Summarize dataset purpose, size, sources, schema, loading and limitations. Group the 26 SFT source lineages into six dataset families, retaining detailed provenance in the manifest. Preserve YAML metadata and all data/audit files; update only README and its manifest hash/size.
Remove 271 reviewed benchmark overlaps and answer-conditioned derivatives
Remove the same 271 stable IDs from full-answer and short-answer SFT datasets: 20 same mathematical tasks and 251 derivatives exposing pinned benchmark content and forward answers. Preserve all retained values, steps, row order and schema. Keep 8 modified variants pending policy review and 3 different mathematical tasks. Update card, manifest, provenance, and validation evidence.
Document source revision, answer replacement, generation provenance, and licensing
