00:00:32 让AI学会“学习”：不再死记硬背的秘密 

00:04:30 给AI装个“外挂”：一个叫“记忆解码器”的U盘

00:08:23 开会吵不完？AI教你找共识

00:13:39 大海捞针？不如自己造根针

00:18:41 AI减肥记：如何让聪明的机器不说废话？

本期介绍的五篇论文：

[CL] Learning Facts at Scale with Active Reading  

[Meta FAIR]  

https://arxiv.org/abs/2508.09494  

---

[CL] Memory Decoder: A Pretrained, Plug-and-Play Memory for Large Language Models  

[Shanghai Jiao Tong University]  

https://arxiv.org/abs/2508.09874  

---

[CL] The Human-AI Hybrid Delphi Model: A Structured Framework for Context-Rich, Expert Consensus in Complex Domains  

[Google Research]  

https://arxiv.org/abs/2508.09349  

---

[LG] SYNAPSE-G: Bridging Large Language Models and Graph Learning for Rare Event Classification  

[Google Research]  

https://arxiv.org/abs/2508.09544  

---

[CL] Sample More to Think Less: Group Filtered Policy Optimization for Concise Reasoning  

[Microsoft Research]  

https://arxiv.org/abs/2508.09726