Python users

This page has information on how to install for users who have some familiarity with Python. R users should see this page, which has more thorough instructions and introductions to Python concepts.

Installation

Users can use whatever package management utility they like, such as conda, venv, pixi, or uv. Both PyTorch and Pyseter can be installed with pip. We’ll assume you’re using conda.

Once you’ve installed conda, open the command line interface. Then you’ll need to do several things:

  • Create the conda environment you’ll be using
  • Active the environment and install pip
  • Install packages
    • Pytorch, which will depend on your operating system and GPU availability
    • gdown (for downloading the the dataset)
    • ipykernel (optional, useful if using Jupyter Lab or Jupyter Notebook)
    • pyseter

Below is an example of the commands necessary to install on a Windows (or Linux) machine equipped with an NVIDIA GPU.

conda create -n pyseter_env -y  # create new environment 
source activate pyseter_env     # activate it
conda install pip -y            # pip is necessary to install torch and pyseter
pip install torch torchvision --index-url https://download.pytorch.org/whl/cu128 
pip install pyseter gdown ipykernel==6.30.1 ipywidgets
python -m ipykernel install --user --name pyseter_env --display-name "Python (Pyseter)"

Verify the installation

Verify the Pyseter installation by running the following command.

import pyseter
pyseter.verify_pytorch()

If you have access to an NVIDIA GPU, you should see something like

✓ PyTorch 2.7.1+cu126 detected
✓ CUDA GPU available: NVIDIA A30 MIG 2g.12gb