Register final five-region RAG checkpoint for development VEP - #565
Register final five-region RAG checkpoint for development VEP#565gonzalobenegas wants to merge 4 commits into
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
Claude encountered an error after 0s —— View job I'll analyze this and get back to you. |
|
Claude encountered an error after 0s —— View job I'll analyze this and get back to you. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 51fd75ad43
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| return overlay | ||
|
|
||
|
|
||
| def main() -> None: |
There was a problem hiding this comment.
This function is now a maintained CLI used by CI and documented for users, but the owning project's [project.scripts] still registers only evals-gpu-runtime-check. Consequently, installed environments expose no stable command for this tool and callers must depend on its internal module path, contrary to the package-entry-point requirement in .agents/skills/develop-snakemake-pipelines/SKILL.md; add an entry point and use it in the workflow and README.
Useful? React with 👍 / 👎.
1 participant
Register the final 100,000-update checkpoint from #550 for combined RAG scoring and frozen probes on the Mendelian, Complex Traits, and SGE development cohorts.
The entry pins the europe-west4 version-9 GCS export path, the existing combined harness and its SHA-256, 255-base windows, 10,240-token documents, and batch size 2.
Biological evaluation waits for the final export and its numerical parity check.
Depends on #554 for the combined backend; evaluation also requires #559 and the strict-fp32 controls in #562.
The registration exposes an existing CI limitation: explicit S3 harness inputs still contact AWS when the default storage provider is disabled.
The CI dry-run now substitutes temporary local harness paths while preserving the full model registry and all other configuration; production storage and checksum verification are unchanged.
All CI checks pass, including the credential-free full pipeline dry-run and all five project test jobs.
Independent review found no blocking issues.
Part of #550.