# NVlabs/NVAE

The Official PyTorch Implementation of "NVAE: A Deep Hierarchical Variational Autoencoder" (NeurIPS 2020 spotlight paper)

Repository: https://github.com/NVlabs/NVAE
Canonical: https://ross.abutalabs.com/products/nvae
Homepage: https://arxiv.org/abs/2007.03898
Language: Python
License: NOASSERTION
License Family: other
Last push: 2022-12-06T03:23:32+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": 2206, "days_push": 1366, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1092, forks 176 (observed 2026-08-28T04:03:33.274592+00:00)

## What it is
Official PyTorch implementation of NVAE, a deep hierarchical variational autoencoder for likelihood-based image generation from a NeurIPS 2020 spotlight paper. It achieves state-of-the-art results among non-autoregressive likelihood models on MNIST, CIFAR-10, CelebA 64, and CelebA HQ datasets.

## Use cases
- train a hierarchical VAE on CIFAR-10 or MNIST
- generate high-quality face images with a VAE
- reproduce NeurIPS 2020 NVAE paper results
- benchmark likelihood-based generative models in bits per dimension
- experiment with deep hierarchical VAE architectures in PyTorch
- train a generative image model on CelebA HQ 256

## When to choose
- you need a strong, well-known VAE baseline for image generation research
- you want tractable, fast sampling with an encoder network unlike flows or autoregressive models
- you are reproducing or extending published hierarchical VAE results

## When to avoid
- you need actively maintained code with modern PyTorch versions - the repo has not been updated since late 2022
- you want state-of-the-art image generation today, since diffusion models now outperform it
- you need a production-ready generative model rather than research code

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing
- domain: deep-learning, machine-learning, image-processing, computer-vision
- platform: python
- tags: variational-autoencoder, pytorch, generative-models, image-generation, research-code, neurips-2020, gpu, linux

## Member repositories
- NVlabs/NVAE (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:33.274592+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 2026-08-30T06:48:59.920051+00:00, confidence not recorded.
  - readme: https://github.com/NVlabs/NVAE (fetched 2026-08-28T04:03:33.274592+00:00, sha daed3004f5c9)
  - homepage: https://arxiv.org/abs/2007.03898 (fetched 2026-08-29T12:51:01.753971+00:00, sha 0f234633100b)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T12:51:01.757772+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T12:51:01.760924+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T12:51:01.762743+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T12:51:01.759378+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.