Efficiency without Compromise: CLIP-aided Text-to-Image GANs with Increased Diversity

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kobayashi, Yuya, Takida, Yuhta, Shibuya, Takashi, Mitsufuji, Yuki
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912409428426752
author Kobayashi, Yuya
Takida, Yuhta
Shibuya, Takashi
Mitsufuji, Yuki
author_facet Kobayashi, Yuya
Takida, Yuhta
Shibuya, Takashi
Mitsufuji, Yuki
contents Recently, Generative Adversarial Networks (GANs) have been successfully scaled to billion-scale large text-to-image datasets. However, training such models entails a high training cost, limiting some applications and research usage. To reduce the cost, one promising direction is the incorporation of pre-trained models. The existing method of utilizing pre-trained models for a generator significantly reduced the training cost compared with the other large-scale GANs, but we found the model loses the diversity of generation for a given prompt by a large margin. To build an efficient and high-fidelity text-to-image GAN without compromise, we propose to use two specialized discriminators with Slicing Adversarial Networks (SANs) adapted for text-to-image tasks. Our proposed model, called SCAD, shows a notable enhancement in diversity for a given prompt with better sample fidelity. We also propose to use a metric called Per-Prompt Diversity (PPD) to evaluate the diversity of text-to-image models quantitatively. SCAD achieved a zero-shot FID competitive with the latest large-scale GANs at two orders of magnitude less training cost.
format Preprint
id arxiv_https___arxiv_org_abs_2506_01493
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Efficiency without Compromise: CLIP-aided Text-to-Image GANs with Increased Diversity
Kobayashi, Yuya
Takida, Yuhta
Shibuya, Takashi
Mitsufuji, Yuki
Computer Vision and Pattern Recognition
Machine Learning
Recently, Generative Adversarial Networks (GANs) have been successfully scaled to billion-scale large text-to-image datasets. However, training such models entails a high training cost, limiting some applications and research usage. To reduce the cost, one promising direction is the incorporation of pre-trained models. The existing method of utilizing pre-trained models for a generator significantly reduced the training cost compared with the other large-scale GANs, but we found the model loses the diversity of generation for a given prompt by a large margin. To build an efficient and high-fidelity text-to-image GAN without compromise, we propose to use two specialized discriminators with Slicing Adversarial Networks (SANs) adapted for text-to-image tasks. Our proposed model, called SCAD, shows a notable enhancement in diversity for a given prompt with better sample fidelity. We also propose to use a metric called Per-Prompt Diversity (PPD) to evaluate the diversity of text-to-image models quantitatively. SCAD achieved a zero-shot FID competitive with the latest large-scale GANs at two orders of magnitude less training cost.
title Efficiency without Compromise: CLIP-aided Text-to-Image GANs with Increased Diversity
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2506.01493