[distillation] - fix: add unit tests for distillation losses and training - #90
Merged
Conversation
- Implement tests to verify the correctness of KL divergence loss calculations between student and teacher models. - Ensure that the seq2seq distillation process correctly handles masked tokens and differentiable zero loss scenarios.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Fixes distillation KL loss so student logits are compared against teacher logits on the right axis.
Seq2seq training now ignores masked targets, keeps all-ignored batches safe for backpropagation, and runs teacher
inference without gradients.
Adds unit coverage plus an offline two-step
Trainer.fitsmoke test for seq2seq distillation.Testing
pytest -q tests/test_distillation_losses.py tests/test_seq2seq_distillation_training.py