{
  "document_end_id": 8,
  "elapsed_seconds": 541.1,
  "format_version": 3,
  "held_out_percent": 1,
  "mix_policy": "token_balanced",
  "per_source_tokens_all_splits": {
    "natural_cosmopedia_v2": 168663969,
    "natural_dclm": 211970663,
    "natural_finemath_4plus": 53640449,
    "natural_fineweb_edu": 570339212,
    "natural_fineweb_hq": 109226534
  },
  "row_targets": {
    "broad_web": 0.2,
    "educational_web": 0.5,
    "high_quality_web": 0.1,
    "mathematics": 0.05,
    "textbook_exposition": 0.15
  },
  "row_tokens_all_splits": {
    "broad_web": 211970663,
    "educational_web": 570339212,
    "high_quality_web": 109226534,
    "mathematics": 53640449,
    "textbook_exposition": 168663969
  },
  "splits": {
    "held_out": {
      "document_ends_file": "held_out_document_ends.npy",
      "documents": 8586,
      "token_file": "held_out_tokens.int32",
      "tokens": 11432331
    },
    "train": {
      "document_ends_file": "train_document_ends.npy",
      "documents": 824997,
      "token_file": "train_tokens.int32",
      "tokens": 1091660174
    },
    "validation": {
      "document_ends_file": "validation_document_ends.npy",
      "documents": 8437,
      "token_file": "validation_tokens.int32",
      "tokens": 10748322
    }
  },
  "tokenizer_path": "D:\\llm\\frost\\artifacts\\runs\\e3_tokenizers\\tok_4k_digit",
  "validation_percent": 1
}