Kevin Musgrave issues

Results 79 issues of


                                            Kevin Musgrave

Make CustomKNN usable in Inference or explain that it can't be used

See #429

enhancement

Create example to show how SelfSupervisedLoss works

Also, the existing MoCo notebook should probably use SelfSupervisedLoss.

documentation

Remove `indices_tuple` argument from CrossBatchMemory

It doesn't make sense anymore

enhancement

Convert MPerClassSampler to a `batch_sampler`

enhancement

Simplify reducers

Reducers are way too complicated

enhancement

Add loss batching code as a utility function

See: - https://github.com/KevinMusgrave/pytorch-metric-learning/issues/547#issuecomment-1304907510 - https://github.com/KevinMusgrave/pytorch-metric-learning/issues/6#issuecomment-1416870096

enhancement

Allow multiple reference embeddings in SelfSupervisedLoss

As originally discussed here https://github.com/KevinMusgrave/pytorch-metric-learning/issues/411

enhancement

Add documentation on how to use losses/miners like in the corresponding papers

See this discussion: https://github.com/KevinMusgrave/pytorch-metric-learning/discussions/561#discussioncomment-4580712

documentation

chore: Remove +0.01. Make example command easier to copy-paste

Removes the `+0.01` from the rank metric in the unmanaged example.

cla-signed

docs: Add a link to `step_optimizer`

## Description Added a link to `step_optimizer` since it wasn't clear to me where the automatic zero grad should be specified. - [x] content meets style guide guidelines - [x]...

documentation

cla-signed