Meta-CoT：提升图像编辑的粒度与泛化能力
Meta-CoT: Enhancing Granularity and Generalization in Image Editing
AI总结：
Meta-CoT通过双重分解提升图像编辑的粒度和泛化能力，通过任务-目标-理解能力三元组分解和五元基本元任务训练策略，显著提高编辑性能。
摘要：
统一的多模态理解和生成模型通过在Chain-of-Thought（CoT）过程中整合细粒度理解，已显示出改进的图像编辑性能。然而，关键问题仍未被充分探索：哪些形式的CoT和训练策略能共同提升理解和泛化能力？为此，我们提出了Meta-CoT，一种对单图像编辑操作进行两级分解的范式，具有两个关键特性：（1）可分解性。我们观察到任何编辑意图都可以表示为三元组-(任务，目标，所需理解能力)。受此启发，Meta-CoT分解了编辑任务和目标，生成任务特定的CoT，并在所有目标上遍历编辑操作。这种分解增强了模型对编辑操作的理解粒度，并在训练中指导学习三元组的每个元素，显著提高了编辑能力。（2）泛化性。在第二级分解中，我们进一步将编辑任务分解为五个基本元任务。我们发现训练这些五个元任务，加上三元组的其他两个元素，足以在各种未见过的编辑任务上实现强大的泛化能力。为了进一步使模型的编辑行为与其CoT推理对齐，我们引入了CoT-Editing一致性奖励，鼓励在编辑过程中更准确和有效地利用CoT信息。实验表明，我们的方法在21个编辑任务上实现了整体15.8%的提升，并且在仅训练少量元任务的情况下，能够有效泛化到未见过的编辑任务。我们的代码、基准和模型已发布在https://shiyi-zh0408.github.io/projectpages/Meta-CoT/
英文摘要：
Unified multi-modal understanding/generative models have shown improved image editing performance by incorporating fine-grained understanding into their Chain-of-Thought (CoT) process. However, a critical question remains underexplored: what forms of CoT and training strategy can jointly enhance both the understanding granularity and generalization? To address this, we propose Meta-CoT, a paradigm that performs a two-level decomposition of any single-image editing operation with two key properties: (1) Decomposability. We observe that any editing intention can be represented as a triplet - (task, target, required understanding ability). Inspired by this, Meta-CoT decomposes both the editing task and the target, generating task-specific CoT and traversing editing operations on all targets. This decomposition enhances the model's understanding granularity of editing operations and guides it to learn each element of the triplet during training, substantially improving the editing capability. (2) Generalizability. In the second decomposition level, we further break down editing tasks into five fundamental meta-tasks. We find that training on these five meta-tasks, together with the other two elements of the triplet, is sufficient to achieve strong generalization across diverse, unseen editing tasks. To further align the model's editing behavior with its CoT reasoning, we introduce the CoT-Editing Consistency Reward, which encourages more accurate and effective utilization of CoT information during editing. Experiments demonstrate that our method achieves an overall 15.8% improvement across 21 editing tasks, and generalizes effectively to unseen editing tasks when trained on only a small set of meta-tasks. Our code, benchmark, and model are released at https://shiyi-zh0408.github.io/projectpages/Meta-CoT/
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
