feat: fused Metal q4 inference for MLX 4-bit models#82

Open

cjchanh wants to merge 3 commits intoevilsocket:mainfrom

cjchanh:q4-metal-patchset

Commits on Apr 13, 2026

Add fused Metal q4 path for MLX 4-bit models
cjchanh
committed

Commits on Apr 30, 2026

chore(cake): apply spec 151 portfolio gitignore sweep (q4-metal-patchset branch)
cjchanh
committed
fix(cake-q4): filter single-file safetensors push by layer (spec 199 row resolution)
cjchanh
committed