Uddeshsingh/q4k fused kernels#20231

Merged

metascroy merged 10 commits into

pytorch:mainfrom

uddeshsingh:uddeshsingh/q4k-fused-kernels

Jun 17, 2026

Commits on Jun 12, 2026

Add fused Q4_K Metal kernels for GGUF lowering (#20172)
uddeshsingh
committed
Guard Q4_K GGUF lowering behind ET_MLX_EMIT_DIRECT_GGUF
uddeshsingh
committed
Merge branch 'pytorch:main' into uddeshsingh/q4k-fused-kernels
uddeshsingh
authored
Merge branch 'pytorch:main' into uddeshsingh/q4k-fused-kernels
uddeshsingh
authored
Extract emit_if_else/emit_sub_int/emit_ceil_div helpers, fix output dtype
uddeshsingh
committed
Move Q4_K env-var dispatch into emit_linear/emit_embedding so patterns stays unchanged.
uddeshsingh
committed
Move Q4_K embedding env-var dispatch fully into emit_embedding.
uddeshsingh
committed

Commits on Jun 13, 2026

Simplify GGUF linear emit via emit_if_else constant folding
uddeshsingh
committed

Commits on Jun 16, 2026

up
metascroy
committed

Commits on Jun 17, 2026

up
metascroy
committed