[None][feat] MTP one-model advanced_sampling_mode: skip redundant top-k / top-p filter kernels with additional config enum#16561

jhaotingc wants to merge 1 commit into

NVIDIA:mainNVIDIA/TensorRT-LLM:mainfrom

jhaotingc:qwen36-temponlyjhaotingc/TensorRT-LLM:qwen36-temponlyCopy head branch name to clipboard