arXiv：cs.AI· Myunghoon Kang, Jungseob Lee, Jaehyung Seo, Heuiseok Lim·· 6 小时前AI 评分37
ThinkFuse：面向小型推理模型的轨迹感知测试时融合
ThinkFuse: Trajectory-Aware Test-Time Fusion for Small Reasoning Models
AI 导读
ThinkFuse 是一个免训练的测试时融合框架，通过对比片段级不确定性与轨迹级不确定性趋势来识别不稳定推理点，并将辅助推理路径融入主模型轨迹。在数学与知识密集型推理基准上，它优于基线，且在不同模型族组合下均有稳定增益。该方法融合触发次数更少、生成 token 更少，主模型较小时仍保持稳健，代码已开源。
正文
Abstract:Small reasoning models (SRMs) have shown strong performance on complex reasoning tasks by generating extended chain-of-thought trajectories, but they often fail to recover once their reasoning enters an erroneous path. Existing test-time fusion methods rely on local fusion signals to determine when to trigger fusion, which can be misled by transient uncertainty fluctuations and may reinforce unstable reasoning trajectories. We propose ThinkFuse, a training-free test-time fusion framework that selectively intervenes in unreliable reasoning segments. ThinkFuse compares segment-level uncertainty shifts with trajectory-level uncertainty trends to identify unstable reasoning points and fuse auxiliary reasoning paths into the primary model's trajectory. Extensive experiments demonstrate that ThinkFuse outperforms baselines on mathematical and knowledge-intensive reasoning benchmarks, with consistent gains across model-family combinations, and remains robust with a smaller primary model. Our analysis shows that ThinkFuse requires fewer fusion triggers and generates fewer tokens, highlighting the efficiency of selective triggering. Our code is available at this https URL.
| Comments: | Accepted to EMNLP 2026 Findings |
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.07803 [cs.AI] |
| (or arXiv:2610.07803v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07803 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Myunghoon Kang [view email]
[v1]
Tue, 6 Oct 2026 05:56:21 UTC (745 KB)
来源：arXiv：cs.AI · arxiv.org