Scalable GANs with Transformers

Researchers introduced GAT, a purely transformer-based Generative Adversarial Network trained in a compact Variational Autoencoder latent space, achieving state-of-the-art single-step class-conditional generation on ImageNet-256 with an FID of 2.18 in just 60 epochs. The model scales from S to XL sizes by addressing failure modes like early-layer underutilization and optimization instability through lightweight intermediate supervision and width-aware learning-rate adjustment.

Computer Science Computer Vision and Pattern Recognition Submitted on 29 Sep 2025 v1 https://arxiv.org/abs/2509.24935v1 , last revised 5 Jun 2026 this version, v3 Title:Scalable GANs with Transformers View PDF /pdf/2509.24935 HTML experimental https://arxiv.org/html/2509.24935v3 Abstract:Scalability has driven recent advances in generative modeling, yet its principles remain underexplored for adversarial learning. We investigate the scalability of Generative Adversarial Networks GANs through two design choices that have proven to be effective in other types of generative models: training in a compact Variational Autoencoder latent space and adopting purely transformer-based generators and discriminators. Training in latent space enables efficient computation while preserving perceptual fidelity, and this efficiency pairs naturally with plain transformers, whose performance scales with computational budget. Building on these choices, we analyze failure modes that emerge when naively scaling GANs. Specifically, we find issues as underutilization of early layers in the generator and optimization instability as the network scales. Accordingly, we provide simple and scale-friendly solutions as lightweight intermediate supervision and width-aware learning-rate adjustment. Our experiments show that GAT, a purely transformer-based and latent-space GANs, can be easily trained reliably across a wide range of capacities S through XL . Moreover, GAT-XL/2 achieves state-of-the-art single-step, class-conditional generation performance FID of 2.18 on ImageNet-256 in just 60 epochs, 4x fewer epochs than strong baselines. Project page: this https URL . Submission history From: Sangeek Hyun view email /show-email/6304dfc3/2509.24935 Mon, 29 Sep 2025 15:36:15 UTC 41,481 KB v1 /abs/2509.24935v1 Tue, 26 May 2026 03:14:03 UTC 16,766 KB v2 /abs/2509.24935v2 v3 Fri, 5 Jun 2026 02:15:59 UTC 16,766 KB Current browse context: cs.CV References & Citations Loading... Bibliographic and Citation Tools Bibliographic Explorer What is the Explorer? https://info.arxiv.org/labs/showcase.html arxiv-bibliographic-explorer Connected Papers What is Connected Papers? https://www.connectedpapers.com/about Litmaps What is Litmaps? https://www.litmaps.co/ scite Smart Citations What are Smart Citations? https://www.scite.ai/ Code, Data and Media Associated with this Article alphaXiv What is alphaXiv? https://alphaxiv.org/ CatalyzeX Code Finder for Papers What is CatalyzeX? https://www.catalyzex.com DagsHub What is DagsHub? https://dagshub.com/ Gotit.pub What is GotitPub? http://gotit.pub/faq Hugging Face What is Huggingface? https://huggingface.co/huggingface ScienceCast What is ScienceCast? https://sciencecast.org/welcome Demos Recommenders and Search Tools Influence Flower What are Influence Flowers? https://influencemap.cmlab.dev/ CORE Recommender What is CORE? https://core.ac.uk/services/recommender arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs https://info.arxiv.org/labs/index.html .