Install PyTorch on Apple M1 (M1, Pro, Max) with GPU (Metal)

Note

• Uninstall Anaconda/Anaconda Navigator and any other conda-based installations.
• Anaconda and Miniforge cannot coexist.
• However, if you have previous installations of PyTorch with Miniforge, you can keep using it without uninstalling it.
• We will install the GPU version in a new conda environment anyway.
• The GPU version is still a nightly build, so expect breaking changes for the next couple of months.

Installing Miniforge

• Install miniforge from brew: brew install miniforge
• Create a conda environment: conda create -n pt122
• Activate the environment: conda activate pt122
• Install Python: conda install python

Installing PyTorch

• Run: pip install --pre torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/nightly/cpu to install PyTorch (Nightly) dependencies
• Execute: pip install pandas jupyterlab to install relevant dependencies to run sample code.
• Run: conda activate pt122

Verifying Installation

• Execute: jupyter-lab to open a Jupyter Notebook and run the following code:

import torch
torch.device("mps")
torch.__version__
torch.tensor([1,2,3], device="mps")

Credits

• Nachiket Sirsikar