arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.16486cs.CLcs.AI
收藏

及时机器:时间意识使测试时间缩放成为代理

Timely Machine: Awareness of Time Makes Test-Time Scaling Agentic

Fudan University Shanghai AI Laboratory(复旦大学上海人工智能实验室);Shanghai AI Laboratory(上海人工智能实验室);The Chinese University of Hong Kong(香港中文大学)

Yichuan Ma, Linyang Li, Yongkang chen, Peiji Li, Xiaozhe Li, Qipeng Guo, Dahua Lin, Kai Chen

AI总结:

Timely Machine通过重新定义测试时间为实时时钟时间,提出Timely-RL方法提升时间预算意识,以应对高频工具调用和时间受限推理的挑战。

摘要:

随着大型语言模型(LLMs)越来越多地处理复杂推理任务,测试时间缩放已成为增强能力的关键。然而,在频繁调用工具的代理场景中,传统的基于生成长度的定义失效:工具延迟使推理时间与生成长度脱节。我们提出了Timely Machine,重新定义测试时间为实时时钟时间,其中模型根据时间预算动态调整策略。我们引入了Timely-Eval,一个涵盖高频工具调用、低频工具调用和时间受限推理的基准测试。通过改变工具延迟,我们发现较小的模型通过更多交互在快速反馈中表现优异,而较大的模型则通过更高质量的交互在高延迟设置中占据主导。此外,现有模型未能适应时间预算的推理。我们提出了Timely-RL来填补这一空白。在冷启动监督微调后,我们使用强化学习来增强时间规划。Timely-RL提高了时间预算意识,并在Timely-Eval中一致提升了性能。我们希望我们的工作为代理时代的测试时间缩放提供新的视角。

英文摘要:

As large language models (LLMs) increasingly tackle complex reasoning tasks, test-time scaling has become critical for enhancing capabilities. However, in agentic scenarios with frequent tool calls, the traditional generation-length-based definition breaks down: tool latency decouples inference time from generation length. We propose Timely Machine, redefining test-time as wall-clock time, where models dynamically adjust strategies based on time budgets. We introduce Timely-Eval, a benchmark spanning high-frequency tool calls, low-frequency tool calls, and time-constrained reasoning. By varying tool latency, we find smaller models excel with fast feedback through more interactions, while larger models dominate high-latency settings via superior interaction quality. Moreover, existing models fail to adapt reasoning to time budgets. We propose Timely-RL to address this gap. After cold-start supervised fine-tuning, we use reinforcement learning to enhance temporal planning. Timely-RL improves time budget awareness and consistently boosts performance across Timely-Eval. We hope our work offers a new perspective on test-time scaling for the agentic era.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑