[Feat] Support native Kimi-K2-Thinking native W4A16 quantized experts weights#4516

Merged

wangxiyuan merged 23 commits intovllm-project:mainfrom

zhoux77899:k2-thinking

Dec 10, 2025

Commits on Nov 28, 2025

feat(quantization): add `W4A16` quantization moe method for native Kimi-K2-Thinking quantized experts weights
zhoux77899
committed
fix(bug): fix missing keyword arguments
zhoux77899
committed

Commits on Dec 1, 2025

Commits on Dec 2, 2025

test(ci): add test case to workflows
zhoux77899
committed