Build with us

We welcome contributions to models, evaluation tasks, data, documentation, and agent applications.

Choose the right repository

• Valen: model implementation, inference, training, visual decisions, and related documentation.
• JevalKit: model adapters, benchmark integrations, scoring, and evaluation results.

Report an issue

Include the repository revision, environment, exact command, and a minimal reproducible example. For evaluation results, include the model checkpoint, dataset revision or hashes, and scoring protocol.

Open a Valen issue · Open a JevalKit issue

Improve evaluation coverage

Suggest a benchmark or a released decision model by linking its source and describing its input/output interface. Read the model integration guide and benchmark protocols.

Improve documentation

Report unclear instructions and broken examples in the relevant project's repository. Technical reference pages on this site are synchronized snapshots of those source documents.