English(EN) Freeze, Don't Clone: Fine-Tuning with Expressive Policies Without Retaining Offline Data

新的微调方法在不保留数据的情况下保持离线强化学习模型性能

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

研究人员开发了一种新的强化学习模型微调方法，该方法避免保留离线数据。这种被称为“冻结，而非克隆”的方法解决了关键的失效模式，即在移除离线数据后继续进行行为克隆会导致表示坍塌和动作多样性丧失。通过冻结预训练的表达性先验并停止在线行为拟合，该方法在不同强化学习算法的各种 OGBench 域上显著提高了性能。 AI

影响这项研究可能有助于在无法保留离线数据的场景中更有效地部署强化学习模型。

排序理由这是一篇详细介绍强化学习新方法的学术论文。[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: 12 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 这是一篇详细介绍强化学习新方法的学术论文。[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: 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) · Sami Nur Islam, Johan Obando-Ceron, Doina Precup, Aaron Courville, Pablo Samuel Castro · 2026-10-28 04:00

冻结而非克隆：在不保留离线数据的情况下使用表达性策略进行微调

arXiv:2610.36247v1 Announce Type: cross Abstract: Offline-to-online reinforcement learning uses offline pretraining to accelerate subsequent online learning. While many methods assume the offline dataset remains available throughout fine-tuning, there are many settings where this…

本文如何被排名

报道来源 [1]

冻结而非克隆：在不保留离线数据的情况下使用表达性策略进行微调

相关实体

相关话题