.bn1.running_mean:torch.Size([21])', 'deconv_layers.0.2.0.bn1.running_var:torch.Size([21])', 'deconv_layers.0.2.0.bn1.num_batches_tracked:torch.Size([])', 'deconv_layers.0.2.0.conv2.weight:torch.Size([21, 21, 3, 3])', 'deconv_layers.0.2.0.bn2.weight:torch.Size([21])', 'deconv_layers.0.2.0.bn2.bias:torch.Size([21])', 'deconv_layers.0.2.0.bn2.running_mean:torch.Size([21])', 'deconv_layers.0.2.0.bn2.running_var:torch.Size([21])', 'deconv_layers.0.2.0.bn2.num_batches_tracked:torch.Size([])']
标签:读取,weight,torch,21,running,fuse,网络结构,efficienthrnetH,Size From: https://www.cnblogs.com/hahaah/p/17525599.html