Skip to content

Conversation

HuanzhiMao
Copy link
Owner

No description provided.

HuanzhiMao and others added 23 commits August 1, 2025 01:02
…il#1126)

1. Reasoning content was not being added in the inference data for both
QwenHandler and QwenFCHandler.
2. The _format_prompt function in QwenFCHandler was not checking for
tool_calls in assistant messages. All those messages where the assistant
requested a tool call were converted to empty messages.

This PR fixes both the above points.

This also means that all those scores for qwen3-prompt and qwen3-FC that
were calculated using local_inference, would need to be calculated
again.

Co-authored-by: Huanzhi Mao <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants