Skip to main content

Installation and release tracks

The documentation targets the current RC/alpha release family, with exact pins. Use a fresh virtual environment so older consumer applications do not silently upgrade their dependency set.

python -m venv .venv
# macOS / Linux
source .venv/bin/activate
# Windows PowerShell: .venv\Scripts\Activate.ps1
python -m pip install --upgrade pip

Choose a backend​

python -m pip install --pre "omnibias-torch==0.5.0rc1"
# Or:
python -m pip install --pre "omnibias-jax==0.5.0rc1"
python -m pip install --pre "omnibias-keras==0.0.2a1"

For Keras, install and select TensorFlow, JAX, or PyTorch before importing Keras. See the Keras example. GPU backend installation follows the backend vendor's platform-specific instructions.

Python 3.11+ is recommended for new environments. Python 3.10 remains an upstream compatibility target but resolves older backend releases. The package's Requires-Python and dependency metadata are authoritative.

Add only what you need​

The package explorer records exact versions, maturity, and licenses. omnibias-core==0.5.0rc1 provides pure-Python derivative algebra. omnibias-fields==0.2.0rc1 adds field state. Most retained extension packages use 0.1.0a2.

python -m pip install --pre "omnibias-partition[torch]==0.1.0a2"

Partition is AGPL-or-commercial; the Torch backend is Apache-2.0. Review the package-specific licensing before choosing a dependency.

Existing stable users​

PyPI's default stable Torch version may differ from this documentation's release candidate. Explicit pins avoid that ambiguity. Existing published consumer users should keep the tested historical constraints until compatible successor releases are available.

From source​

git clone https://github.com/derivon-ai/omnibias.git
cd omnibias
git checkout 9a2d73df0573bc3a907351022669684fa3d41ddc
uv sync --all-packages --group docs

See the upstream release guide for prerelease resolution and distribution validation. Proceed to the quickstart.

Support the project​

If omnibias helps your work, consider starring the repository. Organizational sponsorship through Derivon supports maintenance, testing, and new scientific examples.