gradio
transformers
torch
huggingface_hub
accelerate>=0.26.0'
llama-index
llama-index-embeddings-huggingface
peft
auto-gptq
optimum
bitsandbytes