Disable thinking mode in vLLM instruct chat#457
Open
dmndxld wants to merge 1 commit into
Open
Conversation
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
Qwen3 enables thinking mode by default in its chat template, generating
<think>blocks before the actual response.This interferes with uncertainty estimation, as the generated text contains reasoning traces instead of direct
answers.
When using
WhiteboxModelvLLMwithinstruct=True, themodel.chat()call invokesapply_chat_templateinternally,which enables thinking mode by default for Qwen3.
Fix
Pass
chat_template_kwargs={"enable_thinking": False}tomodel.chat(), which propagates toapply_chat_templateand disables thinking mode:
This uses vLLM's native chat_template_kwargs parameter rather than manually calling apply_chat_template +
model.generate(), keeping the implementation consistent with vLLM's API.
Affected models
Models that enable thinking mode by default in their chat template, such as Qwen3.