data/atom_corpus.json

{
  "tokenizer": "atom-fractus-v1",
  "vocab_size": 266,
  "max_span_bytes": 32,
  "pack_mode": "linguistic",
  "docs": 2,
  "tokens": 1059743,
  "bin": "atom_corpus.i16",
  "dtype": "int16",
  "note": "fresh Atom id stream. Do not mix with GPT-2 50257 shards."
}