交错思维中的模态隔离桥接：通过逐步强化监督模态转换
Bridging Modal Isolation in Interleaved Thinking: Supervising Modality Transitions via Stepwise Reinforcement
AI总结：
提出MoTiF框架，通过反射式SFT和Flow-GRPO优化模态转换保真度，解决交错思维中图像与文本脱节的模态隔离问题，提升跨模态一致性和任务准确性。
摘要：
交错思维是一种统一的多模态模型交替进行文本推理和视觉生成的方法，在空间和物理任务上显示出潜力。然而，在复杂的长链场景中，我们识别出一个基本故障模式：生成的图像偏离文本上下文，而后续文本忽略视觉证据，导致两种模态交替但并未真正相互通知。我们将其称为模态隔离，并归因于模态边界处的信息损失累积。我们将每个推理循环分解为原子操作，并定义模态转换损失，量化每个边界处的跨模态幻觉（文本到图像）和视觉利用不足（图像到文本）。我们提出MoTiF（模态转换保真度），一个两阶段训练框架，直接优化这些转换：反射式SFT训练模型检测和恢复错误的视觉输出；Flow-GRPO通过强化学习提高图像生成保真度。MoTiF中的所有训练信号来自转换级保真度而非最终任务准确性。在四个视觉谜题基准测试中，这种转换级监督显著提高了跨模态一致性和最终任务准确性。结果表明，有效的交错推理需要在模态边界处进行明确的结构监督，而不仅仅是扩展或最终任务优化。
英文摘要：
Interleaved thinking, where a unified multimodal model alternates between textual reasoning and visual generation, has shown promise on spatial and physical tasks. However, in complex long-chain scenarios, we identify a fundamental failure mode: generated images diverge from the textual context while subsequent text ignores the visual evidence, causing the two modalities to alternate without genuinely informing each other. We term this Modal Isolation and attribute it to compounding information loss at modality boundaries. We decompose each reasoning cycle into atomic operations and define modality transition loss, quantifying cross-modal hallucination (text-to-image) and visual utilization deficit (image-to-text) at each boundary. We propose MoTiF (Modality Tiransition Fidelity), a two-stage training framework that directly optimizes these transitions: Reflective SFT trains the model to detect and recover from erroneous visual outputs; Flow-GRPO improves image generation fidelity via reinforcement learning. All training signals in MoTiF derive from transition-level fidelity rather than end-task accuracy. Across four visual puzzle benchmarks, this transition-level supervision substantially improves both cross-modal coherence and final task accuracy. The results demonstrate that effective interleaved reasoning requires explicit structural supervision at modality boundaries, not merely scaling or end-task optimization.
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。
