Adaptive Semantic Branching：利用轨迹分歧点定向扩展推理候选
Adaptive Semantic Branching: Disagreement-Localized Cross-Trajectory Regeneration for Reasoning Candidate Expansion
arXiv 上一个免训练的推理技巧：找出几条推理轨迹的分岔点，借别的轨迹补生成，Qwen2.5-7B 在 MATH 上覆盖率从 89% 提到 91%。
Adaptive Semantic Branching（ASB）是一种无需训练的推理时方法，用于扩展推理候选。它先按最终答案对独立采样的轨迹分组，用单调动态规划对齐推理步骤，再通过 persistent-divergence 目标定位相似路径开始分岔的位置。在 500 道 MATH 问题上配合 Qwen2.5-7B-Instruct 测试，ASB 把 oracle 候选覆盖率从 7 个初始样本下的 89.0%（445/500）提升到 91.0%（455/500）。整个评测集共增加 576 次生成调用，平均每题 8.152 次。
Adaptive Semantic Branching: Disagreement-Localized Cross-Trajectory Regeneration for Reasoning Candidate Expansion
Reasoning trajectories generated for the same problem often arrive at conflicting answers. These disagreements reveal alternative intermediate decisions that can be reused to explore new solutions. We introduce Adaptive Semantic Branching (ASB), a training-free inference procedure that turns cross-trajectory disagreement into targeted candidate generation. ASB first groups independently sampled trajectories by extracted final answer, selects a representative from each group, and aligns their reasoning steps with a monotonic dynamic program. A persistent-divergence objective identifies the point at which initially similar reasoning paths develop different continuations. For each selected directed pair, ASB retains the source trajectory's textual prefix, extracts a short reasoning window from a competing donor trajectory, and generates a fresh solution conditioned on both. A diversity-aware policy allocates up to three additional generations according to the answer-cluster structure. In a 500-problem MATH evaluation with Qwen2.5-7B-Instruct, ASB increases oracle candidate coverage from 445/500 (89.0%) with seven initial samples to 455/500 (91.0%), recovering correct candidates for 10 of the 55 initially uncovered problems. The method uses 576 additional generation calls across the evaluation set, averaging 8.152 calls per problem. We provide a precise specification of the branching operator, alignment objective, generation policy, and experimental configuration.