YINYANG-ALIGN: Benchmarking Contradictory Objectives and Proposing Multi-Objective Optimization based DPO for Text-to-Image Alignment

TL;DR

YinYangAlign introduces a comprehensive benchmarking framework for evaluating and improving the alignment of text-to-image systems, addressing fundamental conflicting objectives to enhance fidelity, ethics, and creativity.

Contribution

It proposes a novel benchmarking framework and multi-objective optimization method to systematically address contradictory alignment objectives in T2I systems.

Findings

Quantifies alignment fidelity across six conflicting objectives.

Provides detailed datasets with human prompts and AI responses.

Demonstrates improved alignment through proposed optimization.

Abstract

Precise alignment in Text-to-Image (T2I) systems is crucial to ensure that generated visuals not only accurately encapsulate user intents but also conform to stringent ethical and aesthetic benchmarks. Incidents like the Google Gemini fiasco, where misaligned outputs triggered significant public backlash, underscore the critical need for robust alignment mechanisms. In contrast, Large Language Models (LLMs) have achieved notable success in alignment. Building on these advancements, researchers are eager to apply similar alignment techniques, such as Direct Preference Optimization (DPO), to T2I systems to enhance image generation fidelity and reliability. We present YinYangAlign, an advanced benchmarking framework that systematically quantifies the alignment fidelity of T2I systems, addressing six fundamental and inherently contradictory design objectives. Each pair represents fundamental tensions in image generation, such as balancing adherence to user prompts with creative modifications or maintaining diversity alongside visual coherence. YinYangAlign includes detailed axiom datasets featuring human prompts, aligned (chosen) responses, misaligned (rejected) AI-generated outputs, and explanations of the underlying contradictions.