restricting sampling in vllm with qwen models to avoid OOVs#1237
Merged
Conversation
facebook-github-bot
added
the
CLA Signed
swarnaHub
approved these changes
Jul 29, 2025
jacklanchantin
pushed a commit
that referenced
this pull request
Aug 22, 2025
Co-authored-by: Ilia Kulikov <kulikov@meta.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
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does this PR do? Please describe:
qwen model has bigger projection layer than the actual tokenizer vocabulary. Because of this it can sample non-existing tokens w.r.t. tokenizer and give vllm errors in RL setup
here we patch that by manually assigning allowed tokens id ONLY for policy actor (that samples the token ids for training) if we use qwen model for policy model.