YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
win11 Python3.11.0 CUDA12.4 PyTorch 2.5.1
Introduction to Installing Custom PyTorch Wheels for ComfyUI This guide outlines the process of installing pre-built PyTorch wheels for use within a ComfyUI virtual environment. These wheels are specifically compiled for Python 3.11, CUDA 12.4, and are designed to enhance performance and compatibility.
Installation Steps:
Activate your ComfyUI virtual environment. This ensures that the custom PyTorch installation is isolated to your ComfyUI setup and avoids conflicts with other Python environments.
Bash E:\ComfyUI>venv\Scripts\activate
Install the custom PyTorch wheel. Assuming you have downloaded the torch-2.5.1+cu124-cp311-cp311-win_amd64.whl file to the E: drive, use the following pip command:
Bash pip install "E:\torch-2.5.1+cu124-cp311-cp311-win_amd64.whl"
This command instructs pip to install the specified wheel file. The quotation marks are used to handle potential spaces in the file path. Install the corresponding Torchvision and Torchaudio extensions. To ensure compatibility with the installed PyTorch version, install the matching Torchvision and Torchaudio extensions using the following command:
Bash pip install torchvision==0.20.1+cu124 torchaudio==2.5.1+cu124 --index-url https://download.pytorch.org/whl/cu124
This command installs specific versions of Torchvision and Torchaudio that are compatible with CUDA 12.4. The --index-url flag specifies the PyTorch wheel index, ensuring that the correct CUDA-enabled versions are downloaded. Key Considerations: Virtual Environment: It is crucial to perform these installations within your ComfyUI virtual environment to maintain a clean and organized setup. File Path: Ensure that the file path to your downloaded PyTorch wheel is correct. CUDA Compatibility: These wheels are specifically built for CUDA 12.4. If you have a different CUDA version, you will need to obtain or build compatible wheels. Version Matching: Always ensure that the versions of PyTorch, Torchvision, and Torchaudio are compatible to avoid potential errors. By following these steps, you can successfully install custom PyTorch wheels and their extensions within your ComfyUI environment, enabling optimized performance for your workflows.