{
  "repo_folder": "quality_base",
  "paper_setting": "QUALITY-BASE",
  "internal_run_name": "quality-base",
  "selection_rule": "The next 5B tokens down the DCLM fastText ranking after the shared anchor is removed from the pool, used verbatim.",
  "content": "raw",
  "composition": {
    "note": "This folder is the complete 10B-token mixture the models actually trained on: the shared 5B raw anchor plus this setting's 5B component, already merged and shuffled. The anchor is included in every setting folder.",
    "shuffle_seed": 42,
    "anchor_docs": 4120164,
    "anchor_tokens": 5000002332,
    "strategy_docs": 3133023,
    "strategy_tokens": 5000000805,
    "strategy_token_target": 5000000000
  },
  "rewriter_model": null,
  "prompt_style": null,
  "doc_count": 7253187,
  "token_count_llama2": 10000003137,
  "shard_count": 15,
  "columns": [
    "orig_doc_id",
    "doc_id",
    "text",
    "url",
    "metadata",
    "fasttext",
    "fineweb-edu",
    "modernbert",
    "topic",
    "fasttext-ranking-v2",
    "fineweb-edu-ranking-v2",
    "modernbert-ranking-v2",
    "tokens-llama2"
  ],
  "provenance": {
    "source_folder": "/scratch/bvandur1/zhuicon1/data_rewrite/pretrain/10B-base/shuffled",
    "tokenized_folder": "/scratch/bvandur1/zhuicon1/nanotron_tokenized/10B-base-shuf42/tokenized",
    "source_evidence": "datatrove executor.json (reader data_folder), confirmed by decoding 50 documents out of nanotron_tokenized/10B-base-shuf42/tokenized and matching them into this folder's 15 shuffled shards (50/50)",
    "trained_from_config": "config_quality-base_10B_1.5B.yaml (seed 42 launched as config_S0_1.5B_0613.yaml)",
    "config_dir": "/scratch/bvandur1/zhuicon1/projects/nanotron/examples",
    "seeds": [
      42,
      43,
      44
    ],
    "doc_content_digest_sha256": "772419b3b8cbf54d3e0eb8fded2b1f966126c7a988e0b70f0b32f7d30e1878fc",
    "doc_id_digest_sha256": "90252ee58ad4b298adadb7dfd0dfd9ba5fe01ed595fa62cd5c1a53947e4c2a22"
  }
}