arXiv：cs.CL· Xinglin Wang, Zishen Liu, Tong Zheng, Shaoxiong Feng, Peiwen Yuan, Yiwei Li, Jiayi Shi, Yueqi Zhang, Chuyi Tan, Ji Zhang, Boyuan Pan, Kan Li·· 3 小时前AI 评分31
PersonTTS：通过摊销式智能体策略发现实现个性化测试时扩展
From Pareto to Preference: Personalized Test-Time Scaling via Amortized Agentic Policy Discovery
AI 导读
研究者提出 PersonTTS，一个摊销式智能体策略发现框架，通过需求匹配的控制器初始化和源蒸馏程序指导复用历史搜索经验，为不同用户画像发现最大化多维需求联合满足率的测试时扩展控制器。在 AIME 和 HMMT 上，PersonTTS 在未见用户画像和保留问题上的联合需求满足率显著超越强 TTS 基线。相同候选评估预算下，跨用户经验复用进一步提升了策略质量并大幅降低发现智能体的时间与成本。
正文
Authors:Xinglin Wang, Zishen Liu, Tong Zheng, Shaoxiong Feng, Peiwen Yuan, Yiwei Li, Jiayi Shi, Yueqi Zhang, Chuyi Tan, Ji Zhang, Boyuan Pan, Kan Li
Abstract:Test-time scaling (TTS) improves the reasoning capabilities of large language models by allocating additional inference computation. Existing approaches to improving TTS efficiency largely optimize accuracy against one resource dimension at a time, advancing either the accuracy--cost or accuracy--latency Pareto frontier. Yet user requirements are multidimensional: users may specify accuracy, latency, and inference-cost requirements jointly, and different requirements can favor different controllers. We formulate Personalized Test-Time Scaling as discovering executable controllers that maximize the joint satisfaction rate of user-specific requirements. To reduce the overhead of repeated policy discovery for new user profiles, we propose PersonTTS, an amortized agentic policy-discovery framework that reuses prior search experience through requirement-matched controller initialization and source-distilled procedural guidance, while retaining target-profile evaluation for every candidate. Experiments on AIME and HMMT show that PersonTTS substantially outperforms strong TTS baselines in joint requirement satisfaction on unseen user profiles and held-out problems. Under the same candidate-evaluation budget, cross-user experience reuse further improves policy quality while substantially reducing discovery-agent time and cost.
| Comments: | Preprint |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.09684 [cs.CL] |
| (or arXiv:2610.09684v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09684 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xinglin Wang [view email]
[v1]
Wed, 7 Oct 2026 08:46:56 UTC (540 KB)
来源：arXiv：cs.CL · arxiv.org