fix: guard against division by zero in GPTRewardModel with empty batches#610
Open
themavik wants to merge 1 commit intoCarperAI:mainfrom
Open
fix: guard against division by zero in GPTRewardModel with empty batches#610themavik wants to merge 1 commit intoCarperAI:mainfrom
themavik wants to merge 1 commit intoCarperAI:mainfrom
Conversation
When the input batch has 0 or 1 samples, bs = input_ids.shape[0] // 2 evaluates to 0, causing ZeroDivisionError at loss = loss / bs and RuntimeError at torch.stack on empty lists. Add an early return when bs == 0, returning zero loss and empty score tensors on the correct device. Fixes CarperAI#609 Co-authored-by: Cursor <cursoragent@cursor.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
GPTRewardModel.forward()crashes with aZeroDivisionErrorwhen the input batch has 0 or 1 samples:Additionally,
torch.stack(chosen_end_scores)fails withRuntimeErroron the empty list whenbs == 0.Changes
Add an early return guard after the batch size calculation. When
bs == 0, the function returns:loss: zero tensor on the input devicechosen_end_scores: empty tensor on the input devicerejected_end_scores: empty tensor on the input deviceTest Plan
Fixes #609