Installation

Requirements

BaseAttentive supports Python 3.10 and newer.

The core package depends on:

  • keras>=3

  • numpy>=1.23

  • scikit-learn>=1.2

Runtime extras

Install the package together with the runtime backend you want to use:

pip install "base-attentive[tensorflow]"
pip install "base-attentive[jax]"
pip install "base-attentive[torch]"

If you are working from a source checkout:

git clone https://github.com/earthai-tech/base-attentive.git
cd base-attentive
pip install -e ".[dev,tensorflow]"

For local development only:

pip install -e ".[dev,tensorflow]"

If you use the repository Makefile, common setup commands are:

make install-dev
make install-tensorflow
make install-jax
make install-torch

Backend selection

The runtime backend must be selected before importing base_attentive or keras:

export KERAS_BACKEND=tensorflow   # default / stable
export KERAS_BACKEND=jax          # experimental
export KERAS_BACKEND=torch        # experimental

You can also set the backend programmatically before the first import:

import os
os.environ["KERAS_BACKEND"] = "tensorflow"
from base_attentive import BaseAttentive

The BASE_ATTENTIVE_BACKEND environment variable can also be used; it takes precedence over KERAS_BACKEND.

Verifying your install

After installation, validate the import path, version, and backend resolution:

import base_attentive
from base_attentive import get_backend, get_available_backends

print(base_attentive.__version__)   # e.g. "1.0.0"
print(get_backend().name)           # e.g. "tensorflow"
print(get_available_backends())     # ['tensorflow', ...]

Build the documentation locally

Install the docs dependencies and build the Sphinx site:

pip install -e ".[docs,tensorflow]"
python -m sphinx -b html docs docs/_build/html

The generated HTML lives in docs/_build/html.

Notes

  • TensorFlow is the current full runtime path used in examples and tests.

  • JAX and Torch are available through the backend abstraction layer, but the full BaseAttentive execution path is still experimental on those runtimes.

  • Python 3.9 is no longer supported as of v1.0.0; use Python 3.10 or newer.