Repository navigation
fix(benchmarks): stabilize restricted setup and output collection - #58
Merged
Merged
Conversation
minixalpha
changed the title
minixalpha
changed the title
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Restricted Harbor runs could time out while downloading dependencies, and a long Pi JSON event could abort output collection before official verification. Add verified PyPI and original uv/Python caches restricted to setup and verification, plus UTF-8 chunked output collection that preserves deadlines and cleanup. Snapshot cache inputs, verify removal before model work, and require phase evidence during integrity review. Original task images, package sources, tests, and limits remain unchanged.
The recorded API-only Pi runs passed WAL (7/7) and Pipeline (4/4), with reviewed effective reward 1.0 each and no observed integrity violation. Pipeline captured all 382 events, including a 296,980-byte final event. Preserve the local-search caveat, private credential-bearing logs, dependency drift, and every previous failure; do not pool these outcomes with historical open-network baselines.
Validation: 410 adapter tests passed; all three opt-in Docker checks passed separately. All three CI checks passed on this PR head. Three complete English/Chinese report pairs and the benchmark README are synchronized; 494 frozen evidence files reverified unchanged and public artifacts contain no exact credential matches.
Changelog: existing [Unreleased] Added and Fixed entries cover the managed caches and long-event collector. Development and research notes are unchanged.