CHATS: Combining Human-Aligned Optimization and Test-Time Sampling for Text-to-Image Generation

TL;DR

CHATS is a novel framework that combines human-aligned optimization with test-time sampling to improve text-to-image generation quality, efficiency, and alignment with human preferences.

Contribution

This work introduces CHATS, a new generative framework that models preferred and dispreferred distributions separately and uses proxy-prompt sampling for better results.

Findings

Surpasses traditional preference alignment methods.

Achieves state-of-the-art performance on benchmarks.

Exhibits high data efficiency with limited fine-tuning data.

Abstract

Diffusion models have emerged as a dominant approach for text-to-image generation. Key components such as the human preference alignment and classifier-free guidance play a crucial role in ensuring generation quality. However, their independent application in current text-to-image models continues to face significant challenges in achieving strong text-image alignment, high generation quality, and consistency with human aesthetic standards. In this work, we for the first time, explore facilitating the collaboration of human performance alignment and test-time sampling to unlock the potential of text-to-image models. Consequently, we introduce CHATS (Combining Human-Aligned optimization and Test-time Sampling), a novel generative framework that separately models the preferred and dispreferred distributions and employs a proxy-prompt-based sampling strategy to utilize the useful information contained in both distributions. We observe that CHATS exhibits exceptional data efficiency, achieving strong performance with only a small, high-quality funetuning dataset. Extensive experiments demonstrate that CHATS surpasses traditional preference alignment methods, setting new state-of-the-art across various standard benchmarks.