Skip to content

Complete public fixtures and four missing answer contracts - #11

Merged
ma7dev merged 1 commit into
mainfrom
codex/fixture-evaluation
Sep 27, 2026
Merged

ma7dev merged 1 commit into
mainfrom
codex/fixture-evaluation

Conversation

@ma7dev

@ma7dev ma7dev commented Sep 27, 2026

Copy link
Copy Markdown
Member

Changes

  • Publish/reproduce version 2026.09.27.1 with all 6,089 cases and 11 public fixture files.
  • Document four source-based Absher annotations: two answer-key repairs and two ambiguous-question corrections using the recorded meanings.
  • Preserve original prompts, blank upstream answer cells, pinned CSV hashes/row indexes, and annotation rationale in the public ledger.
  • Update dataset card, README and reconstruction CI. Historical reconstruction remains unchanged.

Verification

29 tests; Ruff; clean deterministic rebuild; exactly four changed rows and two changed prompts; published file hashes independently downloaded and verified.

@ma7dev
ma7dev merged commit dc4a452 into main Sep 27, 2026
3 checks passed
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.

1 participant