Skip to content

Fix Jina embedding model loading on current dependencies - #18

Merged
M9nx merged 4 commits into
mainfrom
issue-17-jina-attention-fix
Sep 6, 2026
Merged

M9nx merged 4 commits into
mainfrom
issue-17-jina-attention-fix

Conversation

@M9nx

@M9nx M9nx commented Sep 6, 2026

Copy link
Copy Markdown
Owner

Summary

  • use supported eager attention for the Jina code embedding model
  • enable Jina's remote Transformers implementation
  • constrain sentence-transformers and Transformers to compatible major versions
  • support current and legacy MCP server handler APIs
  • ignore generated Rust and local CodexA artifacts

Validation

  • 2670 passed, 7 warnings

Closes #17

M9nx added 4 commits September 7, 2026 00:37
Keep sentence-transformers and Transformers on versions compatible with the Jina model's legacy configuration and remote implementation.
Use eager attention and enable the model's remote implementation for the Jina code embedding model. Add a mocked regression test for the constructor options.
Register MCP handlers through the current Server constructor while retaining compatibility with the legacy decorator API.
Ignore Rust build output and generated local .codexaignore metadata.
@M9nx M9nx self-assigned this Sep 6, 2026
@M9nx M9nx added the bug Something isn't working label Sep 6, 2026
@M9nx
M9nx merged commit 6a84d67 into main Sep 6, 2026
7 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

bug Something isn't working

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Bug] Crash during 'codexa index' due to unsupported attn_implementation="torch"

1 participant