-
Implement Label Smoothing
Write a function to apply label smoothing for classification: - Replace one-hot targets with $$1-\epsilon$$ for true class, $$\epsilon/(K-1)$$ for others. - Use it in cross-entropy training. Show...
-
Save and Load TorchScript Model
Convert a trained PyTorch model to TorchScript via tracing and scripting. Save it to disk. Reload and run inference. Compare outputs with the original model.
-
Mixed Precision Training with autocast
Modify a training loop to use `torch.cuda.amp.autocast`: - Wrap forward + loss in `autocast`. - Use `GradScaler` for backward. Compare training speed vs. full precision.