CoolFace
Modelpublic

mlx-community/Fara1.5-27B-OptiQ-4bit

sourceHugging Facemitupdated 10d agoView on Hugging Face
2likes322downloads
4 commits on main
820a14110d ago

Update kv_config.json: re-measured with `optiq kv-cache` float32 sensitivity (the previous config was ranked on bfloat16 rounding noise). 8-bit layers [31, 47] -> [19, 27].

codelion
01991df12d ago

Add kv_config.json: per-layer KV cache precision from `optiq kv-cache` (target 4.5 bits). Use with `optiq serve --kv-config kv_config.json`.

codelion
1a13be02mo ago

Add files using upload-large-folder tool

codelion
3cd74362mo ago

initial commit

codelion