torchvision
opencv-python
diffusers==0.31.0
transformers==4.49.0
tokenizers>=0.20.3
accelerate>=1.1.1
peft
tqdm
datasets
imageio
imageio-ffmpeg
easydict
ftfy
omegaconf
einops
av==13.1.0
open_clip_torch
sentencepiece
regex
torchao
https://huggingface.co/datasets/multimodalart/zerogpu-blackwell-wheels/resolve/main/wheels/pt211-cu130-cp312/flash_attn-2.8.3-cp312-cp312-linux_x86_64.whl