Finalize pruning by removing the mask
After pruning a model, you need to finalize it before saving or deploying. This means removing the pruning mask so the model no longer tracks which weights were zeroed. You'll use prune.remove() to complete this step.
Den här övningen är en del av kursen
Scalable AI Models with PyTorch Lightning
Övningsinstruktioner
- Finalize the pruning of the weight parameter.
- Use
prune.remove()to remove the mask and store the final weights. - Print the model to confirm pruning removal.
Interaktiv övning med praktiskt arbete
Testa den här övningen genom att slutföra den här exempelkoden.
import torch.nn.utils.prune as prune
# Before pruning
print(model)
# Finalize pruning by removing the pruning mask
prune.____(model[3], "____")
# Print model structure after pruning
print(____)