NVIDIA / TensorRT-LLM

TensorRT-LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and build TensorRT engines that contain state-of-the-art optimizations to perform inference efficiently on NVIDIA GPUs. TensorRT-LLM also contains components to create Python and C++ runtimes that execute those TensorRT engines.
https://nvidia.github.io/TensorRT-LLM
Apache License 2.0
8.71k stars 996 forks source link

Is there any restriction on the weight dimension value? #2433

Closed zhink closed 1 week ago

zhink commented 1 week ago

In cpp/tensorrt_llm/kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu cudaCoreGemm kernel accuracy anomalies will occur, When the weight dimension n or k is not a power of 2.