1BTF
1-Byte Text Format
1BTF is a byte-oriented text representation layer.
Core
- One 1BTF unit = exactly 1 byte.
- Raw UTF-8 byte matching.
- No grapheme segmentation.
- No tokenizer.
- No final token.
- English letters are canonical.
- English words are not stored as 1BTF units.
- Duplicate English representations are not stored in other domains.
- Global IDs belong to the master registry.
- Runtime representation uses 1-byte local IDs.
- Original UTF-8 unit information is stored in the registry.
Package
1btf_master.binโ single master registrytables/โ individual 1BTF sub-tablesmanifest.jsonโ package metadata and checksums
Runtime
Raw UTF-8 โ exact 1BTF unit โ 1-byte representation
1BTF is not a tokenizer and is not a final token format.
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐ Ask for provider support