【赞助商】
OpenClaw快报
每天五分钟,听听 OpenClaw 快报,带你了解最新动态和业内讨论
传送门 www.xiaoyuzhoufm.com
【目录】
本期的 15 篇论文如下:
[] 🧪 ScienceIDE: Turning World's Scientific Codebase into Agent Learnable Environments(ScienceIDE:将全球科学代码库转化为智能体可学习环境)
[] 🧠 LimiX-2: A Contextual Mechanism Network Towards General Structured-Data Intelligence(LimiX-2:面向通用结构化数据智能的上下文机制网络)
[] 📉 Rethinking Critic Learning in PPO: Understanding and Mitigating Value Flattening(重新思考PPO中的评论家学习:理解与缓解价值平坦化)
[] 🧠 Confidence Comes from Experience: Experiential Confidence Estimation from Reasoning to Agents(置信度源于经验:从推理到智能体的经验性置信度估计)
[] 🤖 ProgramDistill: From Interactive Web Apps to Verifiable Reference-Guided SWE Tasks(ProgramDistill:从交互式 Web 应用到可验证的参考引导软件工程任务)
[] 🤖 ActionPiece: Rethinking Action Tokenization for Autoregressive Vision-Language-Action Models(ActionPiece:重新思考自回归视觉-语言-动作模型的动作标记化)
[] 🧠 Agora: Git as Shared Memory for Collective AutoResearch(Agora:将 Git 作为集体自动研究的共享记忆)
[] ⚡ VC-Attention: Value Smoothing and Softmax Casting for Low-bit Attention(VC-Attention:面向低比特注意力的值平滑与 Softmax 转换)
[] 📈 EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents(EvolveTrade:面向自演化LLM交易智能体的经验驱动策略精炼)
[] 🎮 Zing-0.5: Toward Playable Worlds with Real-Time Joint Action and Text Control(Zing-0.5:迈向具备实时联合动作与文本控制的可玩世界)
[] 👀 Gaze as Evidence for Common Grounding: A Cross-Corpus Analysis of MapTask and MUNDEX(注视作为共同基础证据:MapTask与MUNDEX的跨语料库分析)
[] 🎯 A Zeroth-Order Paradigm for LLM Preference Alignment(面向LLM偏好对齐的零阶范式)
[] 🧬 HypoEvolve: Genetic Algorithms Enable Multi-Agent LLMs to Discover Scientific Hypotheses(HypoEvolve:遗传算法使多智能体大语言模型能够发现科学假设)
[] ✋ EventEgoHands++: Event-based Egocentric 3D Hand Mesh Reconstruction with Real Dataset(EventEgoHands++:基于事件的第一人称3D手部网格重建与真实数据集)
[] 🔭 SpectralShift: Effective Context Window Extension of Gated DeltaNet via Spectral Reparameterization(SpectralShift:通过谱重参数化有效扩展 Gated DeltaNet 的上下文窗口)

【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
