Hi,
If I were to prune a pre-trained model (like a resnet18 for example) using a pruning option like torch.nn.utils.prune.random_structured()
, how can I handle dependencies on those layers such as skip connections and concatenations?
Thank you
Hi,
If I were to prune a pre-trained model (like a resnet18 for example) using a pruning option like torch.nn.utils.prune.random_structured()
, how can I handle dependencies on those layers such as skip connections and concatenations?
Thank you