[TRTLLM-7263][fix] Prevent recreation of cublas handles in lora_grouped_gemm every call#7053

amitz-nv merged 1 commit into

NVIDIA:release/1.0NVIDIA/TensorRT-LLM:release/1.0from

amitz-nv:trtllm-7263-prevent-recreation-of-cublas-handles-in-lora-grouped-gemmamitz-nv/TensorRT-LLM:trtllm-7263-prevent-recreation-of-cublas-handles-in-lora-grouped-gemmCopy head branch name to clipboard