{
  "source": "synthetic_empirical_support_power_decay_trigram",
  "source_revision": "03a31e6d03b5039ab3bff78c2519c90d5fa8bab99fce54f798294c1d217ddbd8",
  "source_report_sha256": "03a31e6d03b5039ab3bff78c2519c90d5fa8bab99fce54f798294c1d217ddbd8",
  "source_rows_sha256": "d12e5e751673515d4b3c925d6ca5df829e0293a460e51c598f2a6890455da8a7",
  "tokenizer_name": "eoinf/pile_tokenizer_4096",
  "tokenizer_revision": "155b268a49b69bac2543a3eb0acb635cef3dc4b0",
  "eos_token_id": 2,
  "seed": 40,
  "generation_batch_size": 512,
  "source_conditional_entropy_nats": 3.5714104546868244,
  "training_window_entropy_nats": 3.576508009854634,
  "evaluation_kind": "training_sample",
  "note": "Same 1025-to-257 source-row packing as historical empirical trigrams; synthetic source changes. Training-sample evaluation is not held out.",
  "tokens": 199999488,
  "cache_entries": 781248,
  "tokenizer_size": 4096,
  "packing": "Subwindows of independent source rows; 257 tokens and 256 scored targets per window",
  "row_length": 1025,
  "windows_per_source_row": 4,
  "context_length": 256
}