A Universal Avoidance Method for Diverse Multi-branch Generation

2026-04-19

AI Summary

This work addresses the limited human-like creativity of existing generative models in multi-branch generation, where achieving both diversity and efficiency remains challenging. The authors propose UAG, a model-agnostic and computationally efficient generation strategy that enhances output diversity by dynamically penalizing similarity among generated samples during inference—without requiring additional training or complex architectural modifications. Notably, UAG is the first method to offer unified support for both diffusion models and Transformers, eliminating reliance on specific architectures or high computational overhead. Experimental results demonstrate that UAG improves diversity by 1.9×, accelerates inference by 4.4×, and achieves these gains with only 1/64 of the FLOPs compared to state-of-the-art approaches.

Technology Category

• Natural Language Processing: Generation
• Cognitive Modeling & Cognitive Systems: Computational Creativity
• Computer Vision: Diffusion Models for Vision

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Modern generative models still lack human-level creativity, particularly in multi-branch diversity. Prior approaches to address this problem often incur heavy computation or strong dependency on model architecture. Therefore, we introduce UAG(Universal Avoidance Generation), a model-agnostic and computationally efficient generation strategy that penalizes similarity among previously generated outputs. Thus, UAG can enhance multi-branch diversity across both diffusion and transformer models, with minimal additional computation. In experiments, our method achieves up to 1.9 times higher diversity, runs 4.4 times faster, and requires only 1/64 of the FLOPs compared to state-of-the-art methods. The full code is https://anonymous.4open.science/r/2026_ACL_Universal/.

Problem

• multi-branch generation
• diversity
• generative models
• creativity

Innovation

• Universal Avoidance Generation
• multi-branch diversity
• model-agnostic
• computational efficiency
• generative models