To extract on another machine:
  # download all shards
  huggingface-cli download neversa/markers --repo-type dataset --local-dir ./llava178k_merged

  解压文件

  
  # extract all
  bash extract.sh

  # edit环境

  ## 环境配置
  ```
    conda create  -n visual python=3.10 -y
    conda activate visual
    pip install torch==2.5.1 torchvision==0.20.1 torchaudio==2.5.1 --index-url https://download.pytorch.org/whl/cu121
    
    pip install 'volcengine-python-sdk[ark]'
    
    pip install transformers
    pip install diffusers
    pip install qwen_vl_utils
    pip install tqdm
    pip install accelerate
    sudo apt install -y ffmpeg
    pip install opencv-python
    pip install vllm==0.17.1
    pip install aiohttp
  ```
  ## 执行命令
  `bash edit_sample_qwen_vllm_detect.sh`
  (sh文件中 --detect_json_path 为detect_results_all.json的路径，--model_path qwen image edit的路径, --data_root 存储该数据文件夹的路径 )