# openai/improved-gan

Code for the paper "Improved Techniques for Training GANs"

Repository: https://github.com/openai/improved-gan
Canonical: https://ross.abutalabs.com/products/improved-gan
Homepage: https://arxiv.org/abs/1606.03498
Language: Python
License Family: other
Topics: paper
Archived: true
Last push: 2018-11-21T23:56:07+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3738, "days_push": 2842, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2333, forks 615 (observed 2026-08-28T04:06:37.865797+00:00)

## What it is
Reference implementation of the techniques from the OpenAI paper 'Improved Techniques for Training GANs' (arXiv:1606.03498), including semi-supervised classification and image generation experiments. It contains experiment code for MNIST, SVHN, CIFAR-10, and ImageNet, and is archived with no further updates.

## Use cases
- reproduce GAN training results from the Improved Techniques paper
- implement semi-supervised learning with GANs
- study feature matching and minibatch discrimination techniques
- generate realistic MNIST or CIFAR-10 images with a GAN
- learn how GAN training stability tricks are implemented
- run ImageNet GAN sample generation experiments

## When to choose
- you need the exact code behind the 2016 Improved Techniques for Training GANs paper
- you are studying historical GAN research or reproducing published baselines
- you want a reference implementation of semi-supervised GAN classification

## When to avoid
- you need maintained, production-ready GAN tooling
- you want a license-cleared codebase for commercial use (no license is provided)
- you need modern GAN architectures or up-to-date framework support

## Facets
- artifact type: library
- maturity: abandoned
- function: machine-learning, deep-learning, image-processing
- domain: machine-learning, deep-learning, computer-vision
- platform: python
- tags: gan, generative-adversarial-networks, semi-supervised-learning, research-code, paper-reproduction, tensorflow, mnist, cifar10, imagenet, research, gpu

## Member repositories
- openai/improved-gan (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:37.865797+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 2026-08-30T02:37:55.536580+00:00, confidence not recorded.
  - readme: https://github.com/openai/improved-gan (fetched 2026-08-28T04:06:37.865797+00:00, sha 4042460b50eb)
  - homepage: https://arxiv.org/abs/1606.03498 (fetched 2026-08-29T10:18:28.551049+00:00, sha 413a147dbb11)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T10:18:28.560448+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T10:18:28.563937+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T10:18:28.565652+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T10:18:28.562333+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.