Allow zero loss_mask for remove_sample'd rollouts#60
Draft
nightlessbaron wants to merge 1 commit into
Draft
Conversation
… samples Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
nightlessbaron
force-pushed
the
feat/validator-allow-removed-zero-mask
branch
from
95526d0 to
b98a2bf
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
validate_rollout_for_grpo_training_steprejects any sample with an all-zeroloss_maskas a hard error. But rollout sample filters (mask_truncated/mask_token_truncated) legitimately zero a sample'sloss_maskviaremove_sampleto drop it from the gradient while keeping its reward in the group baseline — and the token-mean reducer clamps the denominator, so an all-zero mask is safe.Downgrade that case to a warning so masking doesn't crash
train_actor.