Layer Wise Relevance Propagation Python Packages

wax-ot

WaX: explainable Wasserstein distances. Attributes a Wasserstein distance to instances, features and subspaces (Naumann, Kauffmann and Montavon, TPAMI 2026).

lrp-pf-auc

Explain Neural Networks using Layer-Wise Relevance Propagation and evaluate the explanations using Pixel-Flipping and Area Under the Curve.

layer-wise-integrated-gradients

Layer-wise Integrated Gradients (LIG): set-to-set IG at Transformer ATT/MLP module boundaries for within-layer token flow analysis. PyPI package & interactive demo. arXiv:2606.21564