Skip to content
TheChatArchiveThe Chat Archive
gpt-4.1-mini-2025-04-14· OpenAIAi2 WildChat-4.8M (AI2)

Share the code using which I can install the below dependencies through python embeded…

  1. User

    Share the code using which I can install the below dependencies through python embeded folder of comfyui.

    comfyui-frontend-package==1.16.8 comfyui-workflow-templates==0.1.1 torch torchsde torchvision torchaudio numpy>=1.25.0 einops transformers>=4.28.1 tokenizers>=0.13.3 sentencepiece safetensors>=0.4.2 aiohttp>=3.11.8 yarl>=1.18.0 pyyaml Pillow scipy tqdm psutil

    #non essential dependencies: kornia>=0.7.1 spandrel soundfile av

  2. Assistant

    To install the specified dependencies using Python embedded within the ComfyUI folder, you typically need to run pip commands using that embedded Python executable.

    Assuming your ComfyUI embedded Python folder is something like ComfyUI/python, and that the Python executable is within it (python.exe on Windows or python on Linux/macOS), you can run the following commands from a script or command line:

    # Navigate to your ComfyUI embedded Python folder
    cd path/to/ComfyUI/python
    
    # On Windows, use python.exe
    .\python.exe -m pip install --upgrade pip
    
    # Install the required dependencies
    .\python.exe -m pip install comfyui-frontend-package==1.16.8 comfyui-workflow-templates==0.1.1 torch torchsde torchvision torchaudio numpy>=1.25.0 einops transformers>=4.28.1 tokenizers>=0.13.3 sentencepiece safetensors>=0.4.2 aiohttp>=3.11.8 yarl>=1.18.0 pyyaml Pillow scipy tqdm psutil kornia>=0.7.1 spandrel soundfile av
    

    If you want a Python script you can run from within the embedded Python which installs all requirements, here is a script you can run:

    import subprocess
    import sys
    
    packages = [
        "comfyui-frontend-package==1.16.8",
        "comfyui-workflow-templates==0.1.1",
        "torch",
        "torchsde",
        "torchvision",
        "torchaudio",
        "numpy>=1.25.0",
        "einops",
        "transformers>=4.28.1",
        "tokenizers>=0.13.3",
        "sentencepiece",
        "safetensors>=0.4.2",
        "aiohttp>=3.11.8",
        "yarl>=1.18.0",
        "pyyaml",
        "Pillow",
        "scipy",
        "tqdm",
        "psutil",
        # non-essential dependencies:
        "kornia>=0.7.1",
        "spandrel",
        "soundfile",
        "av"
    ]
    
    def install(packages):
        subprocess.check_call([sys.executable, "-m", "pip", "install", "--upgrade", "pip"])
        for package in packages:
            subprocess.check_call([sys.executable, "-m", "pip", "install", package])
    
    if __name__ == "__main__":
        install(packages)
    

    How to use:

    1. Save this Python script (e.g., install_deps.py) inside the ComfyUI embedded python folder.
    2. Run it using the embedded Python executable:
      • On Windows:

        python.exe install_deps.py
        
      • On Linux/macOS:

        ./python install_deps.py
        

    This will upgrade pip first and then install all required dependencies.


    Notes:

    • Sometimes torch and related packages may require matching CUDA versions and special indexes. If you run into issues, you may want to install torch separately using the official instructions from pytorch.org.
    • If the embedded Python environment does not have internet access or pip installed, you need to set that up first.

    Let me know if you want me to provide a version with special handling for torch or a requirements.txt file!

  3. User

    share a py file based on the below dependencies too.

    comfyui-frontend-package==1.16.9 comfyui-workflow-templates==0.1.3 torch torchsde torchvision torchaudio numpy>=1.25.0 einops transformers>=4.28.1 tokenizers>=0.13.3 sentencepiece safetensors>=0.4.2 aiohttp>=3.11.8 yarl>=1.18.0 pyyaml Pillow scipy tqdm psutil

    #non essential dependencies: kornia>=0.7.1 spandrel soundfile av

  4. Assistant

    Certainly! Here's a Python script (install_comfyui_deps.py) that you can run within the ComfyUI embedded Python environment to install your specified dependencies, including the updated versions you listed:

    import subprocess
    import sys
    
    packages = [
        "comfyui-frontend-package==1.16.9",
        "comfyui-workflow-templates==0.1.3",
        "torch",
        "torchsde",
        "torchvision",
        "torchaudio",
        "numpy>=1.25.0",
        "einops",
        "transformers>=4.28.1",
        "tokenizers>=0.13.3",
        "sentencepiece",
        "safetensors>=0.4.2",
        "aiohttp>=3.11.8",
        "yarl>=1.18.0",
        "pyyaml",
        "Pillow",
        "scipy",
        "tqdm",
        "psutil",
        # Non-essential dependencies
        "kornia>=0.7.1",
        "spandrel",
        "soundfile",
        "av",
    ]
    
    def upgrade_pip():
        print("Upgrading pip...")
        subprocess.check_call([sys.executable, "-m", "pip", "install", "--upgrade", "pip"])
    
    def install_packages(packages):
        for pkg in packages:
            print(f"Installing {pkg}...")
            subprocess.check_call([sys.executable, "-m", "pip", "install", pkg])
    
    def main():
        try:
            upgrade_pip()
            install_packages(packages)
            print("All packages installed successfully.")
        except subprocess.CalledProcessError as e:
            print(f"Error installing packages: {e}")
            sys.exit(1)
    
    if __name__ == "__main__":
        main()
    

    Usage:

    1. Save this content into a file named install_comfyui_deps.py inside the ComfyUI embedded Python folder.
    2. Open a terminal/command prompt in that folder.
    3. Run the script using the embedded python executable:
    • On Windows:
    python.exe install_comfyui_deps.py
    
    • On Linux/macOS:
    ./python install_comfyui_deps.py
    

    Note on torch and related packages:

    • PyTorch (torch, torchvision, torchaudio) can have specific CUDA or CPU-only versions depending on your environment. The above script installs the default version, which usually matches your system.
    • If you require a specific CUDA version or special repository (like --extra-index-url for NVIDIA), you may need to adjust the script or install those packages manually following PyTorch's official guidance: https://pytorch.org/get-started/locally/

    If you'd like, I can help you adapt this script for specific CUDA versions!

  5. User

    thanks!

  6. Assistant

    You're welcome! If you need any more help, just let me know. Happy coding with ComfyUI! 😊🚀