新的强化学习方法通过密集奖励提升LLM算术推理能力

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-16 04:00

研究人员开发了一种名为Bellman-Consistent Trajectory RL的新强化学习（RL）方法，以提高LLM的算术推理能力，尤其是在处理稀疏奖励时。该框架使用Trajectory Bellman Residual Minimization (TBRM) 将稀疏奖励转换为密集的、token级别的信号。通过整合Reinforce Leave-One-Out (RLOO) 以减少梯度方差，以及目标重标记以加速学习，TBRM-RLOO方法显著提高了算术任务的性能。该方法在Countdown任务上取得了最先进的结果，将Pass@1准确率提高到0.55-0.64，Pass@16准确率提高到76.0-84.0%，优于现有基线，并展示了高单样本效率。 AI

影响增强了LLM在复杂推理任务中的能力，可能提高需要数学问题解决能力的领域的性能。

排序理由关于LLM算术推理新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 17 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 关于LLM算术推理新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, model release; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Lianfa Li · 2026-10-16 04:00

用于LLM算术推理的Bellman一致轨迹强化学习

arXiv:2610.21060v1 Announce Type: cross Abstract: To address sparse terminal rewards in LLM arithmetic reasoning, we propose a Bellman-consistent trajectory RL method. Our framework introduces Trajectory Bellman Residual Minimization (TBRM) to transform sparse rewards into dense …

本文如何被排名

报道来源 [1]

用于LLM算术推理的Bellman一致轨迹强化学习

相关实体

相关话题