decoderstack-d24 / tokenizer

Commit History

  1. tokenizer: keep the pre-fix token_bytes so published bpb numbers stay reproducible

    Verified signature
    4aad0ae
  2. tokenizer: regenerate token_bytes.pt with upstream 2ce972a raw-bytes fix (191/32768 ids corrected)

    Verified signature
    47b3fec
  3. Add model card, meta.json, logs, tokenizer, training source, and nanochat converter

    Verified signature
    11f07f9