用自身预测反馈调整推理过程，提升模型准确率。
Steering Recurrent Reasoners at Inference Time with Readout Feedback
- 利用中间预测生成耦合力，动态调节隐状态演化
- 在六组实验中四组表现超越多步计算或轨迹采样
- 无需重训练，计算成本相当或更低，适合部署优化
循环模型通过共享计算模块反复更新隐状态，已成为解决复杂推理任务的强大架构。现有推理阶段方法通过增加迭代步数或采样更多轨迹来提升性能，但忽略了每条轨迹内部蕴含的信息。本文提出读出反馈（RoFB），一种测试时干预机制，将中间预测转化为逐标记的成对耦合力，注入隐状态动态中，无需重训练即可改进模型表现。在Sudoku和Maze任务上，对三种循环模型（AKOrN、ItrSA++、TRM）的实验显示，RoFB在六组模型-任务组合中，有四组获得显著提升，其性能超越单纯增加迭代步数或选择最优轨迹的方法，且计算开销相当或更低。结果表明，闭环调控隐状态演化可作为循环推理模型的补充推理控制机制。
原文摘要 · Abstract (English)
Recurrent models, which repeatedly update latent states with shared computation blocks, have emerged as powerful architectures for solving complex reasoning tasks. Existing inference-time methods scale computation by running more steps or sampling more trajectories, but ignore information revealed within each trajectory. Here we show that recurrent models can be improved at inference time by using their own readout probabilities to steer latent dynamics without retraining. We introduce Readout Feedback (RoFB), a test-time intervention that converts intermediate predictions into token-wise pairwise coupling forces injected into the latent dynamics. Across three recurrent models (AKOrN, ItrSA++, TRM) on Sudoku and Maze, RoFB yields clear gains in four of six model-task pairs, achieving performance unattainable by merely running more steps or selecting from multiple trajectories, at comparable or lower computational cost. These results suggest that closed-loop steering of latent dynamics can serve as a complementary inference-time control mechanism for recurrent reasoning models.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。