CapField-OPD：通过联合锚定多教师在线策略蒸馏学习连续能力场用于流模型
CapField-OPD: Learning Continuous Capability Fields via Joint-Anchored Multi-Teacher On-Policy Distillation for Flow Models
浏览论文内容
中文总结 AI 辅助
CapField-OPD通过显式能力坐标将多教师整合为连续能力场，解耦能力与提示语义，实现连续能力控制与测试时扩展，在组合生成等任务上保持或超越专家性能。
中文摘要 AI 辅助
奖励特化的后训练为基于流的生成模型产生了强大的专家，而多教师在线策略蒸馏（OPD）将他们的能力整合到一个学生模型中。然而，现有方法根据提示的语义类别将每个提示路由到单个教师，隐式地将所需能力与提示内容绑定。这种耦合使得能力调用容易受到提示扰动的影响，并阻止用户在推理时显式调整所需能力的强度。在本工作中，我们引入了CapField-OPD，一种通过显式能力坐标将多个教师整合到连续能力场中的OPD框架。我们使用教师模型作为锚点来构建该场，坐标决定其输出的组合方式。因此，每种能力配置都获得独特的监督目标，能力控制不再依赖于提示语义。由于训练锚点在推理时可能不是最优的，我们进一步在小型校准集上对学习到的场进行剖析。具有最高平均奖励的坐标作为推荐的默认值，而经常最优的坐标则为测试时扩展提供了有前景的候选集。在组合生成、文本渲染和视觉美学上的广泛实验表明，CapField-OPD将多个特化教师整合到单个学生中，同时保持或超越其性能，在保持语义的提示变化下可靠地调用所需能力，并支持连续能力控制和基于坐标的测试时扩展。
英文摘要
Reward-specialized post-training produces strong experts for flow-based generative models, while multi-teacher on-policy distillation (OPD) consolidates their capabilities into a single student. Existing methods, however, route each prompt to a single teacher according to its semantic category, implicitly binding the desired capability to prompt content. This coupling makes capability invocation vulnerable to prompt perturbations and prevents users from explicitly adjusting the strength of the desired capability at inference time. In this work, we introduce CapField-OPD, an OPD framework that integrates multiple teachers into a continuous capability field through explicit capability coordinates. We use teacher models as anchors to construct this field, with the coordinates determining how their outputs are combined. Each capability configuration thus receives a unique supervision target, and capability control no longer depends on prompt semantics. Since the training anchors may not be optimal at inference time, we further profile the learned field on a small calibration set. The coordinate with the highest mean reward serves as the recommended default, while coordinates that are frequently optimal offer a promising candidate set for test-time scaling. Extensive experiments on compositional generation, text rendering, and visual aesthetics demonstrate that CapField-OPD consolidates multiple specialized teachers into a single student while preserving or surpassing their performance, reliably invokes the desired capabilities under semantics-preserving prompt variations, and supports continuous capability control and coordinate-based test-time scaling.
发表机构
- University of Science and Technology of China（中国科学技术大学）
- Shanghai Jiao Tong University（上海交通大学）
- Fudan University（复旦大学）
- Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）
机构由 AI 辅助整理，请以论文原文为准。
