Skip to content

Deepseek V3.2 implementation fixes#838

Merged
awni merged 2 commits into
ml-explore:mainfrom
sjug:deepseek_v32_fix
Feb 3, 2026
Merged

Deepseek V3.2 implementation fixes#838
awni merged 2 commits into
ml-explore:mainfrom
sjug:deepseek_v32_fix

Conversation

@sjug
Copy link
Copy Markdown
Contributor

@sjug sjug commented Feb 2, 2026

Fixes 837

  • Defensively don't pass tools if None
  • If tools is passed, pop it off kwargs and inject it into the system or developer message
  • Reshape mask to match scores

@awni awni force-pushed the deepseek_v32_fix branch from 0bdd873 to 5d8be24 Compare February 3, 2026 15:45
Copy link
Copy Markdown
Member

@awni awni left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@awni awni merged commit ab050d1 into ml-explore:main Feb 3, 2026
2 checks passed
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.

DeepSeek V3.2 implementation has multiple issues

2 participants