how to truncate a base network

truncate.py

class YourCustomModel(nn.Module):
    def __init__(self):
     	super().__init__()
	# truncated base network, „True“ refers to pretrained
    	self.backbone = nn.Sequential(*list(resnet34(True).children())[:8])
     
	# and your custom layers
	self.features = nn.Sequential(
   	self.backbone,
     	# custom layers:
	AdaptiveConcatPool2d(1),
  	Flatten(),
   	nn.BatchNorm1d(1024),
   	nn.Dropout(0.1),
  	nn.Linear(in_features=1024, out_features=1024, bias=True),
   	#… more layers
        )

    def forward(self, x):
       	return self.features(x)