bug: fix qwen3 reranker vllm precision#4747
Merged
qinxuye merged 1 commit intoxorbitsai:mainfrom Mar 31, 2026
Merged
Conversation
|
@qinxuye Reference to this issue: QwenLM/Qwen3-Embedding#71, include template in VLLM is better than without the template, so I think we shouldn't remove the template... (maybe have a variable to control it) |
Contributor
|
@ZhikaiGuo960110 Can you look at this? |
Collaborator
Author
after testing, result relevance score is indeed a big difference, without template, the score is indeed close and not precise, we should not remove template. However, in the scenario of Arguana Rerank dataset, I use evalscope to evaluate Qwen3 0.6B reranker model(with template, score is 0.699)(without template, score is 0.736, in consistent with vllm serve behavior and precision score), so the better way to approach this issue is to have variable control template opening.
with template,
without template,
|
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
4 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
to fix this issue, and be consistent with vllm reranker dataset precision score, remove this special template for qwen3 reranker family.
fix #4744