# google-research/disentanglement_lib

disentanglement_lib is an open-source library for research on learning disentangled representations.

Repository: https://github.com/google-research/disentanglement_lib
Canonical: https://ross.abutalabs.com/products/disentanglement_lib
Language: Python
License: Apache-2.0
License Family: permissive
Archived: true
Last push: 2021-05-16T02:54:11+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 2762, "days_push": 1935, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1425, forks 202 (observed 2026-08-28T04:04:41.465246+00:00)

## What it is
disentanglement_lib is an open-source Python library for research on learning disentangled representations, supporting models like BetaVAE, FactorVAE, BetaTCVAE, and DIP-VAE along with standard disentanglement metrics and datasets. It provides a pipeline for model training, postprocessing, evaluation, and visualization, and ships with over 10,000 pretrained disentanglement models.

## Use cases
- train and evaluate disentangled representation learning models
- compute disentanglement metrics like MIG, SAP, DCI, and FactorVAE score
- benchmark VAE variants on datasets like dSprites and Shapes3D
- use pretrained disentanglement models for research experiments
- visualize learned latent representations
- reproduce large-scale disentanglement study results

## When to choose
- you are doing academic research on disentangled representation learning
- you need standardized implementations of disentanglement metrics and datasets
- you want to compare VAE-based models under a unified framework
- you need pretrained disentanglement models for downstream experiments

## When to avoid
- you need production-ready representation learning for applications
- you prefer PyTorch over TensorFlow
- you need the latest actively maintained deep learning tooling
- you work on Windows or non-Linux environments

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, benchmarking, data-science, data-visualization
- domain: machine-learning, deep-learning, artificial-intelligence, computer-vision
- platform: python
- tags: disentangled-representations, variational-autoencoder, tensorflow, unsupervised-learning, research, pretrained-models, metrics, linux, gpu

## Member repositories
- google-research/disentanglement_lib (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:41.465246+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 2026-08-30T04:37:31.237480+00:00, confidence not recorded.
  - readme: https://github.com/google-research/disentanglement_lib (fetched 2026-08-28T04:04:41.465246+00:00, sha 2358763cb208)
  - registry_pypi: https://pypi.org/pypi/disentanglement_lib/json (fetched 2026-08-29T11:49:23.863242+00:00, sha 94a81c23e9a6)
- Data as of 2026-08-30T08:39:29.467469+00:00.