Event ·Official first-hand

Google Research 提出持续学习的 sleep 阶段，缓解 LLM 灾难性遗忘

Google Research 在博客介绍一种模拟睡眠阶段的持续学习方法，让 LLM 在训练后继续整合新事实并增强推理能力。该方法针对两个常见问题：灾难性遗忘和知识截止日期静态化。通过类似离线的整合阶段，模型在不破坏原有能力的情况下吸收新知识。

Current read

Google Research 发了个新方法，给 LLM 加了个类似睡觉的整合阶段，让它能持续学新知识还不忘旧的，做模型训练的可以看看。

2 sources38° AI heatLast updated 2026/10/27 17:40:42

Evidence

2 sources

Primary sourceGoogle Research

2026/10/27 17:40:42View source

Related source 1shao__meng

2026/10/29 02:34:55View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story