Glossary · Phase 4 · Deep learning
Checkpoint
A saved copy of a model's weights at some point in training, so you can go back to it or resume later.
Try it now
Every neural network ever trained runs this loop. Forgetting the third step is a classic bug.
Learn it properly
The training loop · Forward, loss, backward, step, repeat. About 35 minutes.
Other terms from the same lesson
- state_dict: PyTorch's dictionary of a model's weights by name. Saving and loading it is how you keep or restore a trained model.
- training loop: The repeated cycle of training: predict, measure the loss, backpropagate, update the weights.
Never trained an AI?
Teach a small one to write, right in your browser. It takes about 20 seconds and needs no account.