论文精选 65已解读
Co-Evolving Acting and Forecasting for Language Agents· 语言代理的共同演化行动与预测
Training language agents is bottlenecked by the high cost of long-horizon rollouts, yet st…
领域：cs.AI作者：Haofei Yu、Yixing Chen、Yining Zhao
📰 AI 研究简报 · 10月16日
今日头条：Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models —— NVIDIA today released Nemotron 3.5 Super VL, an open 120-billion-parameter model that delivers the fastest task completion and highest accuracy for specialized always-on agents. Joining it are two new edge models: Nemotron 3.5 Bolt 74B-A9B 24GB, optimized to balance intelligence and speed per GB for local agentic workloads, and Nemotron 3.5 Bolt E4B 4GB, optimized […]
共 56 条预计阅读 34 分钟Token 消耗 100,800
arXiv 最新论文
论文精选 82
Latent Compaction with Test-Time Training· 测试时训练的潜在压缩
Long contexts enlarge key--value (KV) caches, increasing memory use and attention costs du…
领域：cs.CL作者：Yingfa Chen、Xingyu Shen、Chaojun Xiao
论文精选 81
Manacá-1B-Instruct-2: Reproducible Safety Alignment for a Native Brazilian-Portuguese LLM. Negative DPO Results, a Controlled Composition Scan, and a...· Manacá-1B-Instruct-2: 巴西葡萄牙语大模型的安全对齐
Brazilian Portuguese remains underserved by open-source language models, and the few that …
领域：cs.CL作者：Bruno Leonardo Santos Menezes、Carlos Leonardo Souza Cardoso、Fabio André Machado Porto
论文精选 65已解读
Scaling Discovery Requires Forgetting· 规模化发现需要忘记
Automated research with LLM agents has shown promising results in various domains. Through…
领域：cs.AI作者：Shangyin Tan、Ben Zhou、Lakshya A Agrawal
论文精选 65已解读
PhySGA: A Physics-Space Grounded Agentic Framework for Generalizable and Interactive 3D Composition· PhySGA：物理空间中的代理框架
3D object composition is pivotal to digital content creation and robot simulation, while a…
领域：cs.AI作者：Weichen Lyu、Chong Xia、Mingyu Dong
论文精选 83
PerSo: Investigating Facets of LLMs' Social Intelligence in Real-World Persuasion Scenarios· 大模型社交智能研究
Understanding the social intelligence of large language models (LLMs) is increasingly impo…
领域：cs.CL作者：Ali Zamani、Soroush Esfahanian、Erfan Shahabi
论文精选 82
PrismaCoder: Pushing the Boundaries of Open Multimodal Code Intelligence through Decomposed Rubric Synthesis· 普里玛码农：通过分解评分合成推动开放多模态代码智能的边界
Multimodal code intelligence connects the symbolic world of programs with the visual world…
领域：cs.CL作者：Qiushi Sun、Jingyang Gong、Yang Liu
论文精选 65已解读
ActiveEvidence: Decision-Focused Evidence Graphs with Verifiable Acquisition Gaps· 决策聚焦证据图与可验证获取差距
A multimodal agent must decide whether visual evidence is worth acquiring, which candidate…
领域：cs.AI作者：Miaobo Hu、Shuhao Hu、Xiaobo Guo
论文精选 65已解读
The Wealth of Agents: Scaling Agentic AI Across Developers· 代理的财富：跨开发者扩展代理AI
Today's agentic systems are typically designed, built, and deployed by single developers w…
领域：cs.AI作者：Austin W. Hanjie、Carlos E. Jimenez、Zirui Wang
论文精选 85
VASPO: Self-Distillation with Privileged Value Estimation· VASPO: 增强价值自蒸馏
On-policy self-distillation provides dense supervision for reasoning models via a privileg…
领域：cs.CL作者：Fei Wu、Zhenrong Zhang、Jianshu Zhang
论文精选 65已解读
SafeAdapt-Poker: Counterfactual Risk Calibration for Budgeted Adaptation under Opponent Shift· SafeAdapt-Poker：预算适应下的对手偏移风险校准
Opponent histories can mislead after behavioral shifts. SafeAdapt-Poker formulates budgete…
领域：cs.AI作者：Miaobo Hu、Shuhao Hu、Xiaobo Guo
论文精选 65已解读
SwarmRSI: Recursive Self-Improvement through Evolving Scientific Agent Swarm· SwarmRSI：通过进化科学代理群体实现递归自我改进
Recursive self-improvement (RSI) in scientific research aims to improve agents' research c…
领域：cs.AI作者：Qianru Meng、Xi Zhang、Konstantinos Papakostas
论文精选 65已解读
From Relevance to Validity: Zero-Shot Tool Selection via Functional and Contextual Validation· 从相关性到有效性：零样本工具选择
Large language model (LLM) agents increasingly face tasks from unseen domains and executio…
领域：cs.AI作者：Xiaoxuan Dong、Jin Zhang、Zichen Yuan
论文精选 65已解读
Lifted State Policy Optimization: Internalizing Self-Correction through Transition Geometry· 提升状态策略优化：通过转换几何内部化自校正
Preference optimization ranks final answers and trajectory RL maximizes aggregate returns,…
领域：cs.AI作者：Ren Zhuang、Ben Wang、Shuifa Sun
论文精选 65已解读
TreeFlex: Agents Navigating Their Own Search Trees· TreeFlex：代理自主导航搜索树
A ReAct-style agent accumulates observations, intermediate results, and failed attempts in…
领域：cs.AI作者：Dongzhuo Li、Fangzhou Cheng、Weiwei Li
论文精选 65已解读
Argument Graph Uncertainty: Quantifying Uncertainty from the Logical Structure of Reasoning Chains· 论证图不确定性：从推理链逻辑结构量化不确定性
Reliable uncertainty estimation is essential for deploying large language models in high-s…
领域：cs.AI作者：Simon Baur、Sumire Honda、Wojciech Samek
论文精选 65已解读
Reflect After Failure: Brain-Inspired Self-Evolving Agents for Embodied Spatial Intelligence· Reflect After Failure: 受脑启发的自我进化的具身空间智能体
Vision-Language Models (VLMs) have made substantial progress in spatial reasoning, and sel…
领域：cs.AI作者：Xin Li、Xiaohuan Gu、Yifan Wang
论文精选 85
CASI: Credit Assignment via Set Intervention for Agentic Retrieval· CASI: 代理检索的集合干预信用分配
Large language models increasingly generate multiple retrieval actions jointly, while down…
领域：cs.AI作者：Xinyi Cheng、Tianfu Zhang、Jiwen Li
论文精选 83
Design Space Exploration of Reinforcement Learning Configurations for Post-Legalization Placement· 强化学习配置的设计空间探索
Graph Neural Networks (GNNs) and Reinforcement Learning (RL) have recently been explored f…
领域：cs.AI作者：Foteini Oikonomou、Maria Pantazi-Kypraiou、George Goudroumanis
论文精选 85
Robust Counterfactuals for Markov Decision Processes under Statistical and Causal Uncertainties· 在统计和因果不确定性下稳健的MDP反事实推理
Counterfactual reasoning (asking what would have happened, had a decision been different) …
领域：cs.AI作者：Jessica Lally、Milad Kazemi、Nicola Paoletti
论文精选 82
Drivers of LLM Consensus Formation Under Competing Incentives: A Simulation Study· 竞争激励下大型语言模型共识形成的驱动因素：一项模拟研究
Large Language Models (LLMs) are increasingly deployed in multi-agent environments that pr…
领域：cs.AI作者：Isabel Jones、Matthew Edwards、Janina A. Hoffmann
论文精选 83
HEM-Med: A Hierarchical Agentic Memory Framework for Safe Multi-Center Medication Recommendation· HEM-Med: 多中心药物推荐的分层代理记忆框架
Medication recommendation (MR) is critical for clinical decision support. While existing M…
领域：cs.AI作者：Min Xiao、Jin Li、Shoujin Wang
论文精选 82
Value-Guided Repulsion for Offline Reinforcement Learning· 值引导的离线强化学习排斥机制
Drifting models have emerged as a highly efficient and stable paradigm for policy design i…
领域：cs.AI作者：Yixue Hao、Jiahao Zhang、Jiesong Lian
论文精选 83
LLM-Based Policy Compliance Verification of Loan Applications· 基于大模型的贷款政策合规性验证
Banks rely on extensive internal policy frameworks to ensure that corporate lending decisi…
领域：cs.AI作者：Florentina Voboril
论文精选 82
Maximising Transparency in Context Compaction Improves Retention of Facts and Commitments for Lightweight Coding Agents· 透明压缩提高轻量级编码代理的事实保留
Long coding agent sessions outgrow the model's context window, so agents compact: older hi…
领域：cs.AI作者：Felix Woolford
GitHub 热门开源项目
开源项目精选 75
apenasdouglas/ai-motion-library· AI 动画库
Biblioteca de Motion apenasdouglas: 24 efeitos de animação com o nome em português e o ter…
⭐ 63 stars
开源项目精选 82
tony3liu/eido· tony3liu/eido: macOS 上的 AI 代码助手
AI coding agent workspace for macOS, built on Zed and pi. Native code editing and review, …
⭐ 63 stars语言 Rust
开源项目精选 81
thesysdev/open-intelligent-ui· 开源智能UI
An open-source take on ChatGPT's Intelligent UI: answers stream as interactive UI (maps, p…
⭐ 391 stars语言 TypeScript
开源项目精选 75
sebastianhaba/book-to-course-skill· 书籍转课程技能
Claude skill that turns a book (EPUB, PDF, HTML, DOCX, Markdown) into an interactive begin…
⭐ 66 stars语言 Python
精选 AI 行业资讯
行业资讯精选 75
Mark Wahlberg on using AI in film ‘the right way’ and being open to an AI clone carrying on his legacy· 马克·沃尔伯格谈正确使用 AI 制作电影
Actor and angel investor Mark Wahlberg says he's OK with using AI in filmmaking, provided …
行业资讯精选 82
Replit’s CEO says the future of work is the ‘self-driving company’· Replit CEO：未来的工作是‘自动驾驶公司’
Replit CEO Amjad Masad envisions a future of “self-driving companies,” where AI agents han…
行业资讯精选 83
中控技术接入千问办公，打造流程工业领域AI Native标杆企业
10月16日，中控技术股份有限公司（以下简称“中控技术”）正式接入千问办公，将Agent落地营销全链路。后续，千问办公还将优化中控技术的采购、供应链、财务、人力、研发等场景工作流，…
行业资讯精选 82
Kling 4.0正式上线：10-bit HDR、30秒原生直出 刷新视频模型新基准
10月15日，可灵AI新一代视频生成模型 Kling 4.0 正式上线。 真实、可控、专业，是这一代模型升级的核心方向。Kling 4.0 聚焦高动态、强运镜的视频创作场景，在三个…
行业资讯精选 82
阿里千问上车宝马iX3
10月15日消息，在2026宝马集团中国创新日上，宝马宣布新世代BMW iX3将搭载专为中国市场打造的座舱AI能力。该能力由宝马与阿里巴巴联合开发，以阿里千问大模型为AI底座，并由…
行业资讯精选 75
无人机表演十年后，大漠大想做「会干活」的机群
即便对大漠大这家公司名字感到陌生的人，大概也看过它的无人机编队表演——7次登上央视春晚，多次出现在《新闻联播》里，但迄今为止，大漠大鲜少对外讲述自己的品牌故事。 在刚刚过去的国庆假…
行业资讯精选 75
付费用户占比超 20%，电商商家「涌入」千问办公
一双鞋还没生产出来，商家刘祥龙已经开始琢磨怎么卖了。 准备开一家3D打印鞋店时，他没有急着打样、拍摄、做详情页，而是先让AI帮忙调研市场、制作数字样品，再把商品图和详情页准备好，看…
行业资讯精选 82
5 年前的机密文档被泄露：Anthropic 的所有胜利，早在 2021 年就写好了
<img class="rich_pages wxw-img" data-aistatus="1" data-croporisrc="https://mmbiz.qpic.cn/m…
行业资讯精选 65
When it came to Cal AI’s viral success, a small army of influencers paid off· Cal AI 的病毒式成功得益于付费影响者
The company generated over $30 million in revenue in two years and was acquired by MyFitne…
行业资讯精选 82
Meta launches an open-weight AI model that runs on Macs· Meta 推出在 Mac 上运行的开放权重 AI 模型
Meta and Apple have teamed up for the launch of the social media giant's newest model, Mus…
行业资讯精选 82
Amazon devices chief Panos Panay sees a future where AI devices replace the need to use apps· 亚马逊设备主管：AI 设备将取代应用程序
Speaking at TechCrunch Disrupt, Amazon devices chief Panos Panay described a future where …
行业资讯精选 75
Anthropic is letting METR researchers look under the hood with employee-level access· METR 研究人员获准评估 Anthropic
METR will publish its own assessments of Anthropic’s risks, incidents, and practices, and …
行业资讯精选 80
Bringing GPT-Rosalind to academic research labs· GPT-Rosalind 进驻学术研究实验室
Introducing the Rosalind Academic Plan, offering eligible academic labs sponsored access t…
行业资讯精选 75
Bluesky expands its AI-powered Attie app to let anyone build websites, games, and more· Bluesky 扩展 AI 驱动的 Attie 应用
Bluesky is bringing “social vibe coding” to its experimental Attie platform, allowing user…
行业资讯精选 65
Be There for the Beginning: Celebrate the Launch of NVIDIA RTX Spark· 见证 NVIDIA RTX Spark 的发布
A new beginning for Windows PCs starts tomorrow with the launch of NVIDIA RTX Spark — one …
行业资讯精选 85
Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models· NVIDIA 扩大 Nemotron 家族
NVIDIA today released Nemotron 3.5 Super VL, an open 120-billion-parameter model that deli…
行业资讯精选 82
量子位智库发布Agent落地报告！拆解企业智能化六大断点，附全链路破局框架
Agent究竟该如何深入业务毛细血管？
行业资讯精选 65
2026 TMOT首个AI智能体走进电音现场，让听众成为音乐共创者· 2026 TMOT首个AI智能体走进电音现场
10月17日至18日，顶尖电音节The Magic of Tomorrowland 2026将在上海举行。
行业资讯精选 75
这个双十一，百灵新模型干起活来不烧心
大模型好不好用，百灵拿成品说话
行业资讯精选 75
AI时代，商学院这样培养未来商业领袖
未来十年共创1000个AI课题
行业资讯精选 75
苹果AI国内上线，阿里百度浮出水面
苹果 Apple Intelligence（苹果智能）在中国大陆的上线进入最后阶段。
行业资讯精选 80
前华为小艺大模型负责人创业，押注“主动式物理AI”
华为前大模型负责人创业，专注新型AI技术。
行业资讯精选 75
Sullivan & Cromwell brings legal expertise into agreement review
In collaboration with OpenAI, Sullivan & Cromwell developed Agreement Analyzer, an AI tool…
YouTube 热门 AI 视频
AI 视频精选 80
📖 Explore custom voice narration with Gemini 3.8 Flash TTS· 使用 Gemini 3.8 Flash TTS 探索自定义语音叙述
YouTube · Google for Developers
📺 Google for Developers▶ 135 次观看
AI 视频精选 80
Build more personal sites with Codex· 用 Codex 构建更个性化的网站
YouTube · OpenAI
📺 OpenAI▶ 4.5k 次观看
AI 视频精选 65
Turning raw hardware sensor data into music with Gemini· 用 Gemini 将硬件传感器数据转化为音乐
YouTube · Google for Developers
📺 Google for Developers▶ 291 次观看
AI 视频精选 82
GPT-Live-1.1 has new voices and better accents.· GPT-Live-1.1语音改善
YouTube · OpenAI
📺 OpenAI▶ 7.9k 次观看