arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2602.03773cs.LG
收藏

推理缓存:通过短期限强化学习实现长周期的持续改进

Reasoning Cache: Continual Improvement Over Long Horizons via Short-Horizon RL

Carnegie Mellon University(卡内基梅隆大学)

Ian Wu, Yuxiao Qu, Amrith Setlur, Aviral Kumar

AI总结:

本文提出RC算法,通过利用LLM的响应生成与总结能力差异,实现长周期的持续改进,实验显示其在HMMT 2025任务中性能显著提升。

摘要:

大型语言模型(LLMs)能够通过测试时适应不断解决更困难的问题,这一特性称为外推。然而,标准强化学习(RL)在固定问题分布和训练预算下运行,限制了在测试时分布偏移时的外推能力。为此,我们引入RC,一种迭代解码算法,在训练和推理过程中均取代标准自回归解码。RC利用LLM在响应生成和总结能力上的不对称性,构建持续改进的推理链。训练使用RC的模型在推理周期上比训练时看到的模型提高了数量级。实验证明,使用RC训练的4B模型在16k-token训练预算下,使HMMT 2025任务性能从40%提升到近70%,优于同等规模模型和许多更大规模的推理LLM。最后,我们还显示,训练使用RC的模型能更有效地利用现有框架进一步提升测试时性能,这是通过训练中学习的改进的总结条件生成能力实现的。

英文摘要:

Large Language Models (LLMs) that can continually improve beyond their training budgets are able to solve increasingly difficult problems by adapting at test time, a property we refer to as extrapolation. However, standard reinforcement learning (RL) operates over fixed problem distributions and training budgets, which limits extrapolation amidst distribution shift at test time. To address this, we introduce RC, an iterative decoding algorithm that replaces standard autoregressive decoding during both training and inference. RC exploits an asymmetry between the response generation and summarization capabilities of LLMs to construct reasoning chains that consistently improve across iterations. Models trained to use RC can extrapolate and continually improve over reasoning horizons more than an order of magnitude longer than those seen during training. Empirically, training a 4B model with RC using a 16k-token training budget improves performance on HMMT 2025 from 40% to nearly 70% with 0.5m tokens at test time, outperforming both comparably sized models and many larger reasoning LLMs. Finally, we also show that models trained with RC can more effectively leverage existing scaffolds to further scale test-time performance, due to the improved summary-conditioned generation abilities learned through training.

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。

↑