Repository navigation
Conversation
…the latest run - deploy_hf.py deletes Hub files the allowlist no longer publishes (upload_folder only ever added). 161 are there now, all from the old layout: 108 tasks/harbor/ grader files including reference solutions and rubrics, 41 third-party author scripts under ref/ (not redistributable), paper/, ingestion/, old eval/ files. --dry-run lists them. - The dataset card's results table is generated at deploy time from the latest results/runs/ folder, with the same code as the leaderboard (four models, three refusal rules, easy/hard, cost). The card's old 2026-10-04 numbers, the old grading formula and its Apache-2.0 licence line (the metadata and repo say MIT) are gone; it now says what is and is not in the dataset. - tests/test_deploy.py: the Hub mirrors the allowlist, no task's tests/, solution/ or environment/ is staged, and the card carries the latest results and one licence. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…SIONAL.json Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…t scored Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This was referenced Oct 13, 2026
Merged
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Yes, we deploy to Hugging Face on every push to
main(deploy-hfin CI, after tests and image builds; last deploy today 16:18). Two problems with it:tasks/harbor/…/tests/files including reference solutions and rubrics, 41 third-party author scripts underref/(licences say not redistributable), pluspaper/,ingestion/and oldeval/files — 161 in all.This PR
deploy_hf.pynow mirrors the allowlist: after uploading, it deletes Hub files the deploy no longer publishes (.gitattributeskept).python scripts/deploy_hf.py --dry-runlists both sides: 259 published, 161 deleted.results/runs/folder with the leaderboard's own code, so it can't drift. The card now states what is and isn't in the dataset, the current grading, and one licence (MIT).tests/,solution/orenvironment/is staged; card has the latest results and one licence.🤖 Generated with Claude Code