arXiv：cs.AI· Zixu Wang, Xiaoran Zhao, Yahui Liu, Feng-ao Wang, Junhao Xiao, You Shu, Bing Liu, Yiting Liu, Songlei Jian, Zhiliang Tian, Dongsheng Li·· 22 小时前
GenRevisit：让图像生成智能体学会回退到更优状态
Look Back to Move Forward: Learning to Revisit for Agentic Compositional Image Generation
AI 导读
GenRevisit 是一个面向组合式图像生成的智能体框架，将动作选择扩展为「续接状态+动作」联合选择，使智能体可回退到历史候选图像而非仅从最新图像继续编辑。该框架经 SFT 加智能体强化学习训练，并提出 Normalized Action Credit Assignment（NACA）按各动作对减少剩余目标的贡献重新分配 GRPO 中的正优势。
正文
Authors:Zixu Wang, Xiaoran Zhao, Yahui Liu, Feng-ao Wang, Junhao Xiao, You Shu, Bing Liu, Yiting Liu, Songlei Jian, Zhiliang Tian, Dongsheng Li
Abstract:Text-to-image models have made remarkable progress in visual fidelity and instruction following, yet complex compositional generation remains challenging. Existing agentic approaches address this limitation through iterative generation and editing, adapting their actions based on interaction history. However, they typically fix the continuation state (i.e., the image from which refinement proceeds) to the latest image. Since iterative editing is not necessarily monotonic, the agent may continue refining from the latest image even when it has already degraded. In this work, we introduce GenRevisit, an agentic framework that extends action selection to joint continuation-state and action selection. The agent can not only use interaction history to adjust editing instructions but also select an image candidate to edit, jointly deciding where to continue from and how to refine. We train GenRevisit with SFT followed by agentic reinforcement learning. To address the coarse credit assignment of final outcome reward in GRPO, which assigns the same trajectory-level advantage to all actions, we further propose Normalized Action Credit Assignment (NACA) to redistribute positive advantage according to each action's contribution to reducing the remaining objective. On GenEval2, GenRevisit achieves GM scores of 84.3 and 74.4 with Qwen-Image-2512 and Z-Image-Turbo, surpassing the strongest baselines using the same generators by 12.7 and 10.9 points respectively. GenRevisit achieves state-of-the-art performance on GenEval, GenEval++, and GenEval2 among baselines.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.16224 [cs.CV] |
| (or arXiv:2610.16224v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.16224 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zixu Wang [view email]
[v1]
Mon, 12 Oct 2026 05:59:12 UTC (38,006 KB)
来源：arXiv：cs.AI · arxiv.org