GOVERNATLAS / PRODUCT DOSSIER

Alibi Explain

Alibi documents a source-available Python library for machine-learning model inspection and explanation.

Explore the product ↗Add to comparison ↗

01 / CONTEXT & SCOPE

Understand the proposition.
Define the boundary.

Where it fits

Choose a method whose access and data requirements match the model and the explanation question.

The decision to make

Choose the explanation method and its input requirements explicitly. Local anchors, counterfactuals and global summaries answer different questions. The current README calls Alibi source-available, so review the exact release license before assuming an unrestricted open-source deployment. Keep it distinct from the existing Alibi Detect product.

What this record does not establish

Current primary material describes a source-available explanation library, correcting the initial research candidate wording. License permissions, method-specific dependencies and commercial support need separate review; no detection-product parity or causal assurance is inferred.

1 current source · 4 published assertions
Source coverage is a research boundary, not a quality score.

02 / YOUR EVALUATION

Turn interest into a useful trial.

Open the evaluation playbook ↗

A test worth running

For a synthetic routing classifier, inspect one local explanation and its metadata, then test a documented input change in an authorized environment. Retain fit data, method settings and an unresolved result. Ask a reviewer why the chosen method answers the stated question and why a different method may not.

GovernAtlas editorial proposal. We have not performed this test.

Evidence to bring back

An exact package/license/method/input-requirements manifest.
An explanation object with settings, evidence references and interpretation limits.

THE OPERATING DETAIL

Match the method to the question and the available access.

GovernAtlas analysis: practical review considerations, not findings from a product test.

Define local or global scope

Start with a fictional review question and decide whether it concerns one prediction or behavior across a reference population. Record the selected method, model-access requirement and dataset. A global explanation and a local explanation answer different questions even when they use the same feature names.

Retain method-specific metadata

Preserve initialization, fitting data references, hyperparameters and the returned explanation metadata. Repeat a small synthetic example after changing one setting. Require a reviewer to identify which assumptions changed, rather than comparing only screenshots whose configuration is no longer recoverable.

Separate projects and release rights

Confirm the exact Alibi Explain release and applicable license before integration. Keep outlier and drift-detection requirements separate from explanation requirements associated with Alibi Detect. Historical names or shared maintainers do not establish interchangeable features, deployment rights or support commitments.

03 / DOCUMENTED EVIDENCE

Read the claim. Inspect the basis.

Each statement retains its classification and preserved source. A vendor assertion is not an independent test result.

first party asserted

The current README identifies Alibi as source-available and distinguishes it from Alibi Detect; exact release license rights require separate review.

Source, dates & preserved provenance

Alibi Explain — primary scope and current identity ↗

Authority: first party
Observed 2026-10-16 · Reviewed 2026-10-16 · current until 2026-11-15

Artifact SHA-256: 837ae032cd2069bd7547273a6dc2be2dc5e5f9aa9be7a88dfc31ae6e36447a8e

The hash identifies the preserved capture. The live source may have changed.

first party asserted

Its API separates initialization, fitting and explanation, returning metadata and method-specific explanation data.

Source, dates & preserved provenance

Alibi Explain — primary scope and current identity ↗

Authority: first party
Observed 2026-10-16 · Reviewed 2026-10-16 · current until 2026-11-15

Artifact SHA-256: 837ae032cd2069bd7547273a6dc2be2dc5e5f9aa9be7a88dfc31ae6e36447a8e

The hash identifies the preserved capture. The live source may have changed.

first party asserted

Alibi documents local and global methods for classification and regression, including Anchors, accumulated local effects, counterfactuals and SHAP methods.

Source, dates & preserved provenance

Alibi Explain — primary scope and current identity ↗

Authority: first party
Observed 2026-10-16 · Reviewed 2026-10-16 · current until 2026-11-15

Artifact SHA-256: 837ae032cd2069bd7547273a6dc2be2dc5e5f9aa9be7a88dfc31ae6e36447a8e

The hash identifies the preserved capture. The live source may have changed.

first party asserted

Alibi documents a source-available Python library for machine-learning model inspection and explanation. Scope: primary documentation only; no independent model test, fairness determination, causal finding or compliance approval.

Source, dates & preserved provenance

Alibi Explain — primary scope and current identity ↗

Authority: first party
Observed 2026-10-16 · Reviewed 2026-10-16 · current until 2026-11-15

Artifact SHA-256: 837ae032cd2069bd7547273a6dc2be2dc5e5f9aa9be7a88dfc31ae6e36447a8e

The hash identifies the preserved capture. The live source may have changed.

Source register

Alibi Explain — primary scope and current identity ↗first party · captured 2026-10-16

Brand asset provenance

Organization identity asset from its official website; inclusion does not imply endorsement.

Official website ↗ · Original asset ↗

04 / BEFORE YOU COMMIT

Better questions.
A clearer decision.

Resolve these with the provider and your own evaluation team.

Can you provide an exact package/license/method/input-requirements manifest for the selected configuration?

Can you provide an explanation object with settings, evidence references and interpretation limits for the selected configuration?

Which exact version, entitlements, data paths and responsible roles apply?

Regulatory applicability requires a separate, current determination. Inclusion is not certification.

COMPLIANCE / CLAIMS, SCOPE & GAPS

Start with the requirement.
Inspect the support.

Prepare your vendor review ↗

These labels describe this dossier’s current evidence coverage. A published mention may be a framework mapping, a qualified assertion or a limitation; it is not a compliance finding.