# cd /nfs/ywang29/video_reward/DanceGRPO

# source activate 
# conda create -n videoalign python==3.10 -y
# conda activate videoalign 
# ./env_setup.sh fastvideo


# pip3 install moviepy
# pip3 install protobuf==3.20.0
# # mkdir videos
# pip3 install huggingface_hub==0.24.0 
# pip3 install tf-keras
# pip3 install trl
# # pip3 install transformers==4.51.0

# pip uninstall torchvision torch torchaudio -y 
# # pip install torchvision==0.20.1
# pip install torchvision==0.22.1

# pip install pydantic -U
# pip install qwen_vl_utils
pip install librosa
pip install wandb
pip install trl==0.8.6
# pip install flash-attn==2.5.8 --no-build-isolation
cd /nfs/ywang29/Reward_finetuning/VideoX-Fun

pip install transformers==4.45.2
# cp /nfs/ywang29/Reward_finetuning/VideoX-Fun/scripts/wan2.1/image_processing_qwen2_vl.py /usr/local/lib/python3.12/site-packages/transformers/models/qwen2_vl/image_processing_qwen2_vl.py
cp /nfs/ywang29/Reward_finetuning/VideoX-Fun/scripts/wan2.1/image_processing_qwen2_vl.py /opt/conda/lib/python3.10/site-packages/transformers/models/qwen2_vl/image_processing_qwen2_vl.py