gradio==6.3.0
huggingface_hub
hf_xet
torch
torchvision
transformers==4.57.1
bitsandbytes
git+https://github.com/vllm-project/llm-compressor.git
flash-attn @ https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.8cxx11abiTRUE-cp310-cp310-linux_x86_64.whl