[人人能懂] 超越规模竞赛的五种智慧策略

00:00:35 如何让AI从“学霸”变“侦探”？00:04:35 内卷的AI界：为什么“更快”的工具，没人敢用？00:09:28 让机器人学会“举一反三”的秘密00:15:02 AI训练太烧钱？换个聪明的“刹车”法00:19:13 让机器人看懂世界，只需一副“智能眼镜”本期介绍的五篇论文：[CL] Open Data Synthesis For Deep Research  [Beijing Academy of Artificial Intelligence (BAAI)]  https://arxiv.org/abs/2509.00375  ---[LG] Fantastic Pretraining Optimizers and Where to Find Them  [Stanford University]  https://arxiv.org/abs/2509.02046  ---[RO] Data Retrieval with Importance Weights for Few-Shot Imitation Learning  [Stanford University]  https://arxiv.org/abs/2509.01657  ---[LG] GradES: Significantly Faster Training in Transformers with Gradient-Based Early Stopping  [Boston University Metropolitan College]  https://arxiv.org/abs/2509.018  ---[RO] Manipulation as in Simulation: Enabling Accurate Geometry Perception in Robots  [ByteDance Seed]  https://arxiv.org/abs/2509.02530  在小宇宙查看该单集文稿