[None][fix] Allow fail-early when reuse block and legacy mamba cache - #15540

[None][fix] Allow fail-early when reuse block and legacy mamba cache #15540

Wanli-Jiang wants to merge 1 commit into

NVIDIA:mainNVIDIA/TensorRT-LLM:mainfrom

Wanli-Jiang:user/williamj/fail-fast-for-wrong-combWanli-Jiang/TensorRT-LLM:user/williamj/fail-fast-for-wrong-combCopy head branch name to clipboard