CoolFace
Modelpublic

mlx-community/humanizer-1B-OptiQ-4bit

sourceHugging Faceapache-2.0updated 11d agoView on Hugging Face
5likes258downloads
5 commits on main
8df1a5f11d 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 [4, 12, 13] -> [0, 1, 4].

codelion
59d062612d 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
6744a6012d ago

Add optiq tag

codelion
10c9e683mo ago

Re-publish under correct OptiQ spelling

codelion
2e0dd093mo ago

initial commit

codelion