Skip to content

Conversation

giladgd
Copy link
Member

@giladgd giladgd commented Jun 6, 2025

Description of change

  • feat(JSON Schema Grammar): $defs and $ref support with full inferred types
  • feat(inspect gguf command): format and print the Jinja chat template with --key .chatTemplate
  • fix(JinjaTemplateChatWrapper): first function call prefix detection
  • fix(QwenChatWrapper): improve Qwen chat template detection
  • fix: apply maxTokens on function calling parameters
  • fix: adjust default prompt completion length based on SWA size when relevant
  • fix: improve thought segmentation syntax extraction
  • fix: adapt to llama.cpp changes
  • test: first function call prefix detected and used
  • test: $defs and $ref usage

Fixes #471

Pull-Request Checklist

  • Code is up-to-date with the master branch
  • npm run format to apply eslint formatting
  • npm run test passes with this change
  • This pull request links relevant issues as Fixes #0000
  • There are new or updated unit tests validating the change
  • Documentation has been updated to reflect this change
  • The new commits and pull request title follow conventions explained in pull request guidelines (PRs that do not follow this convention will not be merged)

@giladgd giladgd requested a review from ido-pluto June 6, 2025 22:28
@giladgd giladgd self-assigned this Jun 6, 2025
@giladgd giladgd changed the title fix(JinjaTemplateChatWrapper): first function call prefix detection feat(JSON Schema Grammar): $defs and $ref support with full inferred types Jun 8, 2025
@giladgd giladgd merged commit 9cdbce9 into master Jun 10, 2025
20 checks passed
@giladgd giladgd deleted the gilad/functionCallingFixes branch June 10, 2025 21:15
Copy link

🎉 This PR is included in version 3.10.0 🎉

The release is available on:

Your semantic-release bot 📦🚀

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

bug: tool call sometimes not parsed correctly?
1 participant