huggingface_hub==0.25.2
transformers
accelerate>=0.26.0
llama-cpp-python