Skip to content

Installation

Alice is distributed on PyPI under the package name alice-net. The import name is alice.

Requirements

  • Python ≥ 3.11
  • PyTorch ≥ 2.5
  • Nicole ≥ 0.3.6 (installed automatically)
  • NumPy ≥ 2.0 (installed automatically)

Install from PyPI

pip install alice-net

uv add alice-net

conda create -n alice python=3.11
conda activate alice
pip install alice-net

Verify the Installation

import alice
print(alice.__version__)

You should see 0.1.0 (or whichever version was installed).

Optional Extras

Development and Testing

pip install "alice-net[test]"

Installs pytest, hypothesis, and pytest-benchmark for running the test suite.

Linting and Type Checking

pip install "alice-net[lint]"

Installs ruff and mypy.

Documentation

pip install "alice-net[docs]"

Installs MkDocs, the Material theme, mkdocstrings, and all plugins needed to build and serve this documentation locally:

mkdocs serve

GPU Support

Alice inherits PyTorch's device support from Nicole. To run on CUDA:

import torch
import alice

# Check that a CUDA device is available
print(torch.cuda.is_available())

Pass device="cuda" when deserializing an MPS or Summary, or move tensors to GPU using standard Nicole/PyTorch .to(device) calls on individual site tensors.

Development Install

To contribute to Alice or run the latest unreleased code, clone the repository and install in editable mode:

git clone https://github.com/Ideogenesis-AI/Alice.git
cd Alice
pip install -e ".[test,lint,docs]"

See the Contributing guide for the full development workflow.