Filter padded samples from retrieval metrics

[ManasaDeshagouni]

Summary

This fixes a validation metrics inconsistency in _run_step.

When stride_drop_incomplete=False, zero-padded rows are filtered for flattened metrics, but retrieval metrics were still updated using unfiltered batch tensors. As a result, retrieval_top1 included padded boundary segments during default training runs.

What changed

• kept the existing flat filtering logic for flattened metrics
• added a batch-level mask for retrieval metrics
• filtered retrieval inputs before metric.update(...)

Impact

This does not affect the Pearson metric used for checkpointing and early stopping. It fixes the logged retrieval metric so it reflects only valid samples.

Note

Previous W&B retrieval metrics may have been slightly inflated because zero-padded boundary segments were included.

[meta-cla]

Hi @ManasaDeshagouni!

Thank you for your pull request and welcome to our community.

Action Required

In order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you.

Process

In order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA.

Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with CLA signed. The tagging process may take up to 1 hour after signing. Please give it that time before contacting us about it.

If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks!

[meta-cla]

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Meta Open Source project. Thanks!