Bridges: using Dense-Armor with other libraries

Dense-Armor estimates one sample at a time; scikit-learn estimates in a fit over the whole table. A bridge lets each side use the other without rewriting an algorithm.

What a bridge is

A bridge does not replace an algorithm. It makes a Dense-Armor object usable where the other library expects its own interface, and the other way round, with a test that the numbers agree.

Available bridges

Library	Page	Extra
scikit-learn	sklearn	pip install dense-armor[sklearn]
PyTorch	torch	pip install dense-armor[torch]
TensorFlow	tensorflow	pip install dense-armor[tensorflow]
river	river	pip install dense-armor[river]
LLM agents	llm	none: embed is supplied by the caller
scikit-learn tools on the classes of ml/	ml/sklearn	pip install dense-armor[sklearn]

The imports of a bridge are optional: import dense_armor works without the other library installed, and the bridge module raises a ModuleNotFoundError that names the extra when the other library is missing.