llama-cpp-python
# To install with ROCm support, use:
# CMAKE_ARGS="-DLLAMA_HIPBLAS=on" pip install llama-cpp-python
