arXiv cs.CL· Haotian Wu, Zeming Chen, Hao Zhao, Antoine Bosselut·· 11 小时前AI 评分38
TTT-Consolidate：用测试时训练整合多智能体推理轨迹
Consolidating Reasoning with Test-Time Training
AI 导读
研究者提出 TTT-Consolidate 框架，将多智能体推理轨迹通过测试时训练整合进轻量 LoRA 适配器，再据此生成最终解，在数学与 STEM 基准上比最佳聚合基线高出最多 20 个百分点。
正文 · 原文
Abstract:Effective multi-agent collaboration requires finalizing solutions from reasoning that is distributed across the individual attempts of many agents. However, useful intermediate reasoning can remain unexploited when collaboration is limited to voting on conclusions reached among individual agents, or aggregating long, noisy collections of reasoning trajectories in context. In this work, we propose TTT-Consolidate, a framework that uses multi-agent reasoning trajectories as adaptation data to improve a model's subsequent reasoning. TTT-Consolidate jointly consolidates trajectories into parametric representations through test-time training, then generates a new solution by conditioning on the consolidated information. We meta-learn how to consolidate using bi-level optimization at training time, where an inner loop encodes reasoning trajectories into a lightweight LoRA adapter, and an outer loop learns meta-parameters for the inner loop such that the consolidation is helpful for the model to synthesize a final solution.
As no large-scale corpus of diverse reasoning trajectories exists for training aggregation methods, we construct the OpenParallelThinking corpus composed of 12,810 math problems paired with a mixture of trajectories from six LLMs. Evaluated on challenging math and STEM benchmarks, TTT-Consolidate outperforms the best aggregation baselines by up to 20 percentage points, even under matched compute and runtime settings, with the gap widening at larger model scales. TTT-Consolidate also recovers from unsuccessful individual trajectories more robustly, an ability that improves as the number of available (even incorrect) trajectories increases. Finally, we demonstrate that TTT-Consolidate benefits from both SFT and RL-based outer loop objectives. We will release TTT-Consolidate and OpenParallelThinking for use in future work on reasoning aggregation.
| Comments: | NeurIPS 2026, 25 pages, test-time training, reasoning aggregation, meta-learning |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.02104 [cs.CL] |
| (or arXiv:2611.02104v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.02104 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zeming Chen [view email]
[v1]
Sun, 1 Nov 2026 18:18:22 UTC (741 KB)
来源：arXiv cs.CL · arxiv.org