# Flash Attention - CUDA 12, PyTorch 2.6, Python 3.10
flash-attn @ https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu12torch2.6cxx11abiFALSE-cp310-cp310-linux_x86_64.whl

# Core ML/AI Libraries
torch==2.6.0
torchvision
accelerate>=0.24.0

# Transformers - using version compatible with both sets of models
transformers==4.57.1
tokenizers>=0.20.3
transformers-stream-generator

# Hugging Face
huggingface_hub
hf_xet
spaces>=0.20.0

# Vision & Image Processing
qwen-vl-utils

# Web Interface
gradio==5.9.1
pydantic==2.10.6