本期的 15 篇论文如下:
00:24 🧠 PhysBrain: Human Egocentric Data as a Bridge from Vision Language Models to Physical Intelligence(PhysBrain:以人类第一人称数据为桥梁,从视觉语言模型迈向物理智能)
01:05 🔬 Probing Scientific General Intelligence of LLMs with Scientist-Aligned Workflows(通过科学家对齐的工作流程探究大语言模型的科学通用智能)
01:34 🧠 When Reasoning Meets Its Laws(当推理遇见其定律)
02:16 🧠 Seed-Prover 1.5: Mastering Undergraduate-Level Theorem Proving via Learning from Experience(Seed-Prover 1.5:通过经验学习掌握本科级定理证明)
03:02 🧠 4D-RGPT: Toward Region-level 4D Understanding via Perceptual Distillation(4D-RGPT:通过感知蒸馏实现区域级4D理解)
03:51 🎨 Both Semantics and Reconstruction Matter: Making Representation Encoders Ready for Text-to-Image Generation and Editing(语义与重建皆重要:让表征编码器为文本到图像生成与编辑做好准备)
04:30 ⚖ Are We on the Right Way to Assessing LLM-as-a-Judge?(我们评估LLM作为评判者的方法正确吗?)
05:05 📡 RadarGen: Automotive Radar Point Cloud Generation from Cameras(RadarGen:基于摄像头的汽车雷达点云生成)
05:54 🔬 Physics of Language Models: Part 4.1, Architecture Design and the Magic of Canon Layers(语言模型的物理学:第4.1部分,架构设计与Canon层的魔力)
06:41 🎬 HERBench: A Benchmark for Multi-Evidence Integration in Video Question Answering(HERBench:视频问答中多证据整合的基准测试)
07:26 🔍 GroundingME: Exposing the Visual Grounding Gap in MLLMs through Multi-Dimensional Evaluation(GroundingME:通过多维评估揭示MLLMs中的视觉基础能力差距)
08:06 ⚙ SWE-Bench++: A Framework for the Scalable Generation of Software Engineering Benchmarks from Open-Source Repositories(SWE-Bench++:一种从开源仓库可扩展生成软件工程基准的框架)
08:39 🧠 Turn-PPO: Turn-Level Advantage Estimation with PPO for Improved Multi-Turn RL in Agentic LLMs(Turn-PPO:基于回合级优势估计与PPO的智能体大语言模型多轮强化学习优化)
09:14 ⚡ StageVAR: Stage-Aware Acceleration for Visual Autoregressive Models(StageVAR:面向视觉自回归模型的阶段感知加速)
09:48 🤖 An Anatomy of Vision-Language-Action Models: From Modules to Milestones and Challenges(视觉-语言-动作模型剖析:从模块、里程碑到挑战)

【关注我们】
您还可以在以下平台找到我们,获得播客内容以外更多信息
小红书: AI速递
