# lucidrains/gigagan-pytorch

Implementation of GigaGAN, new SOTA GAN out of Adobe. Culmination of nearly a decade of research into GANs

Repository: https://github.com/lucidrains/gigagan-pytorch
Canonical: https://ross.abutalabs.com/products/gigagan-pytorch
Language: Python
License: MIT
License Family: permissive
Topics: artificial-intelligence, deep-learning, generative-adversarial-network
Last push: 2025-01-12T14:26:17+00:00

## Health v2 (maintenance only)
Score: 21/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 1, release rhythm 8, longevity 90
- inputs: {"age_days": 1272, "days_push": 598, "days_rel": 598, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1942, forks 106 (observed 2026-08-28T04:05:57.500919+00:00)

## What it is
A PyTorch implementation of GigaGAN, Adobe's state-of-the-art generative adversarial network for text-to-image and unconditional image synthesis, including 1k-4k image upsamplers. It incorporates stability and convergence improvements from Lightweight GAN such as skip layer excitation and auxiliary reconstruction losses.

## Use cases
- train a GAN to generate images from text descriptions
- generate high-resolution images with a GAN in pytorch
- upsample images from 1k to 4k resolution
- replicate the GigaGAN paper results
- train an unconditional GAN on my own image dataset
- experiment with generative adversarial network architectures

## When to choose
- you want an open-source PyTorch implementation of GigaGAN to train or fine-tune
- you need GAN-based image upsampling toward 4k resolution
- you want to experiment with or extend a research GAN codebase

## When to avoid
- you need a production-ready, well-supported text-to-image model out of the box
- you lack GPU resources, since training GANs at scale is compute-intensive
- you prefer diffusion-based image generation with mature tooling

## Facets
- artifact type: library
- maturity: experimental
- function: machine-learning, deep-learning, image-processing
- domain: artificial-intelligence, deep-learning, image-processing
- platform: python
- tags: gan, gigagan, image-generation, text-to-image, pytorch, upsampling, generative-models, gpu

## Member repositories
- lucidrains/gigagan-pytorch (main) score 21

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:57.500919+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 2026-08-30T03:08:04.495085+00:00, confidence not recorded.
  - readme: https://github.com/lucidrains/gigagan-pytorch (fetched 2026-08-28T04:05:57.500919+00:00, sha 74e0818ccd66)
- Data as of 2026-08-30T08:39:29.467469+00:00.