HuggingFace 每日AI论文速递

本期的 15 篇论文如下：[00:25] 🧠 General Agentic Memory Via Deep Research（通过深度研究的通用代理记忆）[00:52] 🧪 AutoEnv: Automated Environments for Measuring Cross-Environment Agent Learning（AutoEnv：用于跨环境智能体学习的自动化环境测量）[01:24] 🤖 Computer-Use Agents as Judges for Generative User Interface（以计算机使用代理作为生成式用户界面的评判者）[01:55] 🎨 DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generation（DeCo：用于端到端图像生成的频率解耦像素扩散）[02:24] 🎨 UltraFlux: Data-Model Co-Design for High-quality Native 4K Text-to-Image Generation across Diverse Aspect Ratios（UltraFlux：面向高质量原生4K文本到图像跨多样宽高比的数据-模型协同设计）[03:10] 🔍 DR Tulu: Reinforcement Learning with Evolving Rubrics for Deep Research（DR Tulu：基于演化评分标准的深度研究强化学习）[03:46] 🎬 In-Video Instructions: Visual Signals as Generative Control（视频内指令：视觉信号作为生成控制）[04:24] 📊 Budget-Aware Tool-Use Enables Effective Agent Scaling（预算感知的工具使用实现有效的智能体扩展）[05:12] 🎬 Plan-X: Instruct Video Generation via Semantic Planning（Plan-X：通过语义规划指导视频生成）[05:54] 🧪 M3-Bench: Multi-Modal, Multi-Hop, Multi-Threaded Tool-Using MLLM Agent Benchmark（M3-Bench：多模态、多跳、多线程工具使用MLLM智能体基准）[06:25] 🤖 Multi-Agent Deep Research: Training Multi-Agent Systems with M-GRPO（多智能体深度研究：使用M-GRPO训练多智能体系统）[07:24] 🎬 HunyuanVideo 1.5 Technical Report（混元视频1.5技术报告）[07:56] 🧠 Chain-of-Visual-Thought: Teaching VLMs to See and Think Better with Continuous Visual Tokens（视觉思维链：通过连续视觉标记教导视觉语言模型更好地观察与思考）[08:36] 🧠 MIST: Mutual Information Via Supervised Training（MIST：通过监督训练实现互信息估计）[09:07] 🎨 Controllable Layer Decomposition for Reversible Multi-Layer Image Generation（可控层分解用于可逆多层图像生成）【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

本期的 15 篇论文如下：[00:21] 🧠 OpenMMReasoner: Pushing the Frontiers for Multimodal Reasoning with an Open and General Recipe（OpenMMReasoner：以开放通用方案推动多模态推理前沿）[01:04] 🌍 GeoVista: Web-Augmented Agentic Visual Reasoning for Geolocalization（GeoVista：用于地理定位的Web增强智能视觉推理）[01:41] 🎯 SAM 3: Segment Anything with Concepts（SAM 3：基于概念的通用分割模型）[02:31] 📊 Unveiling Intrinsic Dimension of Texts: from Academic Abstract to Creative Story（揭示文本的内在维度：从学术摘要到创意故事）[03:09] 🧠 O-Mem: Omni Memory System for Personalized, Long Horizon, Self-Evolving Agents（O-Mem：面向个性化、长周期、自进化智能体的全能记忆系统）[03:43] 🦜 Parrot: Persuasion and Agreement Robustness Rating of Output Truth -- A Sycophancy Robustness Benchmark for LLMs（鹦鹉：输出真相的说服与一致性鲁棒性评级——一个面向大语言模型的谄媚鲁棒性基准）[04:26] 🧠 RynnVLA-002: A Unified Vision-Language-Action and World Model（RynnVLA-002：统一的视觉-语言-动作与世界模型）[05:19] 🧠 VisMem: Latent Vision Memory Unlocks Potential of Vision-Language Models（VisMem：潜在视觉记忆解锁视觉语言模型潜力）[05:51] 🌍 WorldGen: From Text to Traversable and Interactive 3D Worlds（WorldGen：从文本到可遍历交互式3D世界）[06:34] 🎨 Loomis Painter: Reconstructing the Painting Process（Loomis Painter：重建绘画过程）[07:06] 🔮 Mantis: A Versatile Vision-Language-Action Model with Disentangled Visual Foresight（Mantis：具有解耦视觉预测能力的多功能视觉-语言-动作模型）[07:48] 🎨 InstructMix2Mix: Consistent Sparse-View Editing Through Multi-View Model Personalization（InstructMix2Mix：通过多视图模型个性化实现一致的稀疏视图编辑）[08:21] 🔬 OmniScientist: Toward a Co-evolving Ecosystem of Human and AI Scientists（全能科学家：迈向人类与AI科学家共同进化的生态系统）[09:07] 🧬 MergeDNA: Context-aware Genome Modeling with Dynamic Tokenization through Token Merging（MergeDNA：基于动态标记化的上下文感知基因组建模）[09:41] 🔍 Video-R4: Reinforcing Text-Rich Video Reasoning with Visual Rumination（Video-R4：通过视觉反刍增强文本丰富视频推理）【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

本期的 5 篇论文如下：[00:41] TOP1(🔥171) | 🎨 Kandinsky 5.0: A Family of Foundation Models for Image and Video Generation（Kandinsky 5.0：用于图像和视频生成的基础模型家族）[02:02] TOP2(🔥150) | 🚀 MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling（MiroThinker：通过模型、上下文与交互扩展，将开源研究智能体性能推向新边界）[04:31] TOP3(🔥127) | 🏅 P1: Mastering Physics Olympiads with Reinforcement Learning（用强化学习攻克物理奥赛）[06:43] TOP4(🔥126) | 🍲 Souper-Model: How Simple Arithmetic Unlocks State-of-the-Art LLM Performance（“汤”级模型：简单加权平均即可让大语言模型性能跃升）[08:09] TOP5(🔥104) | 🧠 VIDEOP2R: Video Understanding from Perception to Reasoning（VIDEOP2R：从感知到推理的视频理解）【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

本期的 15 篇论文如下：[00:22] 📊 V-ReasonBench: Toward Unified Reasoning Benchmark Suite for Video Generation Models（V-ReasonBench：面向视频生成模型的统一推理基准套件）[01:06] 🧠 Step-Audio-R1 Technical Report（Step-Audio-R1技术报告）[01:48] 🧭 Scaling Spatial Intelligence with Multimodal Foundation Models（通过多模态基础模型扩展空间智能）[02:18] 🎬 First Frame Is the Place to Go for Video Content Customization（首帧是实现视频内容定制化的关键所在）[02:49] 🎬 Video-as-Answer: Predict and Generate Next Video Event with Joint-GRPO（视频即答案：使用联合GRPO预测并生成下一视频事件）[03:29] 🔮 SAM 3D: 3Dfy Anything in Images（SAM 3D：图像中任意物体的三维化）[04:03] 🚀 MiMo-Embodied: X-Embodied Foundation Model Technical Report（MiMo-Embodied：跨具身基础模型技术报告）[04:38] 🧠 Thinking-while-Generating: Interleaving Textual Reasoning throughout Visual Generation（边生成边思考：在视觉生成中交织文本推理）[05:10] 🏆 TurkColBERT: A Benchmark of Dense and Late-Interaction Models for Turkish Information Retrieval（TurkColBERT：土耳其信息检索中稠密与延迟交互模型的基准研究）[05:53] 🌀 Nemotron Elastic: Towards Efficient Many-in-One Reasoning LLMs（Nemotron Elastic：迈向高效多合一推理大语言模型）[06:26] 🚀 SRPO: Self-Referential Policy Optimization for Vision-Language-Action Models（自参考策略优化：面向视觉-语言-动作模型）[07:09] 🎬 TimeViper: A Hybrid Mamba-Transformer Vision-Language Model for Efficient Long Video Understanding（TimeViper：一种用于高效长视频理解的混合Mamba-Transformer视觉语言模型）[07:46] 🔬 SAM2S: Segment Anything in Surgical Videos via Semantic Long-term Tracking（SAM2S：通过语义长期跟踪实现手术视频中的任意分割）[08:23] 🎨 NaTex: Seamless Texture Generation as Latent Color Diffusion（NaTex：作为潜在颜色扩散的无缝纹理生成）[08:58] 📐 PartUV: Part-Based UV Unwrapping of 3D Meshes（PartUV：基于部件分割的3D网格UV展开方法）【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

本期的 4 篇论文如下：[00:23] 🎬 Reasoning via Video: The First Evaluation of Video Models' Reasoning Abilities through Maze-Solving Tasks（通过视频进行推理：基于走迷宫任务对视频模型推理能力的首次评测）[01:17] 🔄 VisPlay: Self-Evolving Vision-Language Models from Images（VisPlay：基于无标注图像自我进化的视觉-语言模型）[01:54] 📚 ARC-Chapter: Structuring Hour-Long Videos into Navigable Chapters and Hierarchical Summaries（ARC-Chapter：将超长视频结构化导航章节与分层摘要）[02:45] 🦴 MHR: Momentum Human Rig（MHR：动量人体绑定模型）【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

本期的 11 篇论文如下：[00:23] 🧠 Can World Simulators Reason? Gen-ViRe: A Generative Visual Reasoning Benchmark（世界模拟器会推理吗？Gen-ViRe生成式视觉推理基准）[01:03] 🕵 MVI-Bench: A Comprehensive Benchmark for Evaluating Robustness to Misleading Visual Inputs in LVLMs（MVI-Bench：评估大型视觉语言模型对误导性视觉输入鲁棒性的综合基准）[01:49] 🎞 REVISOR: Beyond Textual Reflection, Towards Multimodal Introspective Reasoning in Long-Form Video Understanding（REVISOR：超越文本反思，迈向长视频理解中的多模态内省推理）[03:02] 🧪 ATLAS: A High-Difficulty, Multidisciplinary Benchmark for Frontier Scientific Reasoning（ATLAS：面向通用人工智能的高难度跨学科科学推理基准）[03:43] 🔍 Large Language Models Meet Extreme Multi-label Classification: Scaling and Multi-modal Framework（大语言模型遇上极端多标签分类：可扩展多模态框架）[04:16] 🤖 Agent-R1: Training Powerful LLM Agents with End-to-End Reinforcement Learning（Agent-R1：以端到端强化学习训练强大语言模型智能体）[05:02] 🤖 Orion: A Unified Visual Agent for Multimodal Perception, Advanced Visual Reasoning and Execution（Orion：统一视觉智能体，实现多模态感知、高级视觉推理与执行）[05:32] ⚖ Mitigating Label Length Bias in Large Language Models（缓解大语言模型中的标签长度偏差）[06:14] 🧠 Agent READMEs: An Empirical Study of Context Files for Agentic Coding（智能体README：面向代理编程的上下文文件实证研究）[06:49] 🎧 Proactive Hearing Assistants that Isolate Egocentric Conversations（主动式听力助手：以自我为中心的对话自动分离技术）[07:20] 🎯 Error-Driven Scene Editing for 3D Grounding in Large Language Models（面向3D大模型的误差驱动场景编辑实现精准视觉定位）【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

本期的 14 篇论文如下：[00:17] 🏅 P1: Mastering Physics Olympiads with Reinforcement Learning（用强化学习攻克物理奥赛）[00:56] 🌐 Uni-MoE-2.0-Omni: Scaling Language-Centric Omnimodal Large Model with Advanced MoE, Training and Data（Uni-MoE 2.0 Omni：以语言为中心的万模态大模型，通过先进MoE、训练与数据实现规模跃升）[01:42] 🧩 Part-X-MLLM: Part-aware 3D Multimodal Large Language Model（Part-X-MLLM：面向部件感知的3D多模态大语言模型）[02:22] 🧠 TiViBench: Benchmarking Think-in-Video Reasoning for Video Generative Models（TiViBench：视频生成模型思维推理基准测试）[03:08] 🚀 GroupRank: A Groupwise Reranking Paradigm Driven by Reinforcement Learning（GroupRank：一种由强化学习驱动的分组重排范式）[03:49] 🧩 PhysX-Anything: Simulation-Ready Physical 3D Assets from Single Image（PhysX-Anything：单张图像生成可仿真物理3D资产）[04:28] 🌌 UFO$^3$: Weaving the Digital Agent Galaxy（UFO³：编织数字智能体银河）[04:59] 🍲 Souper-Model: How Simple Arithmetic Unlocks State-of-the-Art LLM Performance（“汤”级模型：简单加权平均即可让大语言模型性能跃升）[05:38] 🌍 OlmoEarth: Stable Latent Image Modeling for Multimodal Earth Observation（OlmoEarth：面向多模态地球观测的稳定潜变量图像建模）[06:19] 🔄 Live-SWE-agent: Can Software Engineering Agents Self-Evolve on the Fly?（Live-SWE-agent：软件工程智能体能否实时自我进化？）[06:51] 🚀 MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling（MiroThinker：通过模型、上下文与交互扩展，将开源研究智能体性能推向新边界）[07:36] 🎯 Test-Time Spectrum-Aware Latent Steering for Zero-Shot Generalization in Vision-Language Models（测试时谱感知潜变量引导实现视觉-语言模型零样本泛化）[08:19] 🧠 WebCoach: Self-Evolving Web Agents with Cross-Session Memory Guidance（WebCoach：具备跨会话记忆引导的自进化网页智能体）[09:10] 🧬 Evolve the Method, Not the Prompts: Evolutionary Synthesis of Jailbreak Attacks on LLMs（进化方法而非提示：面向大模型的越狱攻击演化合成）【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

本期的 13 篇论文如下：[00:24] 🧹 DoPE: Denoising Rotary Position Embedding（DoPE：面向旋转位置嵌入的去噪处理）[00:58] 🧪 AIonopedia: an LLM agent orchestrating multimodal learning for ionic liquid discovery（AIonopedia：面向离子液体发现的LLM智能体多模态学习编排）[01:44] 🖼 UI2Code^N: A Visual Language Model for Test-Time Scalable Interactive UI-to-Code Generation（UI2Code^N：面向测试时可扩展交互式UI转代码生成的视觉语言模型）[02:20] 🚀 Virtual Width Networks（虚拟宽度网络）[02:56] ⚡ LiteAttention: A Temporal Sparse Attention for Diffusion Transformers（LiteAttention：面向扩散Transformer的时序稀疏注意力机制）[03:32] 🌐 Simulating the Visual World with Artificial Intelligence: A Roadmap（用人工智能模拟视觉世界：路线图）[04:12] 📐 GGBench: A Geometric Generative Reasoning Benchmark for Unified Multimodal Models（GGBench：面向统一多模态模型的几何生成推理基准）[05:00] 🧏 HI-TransPA: Hearing Impairments Translation Personal Assistant（HI-TransPA：面向听障者的语音-唇形翻译个人助手）[05:35] 🚀 MarsRL: Advancing Multi-Agent Reasoning System via Reinforcement Learning with Agentic Pipeline Parallelism（MarsRL：基于智能体流水线并行强化学习的多智能体推理系统进阶研究）[06:38] 🎭 EmoVid: A Multimodal Emotion Video Dataset for Emotion-Centric Video Understanding and Generation（EmoVid：面向情感中心视频理解与生成的大规模多模态情感视频数据集）[07:18] 🧭 SpatialThinker: Reinforcing 3D Reasoning in Multimodal LLMs via Spatial Rewards（SpatialThinker：用空间奖励强化多模态大模型的3D推理）[07:55] 📊 Workload Schedulers -- Genesis, Algorithms and Differences（工作负载调度器——起源、算法与差异）[08:51] 🚗 CATS-V2V: A Real-World Vehicle-to-Vehicle Cooperative Perception Dataset with Complex Adverse Traffic Scenarios（CATS-V2V：面向复杂恶劣交通场景的真实车车协同感知数据集）【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

本期的 5 篇论文如下：[00:38] TOP1(🔥135) | 🌍 Lumine: An Open Recipe for Building Generalist Agents in 3D Open Worlds（Lumine：在3D开放世界中打造通才智能体的开源方案）[02:47] TOP2(🔥97) | 🖥 Grounding Computer Use Agents on Human Demonstrations（基于人类演示的计算机使用智能体定位研究）[04:44] TOP3(🔥89) | 🧠 Tiny Model, Big Logic: Diversity-Driven Optimization Elicits Large-Model Reasoning Ability in VibeThinker-1.5B（小模型大逻辑：多样性驱动优化唤醒VibeThinker-1.5B的大模型推理力）[06:33] TOP4(🔥84) | 🧠 HaluMem: Evaluating Hallucinations in Memory Systems of Agents（HaluMem：智能体记忆系统幻觉评估基准）[08:56] TOP5(🔥67) | 🧩 IterResearch: Rethinking Long-Horizon Agents via Markovian State Reconstruction（IterResearch：基于马尔可夫状态重构的长程智能体再思考）【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

本期的 4 篇论文如下：[00:24] 🎬 UniVA: Universal Video Agent towards Open-Source Next-Generation Video Generalist（UniVA：面向开源下一代视频通才的通用视频智能体）[00:59] 🌐 Depth Anything 3: Recovering the Visual Space from Any Views（Depth Anything 3：从任意视角恢复视觉空间）[01:50] 🔍 AlphaResearch: Accelerating New Algorithm Discovery with Language Models（AlphaResearch：用语言模型加速全新算法发现）[02:21] 🔍 MuSc-V2: Zero-Shot Multimodal Industrial Anomaly Classification and Segmentation with Mutual Scoring of Unlabeled Samples（MuSc-V2：无需标注样本的零样本多模态工业异常分类与分割）【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

本期的 9 篇论文如下：[00:19] 🌍 Lumine: An Open Recipe for Building Generalist Agents in 3D Open Worlds（Lumine：在3D开放世界中打造通才智能体的开源方案）[00:54] 🎬 Time-to-Move: Training-Free Motion Controlled Video Generation via Dual-Clock Denoising（Time-to-Move：无需训练的双时钟去噪运动控制视频生成）[01:31] ⚡ TiDAR: Think in Diffusion, Talk in Autoregression（TiDAR：扩散式思考，自回归式表达）[02:15] 🔄 LoopTool: Closing the Data-Training Loop for Robust LLM Tool Calls（LoopTool：闭合数据-训练循环，铸就鲁棒LLM工具调用）[02:51] 🤖 WMPO: World Model-based Policy Optimization for Vision-Language-Action Models（基于世界模型的视觉-语言-动作策略优化）[03:33] 🖥 WebVIA: A Web-based Vision-Language Agentic Framework for Interactive and Verifiable UI-to-Code Generation（WebVIA：可交互可验证的网页端视觉-语言智能体UI代码生成框架）[04:19] 🎯 Toward the Frontiers of Reliable Diffusion Sampling via Adversarial Sinkhorn Attention Guidance（迈向对抗式Sinkhorn注意力引导的可靠扩散采样新前沿）[04:55] 🤖 Agentic Refactoring: An Empirical Study of AI Coding Agents（智能体重构：AI编程智能体的大规模实证研究）[05:31] 🛡 Stemming Hallucination in Language Models Using a Licensing Oracle（利用许可证预言机遏制语言模型幻觉）【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

本期的 9 篇论文如下：[00:24] 🧠 Tiny Model, Big Logic: Diversity-Driven Optimization Elicits Large-Model Reasoning Ability in VibeThinker-1.5B（小模型大逻辑：多样性驱动优化唤醒VibeThinker-1.5B的大模型推理力）[00:59] 🤝 Adaptive Multi-Agent Response Refinement in Conversational Systems（对话系统中自适应多智能体响应精炼机制）[01:30] 🧩 Wasm: A Pipeline for Constructing Structured Arabic Interleaved Multimodal Corpora（Wasm：构建结构化阿拉伯交错型多模态语料的流水线）[02:17] ⚡ KLASS: KL-Guided Fast Inference in Masked Diffusion Models（KLASS：基于KL散度引导的掩码扩散模型快速采样）[02:53] 🖥 Grounding Computer Use Agents on Human Demonstrations（基于人类演示的计算机使用智能体定位研究）[03:37] 🎥 VideoSSR: Video Self-Supervised Reinforcement Learning（VideoSSR：视频自监督强化学习）[04:19] 🚪 The Path Not Taken: RLVR Provably Learns Off the Principals（未被选择的路径：RLVR确实沿非主方向学习）[05:14] 🔗 BiCA: Effective Biomedical Dense Retrieval with Citation-Aware Hard Negatives（BiCA：面向引文感知难负样本的生物医学稠密检索）[05:56] 🤹 Walking the Tightrope of LLMs for Software Development: A Practitioners' Perspective（游走于大型语言模型的钢丝绳——开发者视角的平衡之道）【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

本期的 13 篇论文如下：[00:25] 🧩 IterResearch: Rethinking Long-Horizon Agents via Markovian State Reconstruction（IterResearch：基于马尔可夫状态重构的长程智能体再思考）[01:16] 🏆 DRIVE: Data Curation Best Practices for Reinforcement Learning with Verifiable Reward in Competitive Code Generation（DRIVE：面向可验证奖励强化学习的竞赛级代码生成数据精选最佳实践）[02:03] 🔬 The Station: An Open-World Environment for AI-Driven Discovery（“站”：面向AI驱动科学发现的开放世界环境）[02:43] 🚀 RedOne 2.0: Rethinking Domain-specific LLM Post-Training in Social Networking Services（RedOne 2.0：社交网络场景下领域大模型后训练新范式）[03:15] 🧠 SofT-GRPO: Surpassing Discrete-Token LLM Reinforcement Learning via Gumbel-Reparameterized Soft-Thinking Policy Optimization（SofT-GRPO：用Gumbel重参数化软思考策略优化让离散Token强化学习望尘莫及）[03:53] 🧭 Routing Manifold Alignment Improves Generalization of Mixture-of-Experts LLMs（路由流形对齐提升混合专家大语言模型的泛化能力）[04:30] 🔍 Reasoning with Confidence: Efficient Verification of LLM Reasoning Steps via Uncertainty Heads（以置信度推理：通过不确定性头高效验证大模型推理步骤）[05:10] 🎬 MVU-Eval: Towards Multi-Video Understanding Evaluation for Multimodal LLMs（MVU-Eval：面向多模态大模型的多视频理解评测基准）[05:50] 🎨 MPJudge: Towards Perceptual Assessment of Music-Induced Paintings（MPJudge：面向音乐诱发绘画的感知评估）[06:57] 🔄 RLoop: An Self-Improving Framework for Reinforcement Learning with Iterative Policy Initialization（RLoop：一种通过迭代策略初始化自我提升的强化学习框架）[07:36] 🤖 Robot Learning from a Physical World Model（基于物理世界模型的机器人学习）[08:21] 🛠 NURBGen: High-Fidelity Text-to-CAD Generation through LLM-Driven NURBS Modeling（NURBGen：基于大模型驱动NURBS建模的高保真文本转CAD生成）[08:52] 🚀 SWE-fficiency: Can Language Models Optimize Real-World Repositories on Real Workloads?（SWE-fficiency：语言模型能否在真实工作负载下优化真实仓库性能？）【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

本期的 7 篇论文如下：[00:21] 🧠 DeepEyesV2: Toward Agentic Multimodal Model（DeepEyesV2：迈向智能体多模态模型）[01:13] 🧭 Visual Spatial Tuning（视觉空间调优）[01:54] 🦹 Too Good to be Bad: On the Failure of LLMs to Role-Play Villains（过于完美以致无法邪恶：大语言模型反派角色扮演的失败）[02:27] 🧠 Towards Mitigating Hallucinations in Large Vision-Language Models by Refining Textual Embeddings（通过精炼文本嵌入减轻大型视觉-语言模型中的幻觉）[03:13] 🪡 Jailbreaking in the Haystack（干草堆中的越狱攻击）[03:48] 🎯 CritiCal: Can Critique Help LLM Uncertainty or Confidence Calibration?（CritiCal：语言批判能否校准大模型置信度？）[04:23] 🏃 Dense Motion Captioning（密集动作字幕生成）【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

本期的 5 篇论文如下：[00:31] TOP1(🔥137) | 🎬 Thinking with Video: Video Generation as a Promising Multimodal Reasoning Paradigm（用视频思考：视频生成作为统一多模态推理新范式）[02:43] TOP2(🔥95) | 🖼 VCode: a Multimodal Coding Benchmark with SVG as Symbolic Visual Representation（VCode：以SVG为符号视觉表征的多模态代码评测基准）[05:12] TOP3(🔥90) | 🚀 Diffusion Language Models are Super Data Learners（扩散语言模型是超级数据学习者）[07:18] TOP4(🔥88) | 👁 Don't Blind Your VLA: Aligning Visual Representations for OOD Generalization（别让VLA变盲：对齐视觉表征实现分布外泛化）[09:24] TOP5(🔥79) | 🧠 Every Activation Boosted: Scaling General Reasoner to 1 Trillion Open Language Foundation（全激活赋能：将通用推理模型扩展到万亿参数的开放语言基座）【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