Write high performance operators for LLMs with CUDA/OpenCL/Triton
CUDA kernels are implemented through pycuda, and Colab is recommended for trying：
| Name | Name | Last commit date | ||
|---|---|---|---|---|
Write high performance operators for LLMs with CUDA/OpenCL/Triton
CUDA kernels are implemented through pycuda, and Colab is recommended for trying：