Fix folding transposed GPT-OSS and Llama 4 expert weights - #2367

Open

MrCapricornLiu wants to merge 6 commits into

NVIDIA:mainfrom

MrCapricornLiu:MrCapricornLiu/fix/fold-transposed-expert-weights

Commits on Sep 10, 2026

Fix folding transposed GPT-OSS and Llama 4 expert weights
MrCapricornLiu
committed
c15ec0d
fix: materialize offloaded weights while folding quantizers
MrCapricornLiu
committed
4b96b51

Commits on Sep 11, 2026

test: explain optional Accelerate import
MrCapricornLiu
committed
424c3e2

Commits on Sep 13, 2026

Merge upstream main to resolve changelog conflict
MrCapricornLiu
committed
f624cff

Commits on Sep 14, 2026

chore: merge latest main
MrCapricornLiu
committed
932d5a4

Commits on Sep 17, 2026

chore: merge main and resolve changelog conflicts
MrCapricornLiu
committed
2f9d82d