fix: prevent division by zero in sampling when temperature is 0.0 (#562)#585
Open
irhyl wants to merge 1 commit intogoogle-deepmind:mainfrom
Open
fix: prevent division by zero in sampling when temperature is 0.0 (#562)#585irhyl wants to merge 1 commit intogoogle-deepmind:mainfrom
irhyl wants to merge 1 commit intogoogle-deepmind:mainfrom
Conversation
…ogle-deepmind#562) Add guard clause in RandomSampling, TopkSampling, and TopPSampling to fall back to greedy (argmax) when temperature < 1e-6, preventing inf/NaN from division by zero. Add tests to verify zero-temperature behavior matches Greedy sampling.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 #562 — Setting
temperature=0.0inRandomSampling,TopkSampling, orTopPSamplingcauses division by zero, producing inf/NaN logits and incorrecttoken predictions.
Changes
gemma/gm/text/_sampling.py— Add a guard clause at the top ofget_next_tokens()in all three temperature-dependent sampling classes: