--extra-index-url https://pypi.ngc.nvidia.com
nvidia-cuda-runtime
nvidia-cublas
llama-cpp-python @ https://github.com/OpenAccess-AI-Collective/ggml-webui/releases/download/v0.1.53/llama_cpp_python-gpu-0.1.53-cp38-cp38-linux_x86_64.whl
pyyaml
torch