{
  "schema": "mt11-olmo3-think-omr-public-data-provenance/1",
  "source": {
    "repo_id": "nvidia/OpenMathReasoning",
    "revision": "d3d08664755704f422af97d43a7ff0ded4bd95df",
    "revision_kind": "huggingface_git_commit",
    "license": "cc-by-4.0",
    "family": "OPENMATHREASONING",
    "slice_id": "OMR_DSR1_COT",
    "selection": "COT only; DeepSeek-R1 generated solutions; no TIR/tool rows"
  },
  "row_counts": {
    "accepted_after_source_audit": 4826,
    "selected_for_training": 4096
  },
  "selected_public_manifest": {
    "sha256": "1ba8c53a70914097863de4dfb19b65246d6a4c9acc35d090a8f00f91d85f19b1",
    "rows": 4096,
    "included_in_repository": false,
    "reason": "Raw rows and the complete training set are intentionally excluded; only hashes and aggregate profiles are public."
  },
  "ordered_ids": {
    "included_in_repository": false
  },
  "update_plan": {
    "sha256": "8b5e60c9e384e81c57433405bdcb965c00bbd806307fdf5f73e8112cd6383a66",
    "included_in_repository": false
  },
  "data_profile": {
    "source_file_sha256": "4569c25e53fe60cd8c8bd3f3df0de8fb1b473abad6dfc0edce75e5630bd2afb7",
    "public_summary_file": "DATA_PROFILE.json"
  },
  "sensitive_or_restricted_data_included": false
}