head-quant docs: per-block-32 absmax ship shape (per-channel = beta delegate bug; naming note)

Upload README.md with huggingface_hub

card: gpu-pipelined int8lin bundle (iPhone 50.3-51.5 / Mac 204 tok/s) + run instructions

int8 fused-kernel monolith (42.5-45.4 tok/s) + q16 chunked-prefill companion (147 tok/s) — new release config

Card: category layout (best verified config per platform x compute-unit)

Model card