{
  "dataset": "roneneldan/TinyStories",
  "revision": "f54c09fd23315a6f9c86f9dc80f725de7d8f9c64",
  "source_split": "train",
  "source_records_scanned": 100000,
  "counts": {
    "scanned": 100000,
    "train": 96704,
    "filtered": 1303,
    "validation": 498,
    "test": 496,
    "duplicates": 999
  },
  "corpus_tokens": {
    "train": 20097225,
    "validation": 105076,
    "test": 102851
  },
  "split_policy": "normalized-text SHA256 modulo 1000: test 0-4, validation 5-9, train 10-999",
  "filter_version": 1,
  "source_checkpoint_sha256": "e8ce8d530bb679e9907b274811888cc15949606666f116ebf17cc84f4fc84386",
  "data_manifest_sha256": "8a7366d25a937781de35b19d0a9ab4b63981a8e401102b414ce40a71665a4ef5",
  "checkpoint_step": 23000,
  "stage_tokens": 94208000,
  "previous_tokens": 54702080,
  "training_tokens": 148910080,
  "sampled_validation_loss": 1.7479398322105408,
  "export_dtype": "float32",
  "optimizer_state_included": false
}