# mnist mlp

## Summary
- Task: classification
- Epochs: 10
- Final val loss: 0.078887
- Final val accuracy: 0.981200

## Artifacts
- model.pt: model weights
- history.json: per-epoch training and validation metrics
- train_features.pt: training-set feature embeddings and targets
- val_features.pt: validation-set feature embeddings and targets
- meta.json: run metadata

## Metric Keys
- epoch, train_loss, val_loss, train_acc, val_acc