PerceptualNet

class torchelie.models.PerceptualNet(layers: List[str], use_avg_pool: bool = True, remove_unused_layers: bool = True)

Make a VGG16 with appropriately named layers that records intermediate activations.

Parameters

  • layers (list of str) – the names of the layers for which to save the activations.

  • use_avg_pool (bool) – Whether to replace max pooling with averange pooling (default: True)

  • remove_unused_layers (bool) – whether to remove layers past the last one used (default: True)

training: bool