# thu-ml/zhusuan

A probabilistic programming library for Bayesian deep learning, generative models, based on Tensorflow

Repository: https://github.com/thu-ml/zhusuan
Canonical: https://ross.abutalabs.com/products/zhusuan
Homepage: http://zhusuan.readthedocs.io
Language: Python
License: MIT
License Family: permissive
Topics: bayesian-inference, probabilistic-programming, graphical-models, generative-models, deep-learning
Last push: 2022-12-17T20:33:19+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3698, "days_push": 1355, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2219, forks 413 (observed 2026-08-28T04:06:27.403838+00:00)

## What it is
ZhuSuan is a Python probabilistic programming library for Bayesian deep learning built on TensorFlow. It provides primitives for building probabilistic models and inference algorithms such as variational inference, importance sampling, and MCMC methods.

## Use cases
- build Bayesian neural networks with variational inference
- train deep generative models like VAEs
- run Hamiltonian Monte Carlo on probabilistic models
- apply stochastic gradient MCMC to deep learning models
- evaluate generative models with importance sampling
- prototype hierarchical Bayesian models

## When to choose
- you need Bayesian inference combined with deep learning in TensorFlow
- you want advanced gradient estimators like VIMCO or REINFORCE
- you need programmable variational posteriors or importance sampling proposals

## When to avoid
- you work with PyTorch or JAX instead of TensorFlow
- you need a library with frequent updates and active community support
- you only need standard supervised deep learning without probabilistic modeling

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, simulation
- domain: machine-learning, deep-learning, artificial-intelligence
- platform: python
- tags: probabilistic-programming, bayesian-inference, generative-models, tensorflow, variational-inference, mcmc

## Member repositories
- thu-ml/zhusuan (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:27.403838+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 2026-08-30T02:45:52.840725+00:00, confidence not recorded.
  - readme: https://github.com/thu-ml/zhusuan (fetched 2026-08-28T04:06:27.403838+00:00, sha 759567101eeb)
- Data as of 2026-08-30T08:39:29.467469+00:00.