Repository navigation
chore(benchmarks): record flash 1017 dev34 baseline and recovery - #52
Merged
Merged
Conversation
minixalpha
changed the title
minixalpha
marked this pull request as ready for review
October 19, 2026 18:29
minixalpha
changed the title
Record the retained solution passing five tests with independent preparation and native verifier deadlines. Give the timing regression fixture more scheduling tolerance without changing benchmark time limits.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Record the Flash 1017 dev34 baseline on frozen agent/prompt
ce28b1e: 26 passes, five scored failures, and three unscored verifier errors. Add the same-agent recovery summary of 28/34 passes (82.35%), six failures, and no unscored tasks, using 31 original scored outputs and three regenerated outputs. Preserve original records and per-task provenance; this summary includes resampling rather than first-attempt pass@1.Add retained-workspace verification, reviewed dependency profiles, and independent preparation deadlines. The retained PyTorch solution passes all five tests after 1,127.589 seconds of preparation under a 3,600-second limit, followed by 49.934 seconds of official verification under the unchanged 900-second limit, with zero model calls. Collection-error recovery is opt-in; valid scores including zero stop retries, and original trial evidence is restored.
Summarize the results and six remaining failures in the Chinese 0.10.x development notes, regenerate the complete English version, and add the changelog entry. Preserve four complete bilingual report pairs and aligned Harbor READMEs; root READMEs and research notes are unchanged.
Validation: 209 local Harbor tests passed. Summary counts, result provenance, artifact hashes, translations, and links were checked. CI run 40202580370 passes Python 3.13, Python 3.14, and Harbor workflow contracts.