ML Katas

Implement Early Stopping

medium (<1 hr) pytorch training basics early stopping

this month by E

Add early stopping to a training loop:

Monitor validation loss.
Stop training if no improvement after 5 epochs.
Save best model checkpoint.

Demonstrate on MNIST subset.