[人人能懂] AI寻宝记：从调度员、梦之队到神秘指南针

00:01:44 AI偷懒的艺术：好钢如何用在刀刃上00:05:46 组个“AI梦之队”，比单打独斗强在哪？00:10:51 AI玩“跑团”：下一个世界是如何被设计出来的？00:14:40 炼成“大模型”高手，需要计划表还是指南针？00:19:09 让机器人学会“试错”：不是瞎猜，而是高手过招今天介绍的五篇论文：[CL] Mixture-of-Recursions: Learning Dynamic Recursive Depths for Adaptive Token-Level Computation  [KAIST AI & Mila]  https://arxiv.org/abs/2507.10524  ---[LG] Fusing LLM Capabilities with Routing Data  [University of Illinois Urbana-Champaign]  https://arxiv.org/abs/2507.10540  ---[AI] Multi-Actor Generative Artificial Intelligence as a Game Engine  [Google DeepMind]  https://arxiv.org/abs/2507.08892  ---[LG] Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training  [KAIST & Seoul National University & Microsoft Research]  https://arxiv.org/abs/2507.09846  ---[LG] Behavioral Exploration: Learning to Explore via In-Context Adaptation  [UC Berkeley]  https://arxiv.org/abs/2507.09041  在小宇宙查看该单集文稿