Quazim0t0 / Byrne-TriAtn-86M

Text Generation97M181Updated Jul 8, 2026

Downloads

18

Likes

1

Last updated

Jul 8, 2026

License

apache-2.0

A TriAttention-enabled variant of Byrne-86M. Same weights, same chat behavior — plus a training-free KV-cache compression path (TriAttention) that bounds attention memory at long context. This repo does not replace Byrne-86M; it is a separate, drop-in model.

Tasks & library

  • text generation
  • transformers

Tags

  • transformers
  • spike_whale
  • feature-extraction
  • text-generation
  • small-models
  • chat

Technical specifications

Parameters97M
Librarytransformers
Languagesen
Licenseapache-2.0