-
Checkpointing with torch.save
Train a simple feedforward model for 1 epoch. Save: 1. Model state dict. 2. Optimizer state dict. 3. Epoch number. Then load the checkpoint and resume training seamlessly.
1