Oct 16 16 days ago 07:49 Simplify README wording and put rule-based checks last Use kept instead of retained in README prose. Order per-stage breakdowns as build/setup, no-op, reference solution, then rule-based checks. Preserve task counts, validation findings and schema field names. Oct 15 17 days ago 22:10 Keep CalibForge pilot review results in GitHub 22:00 Summarize v1 and v2 task totals and source proportions 21:48 Use consistent short dataset names with v1 and v2 folders 15:17 Keep dataset naming guidance concise 15:14 Name dataset folders by filtering level and document the three-stage protocol 09:53 Simplify image download explanation 09:52 Download retained tasks and cached images together 09:44 Shorten dataset download instructions 09:42 Update README.md 09:29 Explain dataset filtering groups and image downloads Oct 7 25 days ago 08:53 Explain datasource downloads and automatic image caching Sep 29 20:04 Remove notes below the keep-or-archive rules 20:03 Add dataset card and CrossCodeEval source descriptions 
