English(EN) Bayes-Adaptive MDPs for Rollout Allocation to Enhance GRPO Training

新方法AdaRoll-GRPO通过难度感知Rollout分配优化LLM训练

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-28 04:00

研究人员开发了AdaRoll-GRPO，一种用于增强大型语言模型（LLM）的Group Relative Policy Optimization（GRPO）训练的新方法。与分配统一Rollout预算的标准GRPO不同，AdaRoll-GRPO使用贝叶斯自适应马尔可夫决策过程，根据问题难度动态调整预算。这种方法确保更具挑战性的问题得到充分关注，从而在固定的总预算内最大化训练效用。实验表明，AdaRoll-GRPO在各种推理数据集和LLM架构上始终能提高训练性能。 AI

影响通过优化对挑战性问题的资源分配来提高LLM训练效率。

排序理由该集群包含一篇详细介绍LLM训练新方法的论文。[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: 11 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇详细介绍LLM训练新方法的论文。[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: Same-day; Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Faeze Moradi Kalarde, Armin Toroghi, Nam Ho Koh, Doo Hwan Hwang, Seungjin Kim, Kee-Eung Kim, Scott Sanner · 2026-10-28 04:00

用于增强 GRPO 训练的贝叶斯自适应 MDP 滚动分配

arXiv:2610.37429v1 Announce Type: cross Abstract: Group Relative Policy Optimization (GRPO) has demonstrated strong performance in LLM post-training. However, it allocates a uniform rollout budget to every training question, ignoring variations in question difficulty. As a result…

本文如何被排名

报道来源 [1]

用于增强 GRPO 训练的贝叶斯自适应 MDP 滚动分配

相关实体

相关话题