Research

Paper

AI LLM March 19, 2026

ADAPT: Attention Driven Adaptive Prompt Scheduling and InTerpolating Orthogonal Complements for Rare Concepts Generation

Authors

Kwanyoung Lee, Hyunwoo Oh, SeungJu Cha, Sungho Koh, Dong-Jin Kim

Abstract

Generating rare compositional concepts in text-to-image synthesis remains a challenge for diffusion models, particularly for attributes that are uncommon in the training data. While recent approaches, such as R2F, address this challenge by utilizing LLM for prompt scheduling, they suffer from inherent variance due to the randomness of language models and suboptimal guidance from iterative text embedding switching. To address these problems, we propose the ADAPT framework, a training-free framework that deterministically plans and semantically aligns prompt schedules, providing consistent guidance to enhance the composition of rare concepts. By leveraging attention scores and orthogonal components, ADAPT significantly enhances compositional generation of rare concepts in the RareBench benchmark without additional training or fine-tuning. Through comprehensive experiments, we demonstrate that ADAPT achieves superior performance in RareBench and accurately reflects the semantic information of rare attributes, providing deterministic and precise control over the generation of rare compositions without compromising visual integrity.

Metadata

arXiv ID: 2603.19157
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-03-19
Fetched: 2026-03-20 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.19157v1</id>\n    <title>ADAPT: Attention Driven Adaptive Prompt Scheduling and InTerpolating Orthogonal Complements for Rare Concepts Generation</title>\n    <updated>2026-03-19T17:11:49Z</updated>\n    <link href='https://arxiv.org/abs/2603.19157v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.19157v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Generating rare compositional concepts in text-to-image synthesis remains a challenge for diffusion models, particularly for attributes that are uncommon in the training data. While recent approaches, such as R2F, address this challenge by utilizing LLM for prompt scheduling, they suffer from inherent variance due to the randomness of language models and suboptimal guidance from iterative text embedding switching. To address these problems, we propose the ADAPT framework, a training-free framework that deterministically plans and semantically aligns prompt schedules, providing consistent guidance to enhance the composition of rare concepts. By leveraging attention scores and orthogonal components, ADAPT significantly enhances compositional generation of rare concepts in the RareBench benchmark without additional training or fine-tuning. Through comprehensive experiments, we demonstrate that ADAPT achieves superior performance in RareBench and accurately reflects the semantic information of rare attributes, providing deterministic and precise control over the generation of rare compositions without compromising visual integrity.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <published>2026-03-19T17:11:49Z</published>\n    <arxiv:comment>Accepted in CVPR 2026 (findings). 10 pages, 4 figures; supplementary material included (8 pages, 10 figures)</arxiv:comment>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Kwanyoung Lee</name>\n    </author>\n    <author>\n      <name>Hyunwoo Oh</name>\n    </author>\n    <author>\n      <name>SeungJu Cha</name>\n    </author>\n    <author>\n      <name>Sungho Koh</name>\n    </author>\n    <author>\n      <name>Dong-Jin Kim</name>\n    </author>\n  </entry>"
}