Installation¶

Install with pip or uv¶

For users working with the pip package manager, HyperTorch can be installed from PyPI.

pip install hypertorch
# if you want to install optional dependencies for tensorboard support:
pip install "hypertorch[tensorboard]"

Additionally, it is recommended to install and run HyperTorch from a virtual environment, for example, using the Python standard library’s venv. Internally, we use uv as a build and development tool, which also provides a convenient way to manage virtual environments and dependencies. After installing uv, you can create a project environment and add HyperTorch with:

uv init

source test-env/bin/activate || source test-env/Scripts/activate

# install desired torch
uv pip install torch --index-url https://download.pytorch.org/whl/cpu
# install desired pyg-lib 
uv pip install pyg-lib --find-links https://data.pyg.org/whl/torch-2.12.0+cpu.html


uv add hypertorch # or uv pip install hypertorch

# For optional dependencies
uv add "hypertorch[tensorboard]"

Python version support¶

See Python support policy in Policies.

Install from source¶

Use the development installation for contributing or if you want to use the latest features that haven't been released yet. See the Development guide for instructions on setting up a development environment.

Required dependencies¶

HyperTorch declares compatibility ranges for direct dependencies in pyproject.toml.

Optional dependencies¶

Development dependencies¶