How to install Py-Feat¶

Py-Feat requires Python 3.11+ (3.11, 3.12, and 3.13 are tested).

Basic setup¶

We recommend uv — a fast, modern Python package manager. Create an environment and install the latest stable release from PyPI:

uv venv --python 3.13        # creates .venv on Python 3.13 (3.11+ supported)
uv pip install py-feat

Then either activate the environment (source .venv/bin/activate) or run commands through it with uv run, e.g. uv run python my_script.py.

To get the latest development version instead, install directly from GitHub:

uv pip install "git+https://github.com/cosanlab/py-feat.git"

Note

Prefer plain pip? Every command above works by dropping uv: pip install py-feat or pip install "git+https://github.com/cosanlab/py-feat.git".

Note

Py-Feat supports CPU and NVIDIA (CUDA) GPUs, and — since v0.7 — Apple Silicon GPUs via Metal (MPS). Pass device='auto' (or device='mps' / device='cuda') when constructing a Detectorv1 to use the GPU; the default is cpu.

Running tutorials online (molab)¶

Every tutorial page has an Open in molab button at the top that opens the notebook in molab, marimo's free hosted runtime — run the tutorials in your browser with no local install required.

Development setup¶

To contribute to py-feat, install an editable copy so your source changes are picked up immediately:

git clone https://github.com/cosanlab/py-feat.git
cd py-feat
uv venv
uv pip install -e .
uv pip install -r requirements-dev.txt

Run the test suite with uv run pytest (use -m "not network" to skip tests that download models from the HuggingFace Hub).