how do i load saved “state_dict” on 3rd and 4th layer of ResNet-152

[prashant_bahuguna]

1. i tried copying weight manually from the state dictionary however I am only able to load a single layer

   

2. then i thought of manually adding dictionary words but the process is impractical

[ptrblck]

If some layer names are equal, you could load these parameters via load_state_dict(..., strict=False).

class MyModel(nn.Module):
    def __init__(self):
        super(MyModel, self).__init__()
        self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False)
        self.new_layer = nn.Conv2d(64, 64, 3, 1, 1)
    
    def forward(self, x):
        x = self.conv1(x)
        x = self.new_layer(x)
        return x

ref_model = models.resnet152()
model = MyModel()
model.load_state_dict(ref_model.state_dict(), strict=False)

# compare
print((model.conv1.weight == ref_model.conv1.weight).all())
> tensor(True)

Note that strict=False will ignore all incompatible keys and you should double check, that the wanted parameters were indeed copied.