Skip to content

Conversation

@relh
Copy link
Contributor

@relh relh commented Dec 21, 2025

Summary

  • Evaluator submission zip now built from checkpoint bundle paths.
  • Avoid absolute data_path leaks in submission.zip.
  • Update cogames run_evaluation to use policy_spec + safetensors.

Changes

  • Evaluator reads policy_spec/data from bundle dir (local or S3) and rewrites absolute data_path if present.
  • run_evaluation uses policy_spec_from_uri + safetensors.

Tests

  • not run (stack recut)

Copy link
Contributor Author

relh commented Dec 21, 2025

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

This stack of pull requests is managed by Graphite. Learn more about stacking.

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