gradio
llama_cpp_python
torch
transformers
accelerate
safetensors