Skip to content

fix(parallel): shard SwiGLU weights in six model families - #1237

Open
chaofengw-nv wants to merge 6 commits into
NVIDIA:mainfrom
chaofengw-nv:fix/nightly-swiglu-shards
Open

fix(parallel): shard SwiGLU weights in six model families#1237
chaofengw-nv wants to merge 6 commits into
NVIDIA:mainfrom
chaofengw-nv:fix/nightly-swiglu-shards