# albermax/innvestigate

A toolbox to iNNvestigate neural networks' predictions!

Repository: https://github.com/albermax/innvestigate
Canonical: https://ross.abutalabs.com/products/innvestigate
Language: Python
License: NOASSERTION
License Family: other
Last push: 2025-04-11T14:13:23+00:00

## Health v2 (maintenance only)
Score: 30/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 16, release rhythm 8, longevity 100
- inputs: {"age_days": 3185, "days_push": 509, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1309, forks 230 (observed 2026-08-28T04:04:19.521217+00:00)

## What it is
iNNvestigate is a Python toolbox providing a common interface and out-of-the-box implementations of many neural network explanation methods such as Saliency, GuidedBackprop, SmoothGrad, IntegratedGradients, LRP, PatternNet, and PatternAttribution. It is built on TensorFlow/Keras and aims to make analyzing neural network predictions easy and comparable.

## Use cases
- explain neural network predictions with saliency maps
- compute LRP relevance scores for a Keras model
- compare different XAI methods like IntegratedGradients and SmoothGrad
- visualize which pixels influenced an image classifier's decision
- interpret deep learning model decisions for research
- generate attribution maps for TensorFlow 2 models

## When to choose
- you use TensorFlow/Keras and need reference implementations of many explanation algorithms under one interface
- you want to compare XAI methods like LRP, Deep Taylor, or GuidedBackprop
- you need a maintained, citable library for neural network interpretability research

## When to avoid
- you work primarily in PyTorch (consider Captum instead)
- you need model-agnostic explanations for arbitrary black-box APIs
- you need explanations for non-neural-network models like gradient-boosted trees

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, nlp, image-processing
- domain: machine-learning, deep-learning, artificial-intelligence, data-science
- platform: python
- tags: xai, explainability, interpretability, saliency-maps, lrp, layer-wise-relevance-propagation, integrated-gradients, smoothgrad, tensorflow, neural-network-analysis

## Member repositories
- albermax/innvestigate (main) score 30

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:19.521217+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:49:58.566235+00:00, confidence not recorded.
  - readme: https://github.com/albermax/innvestigate (fetched 2026-08-28T04:04:19.521217+00:00, sha 9dfb8c517631)
  - registry_pypi: https://pypi.org/pypi/innvestigate/json (fetched 2026-08-29T12:08:12.162789+00:00, sha 72ad5f893649)
- Data as of 2026-08-30T08:39:29.467469+00:00.