Your model made a decision.Can you explain it?
SHAP values, attention maps, counterfactual traces, and audit-ready compliance reports — every decision pathway laid bare on a single screen. For the engineers, scientists, and compliance officers who are accountable for the output.
Every gear. Every wire.
Every reason.
This isn't a feature list — it's a preview of the product. Hover each cell to see the feature in motion.
SHAP Waterfall Explanations
Decompose any prediction into signed feature contributions. Understand exactly why sample #47821 was approved — not just that it was.
Attention Visualization
Map exactly which tokens, pixels, or nodes drove the model's attention. Auditable at the architecture level.
Counterfactual Generation
Generate the minimum-change path from DENIED to APPROVED. Produce the "right to explanation" output regulators require.
Decision Path Tracing
Follow every branch decision in gradient-boosted or tree-based models. Highlight the exact path taken for any input.
Compliance Report Export
One-click export of audit-ready documentation. PDF, JSON, HTML. Passes every regulatory framework your legal team will ask about.
Model Drift Detection
Monitor feature distributions and prediction drift in production. Get alerted before your model silently degrades.
Supported Frameworks
Works with any scikit-learn compatible model. Black-box or glass-box.
Bundled Explainers
12 explainer types. Glass-box and black-box. Global and local.
If you want it,
it's already on screen.
No form fields. No email gates. The install command is right here — and repeated at the header, mid-scroll, and footer.
# Install from PyPIpip install interpret# Quick startfrom interpret import showfrom interpret.glassbox import ExplainableBoostingClassifierebm = ExplainableBoostingClassifier()ebm.fit(X_train, y_train)show(ebm.explain_global())