ignite
https://github.com/pytorch/ignite
Python
High-level library to help with training and evaluating neural networks in PyTorch flexibly and transparently.
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported3 Subscribers
Add a CodeTriage badge to ignite
Help out
- Issues
- fix(metrics): reject normalized confusion matrices in derived metrics
- JSDivergence returns NaN for finite logits after softmax underflow
- Entropy and MutualInformation return NaN for zero probabilities from finite logits
- Avoid NaN entropy contributions from zero probabilities
- Fix `JSDivergence` NaN from softmax underflow
- cmAccuracy, cmPrecision, cmRecall and DiceCoefficient accept a normalized ConfusionMatrix (average="recall"/"precision") and silently return wrong values, unlike IoU
- `load_state_dict` fails for metrics that accumulate into lists, breaking the documented resume recipe
- `BatchSizeScheduler` silently re-iterates the dataset because `epoch_length` is never recomputed
- Fix non-tuple sequence indexing in all_gather_tensors_with_shapes
- Fall back to first visible GPU when local rank exceeds device count
- Docs
- Python not yet supported