00:01:19 AI的“偏科”难题：学好数理化，走遍天下真的不怕吗？

00:05:08 AI 也会“复盘”？聊聊如何让机器像高手一样思考  

00:09:19 语言的“橡皮泥”：我们如何“捏”出更智能的AI？

00:13:57 AI科学家的新玩法：它不猜答案，专找“意外”  

00:17:42 AI“长篇阅读”的秘密：如何让机器像螺旋一样思考？

本期介绍的五篇论文：

[LG] Does Math Reasoning Improve General LLM Capabilities? Understanding Transferability of LLM Reasoning  

[CMU & University of Washington & M-A-P]  

https://arxiv.org/abs/2507.00432 

---

[LG] ASTRO: Teaching Language Models to Reason by Reflecting and Backtracking In-Context  

[AI at Meta]  

https://arxiv.org/abs/2507.00417 

---

[LG] Flexible Language Modeling in Continuous Space with Transformer-based Autoregressive Flows  

[Apple]  

https://arxiv.org/abs/2507.00425 

---

[LG] Open-ended Scientific Discovery via Bayesian Surprise  

[University of Massachusetts Amherst & Allen Institute for AI]  

https://arxiv.org/abs/2507.00310 

---

[LG] HelixPipe: Efficient Distributed Training of Long Sequence Transformers with Attention Parallel Pipeline Parallelism  

[National University of Singapore]  

https://arxiv.org/abs/2507.00394