# dorarad/gansformer

Generative Adversarial Transformers

Repository: https://github.com/dorarad/gansformer
Canonical: https://ross.abutalabs.com/products/gansformer
Language: Python
License: MIT
License Family: permissive
Topics: transformers, gans, generative-adversarial-networks, image-generation, scene-generation, compositionality, attention
Last push: 2022-06-14T16:57:20+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 2011, "days_push": 1541, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1344, forks 152 (observed 2026-08-28T04:04:27.214032+00:00)

## What it is
GANformer is a research implementation of a generative adversarial transformer that uses a bipartite attention structure for efficient high-resolution image generation. It supports both PyTorch and TensorFlow and generalizes StyleGAN with multiplicative region-based modulation to encourage compositional scene representations.

## Use cases
- generate images with a transformer-based GAN
- train a GAN on custom image datasets
- experiment with compositional scene generation
- sample images from pretrained GANformer models
- research attention-based generative adversarial networks
- compare against StyleGAN architectures

## When to choose
- you need a research-grade GAN with transformer attention for image synthesis
- you want to study compositional and region-based generative modeling
- you want pretrained models for image generation experiments

## When to avoid
- you need a production image-generation service with active support
- you require the latest GPU stack since the code targets older PyTorch/TensorFlow and CUDA versions
- you want a simple off-the-shelf text-to-image generator

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing, graphics
- domain: machine-learning, deep-learning, computer-vision, image-processing
- platform: python
- tags: gan, transformer, generative-models, stylegan, image-synthesis, research-code, pytorch, tensorflow, gpu

## Member repositories
- dorarad/gansformer (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:27.214032+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 2026-08-30T04:42:49.747793+00:00, confidence not recorded.
  - readme: https://github.com/dorarad/gansformer (fetched 2026-08-28T04:04:27.214032+00:00, sha d0797a607c98)
- Data as of 2026-08-30T08:39:29.467469+00:00.