HuggingFace 每日AI论文速递

2026.01.13 | VideoDR让模型边搜边推理；BabyVision揭视觉短板

13 Jan 2026

Contributed by Lukas

本期的 15 篇论文如下：[00:20] 🔍 Watching, Reasoning, and Searching: A Video Deep Research Benchmark on Open Web for Agentic Video Reasonin...

2026.01.12 | 地图AI强化寻位；多模态Lean形式化

12 Jan 2026

Contributed by Lukas

本期的 15 篇论文如下：[00:20] 🗺 Thinking with Map: Reinforced Parallel Map-Augmented Agent for Geolocalization（借助地图思考：用于...

【周末特辑】1月第2周最火AI论文 | GDPO分灶吃饭稳优化；NeoVerse单目视频建4D

11 Jan 2026

Contributed by Lukas

本期的 5 篇论文如下：[00:39] TOP1(🔥126) | 📈 GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimiza...

2026.01.09 | GDPO解耦奖励优化多任务；可学习乘数解锁矩阵尺度

09 Jan 2026

Contributed by Lukas

本期的 15 篇论文如下：[00:21] 📈 GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization（GDPO：面...

2026.01.08 | 熵加权微调保旧学；演化技能网络不断进阶

08 Jan 2026

Contributed by Lukas

本期的 15 篇论文如下：[00:21] ⚖ Entropy-Adaptive Fine-Tuning: Resolving Confident Conflicts to Mitigate Forgetting（熵自适应微调：解...

2026.01.07 | 无限深度任意采样；端到端语音转录分离

07 Jan 2026

Contributed by Lukas

本期的 15 篇论文如下：[00:25] 🔍 InfiniDepth: Arbitrary-Resolution and Fine-Grained Depth Estimation with Neural Implicit Fields（InfiniDe...

2026.01.06 | K-EXAONE MoE；NextFlow统一序列建模多模态

06 Jan 2026

Contributed by Lukas

本期的 15 篇论文如下：[00:21] 🧠 K-EXAONE Technical Report（K-EXAONE技术报告）[00:56] 🚀 NextFlow: Unified Sequential Modeling Acti...

2026.01.05 | Agent流水线提速；4D建模平民化

05 Jan 2026

Contributed by Lukas

本期的 12 篇论文如下：[00:22] 🤖 Youtu-Agent: Scaling Agent Productivity with Automated Generation and Hybrid Policy Optimization（Youtu-A...

【周末特辑】1月第1周最火AI论文 | mHC 稳梯度；思维景观 RAG 读长文

03 Jan 2026

Contributed by Lukas

本期的 5 篇论文如下：[00:33] TOP1(🔥132) | 🧠 mHC: Manifold-Constrained Hyper-Connections（mHC：流形约束的超连接）[02:32] TOP2...

2026.01.02 | 语义密度压缩；扩散边画边想

02 Jan 2026

Contributed by Lukas

本期的 3 篇论文如下：[00:19] 🧠 Dynamic Large Concept Models: Latent Reasoning in an Adaptive Semantic Space（动态大型概念模型：自...

2026.01.01 | 小模型也能原生外挂；30B-MoE智体逼近大模型

01 Jan 2026

Contributed by Lukas

本期的 15 篇论文如下：[00:22] 🚀 Youtu-LLM: Unlocking the Native Agentic Potential for Lightweight Large Language Models（Youtu-LLM：解锁...

【月末特辑】12月最火AI论文 | 代码智能全链路落地；开源模型推理代理双突破

01 Jan 2026

Contributed by Lukas

本期的 10 篇论文如下：[00:29] TOP1(🔥279) | 🧠 From Code Foundation Models to Agents and Applications: A Practical Guide to Code Intellig...

2025.12.31 | 粗模精雕UltraShape；涂鸦编辑DreamOmni3

31 Dec 2025

Contributed by Lukas

本期的 6 篇论文如下：[00:24] 🧊 UltraShape 1.0: High-Fidelity 3D Shape Generation via Scalable Geometric Refinement（UltraShape 1.0：通过...

2025.12.30 | ERC耦合路由与专家；LiveTalk实时视频对话

30 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:24] 🔗 Coupling Experts and Routers in Mixture-of-Experts via an Auxiliary Loss（通过辅助损失耦合专家混...

2025.12.29 | 鸟瞰式检索提效小模型；4D扩散一键插入逼真物体

29 Dec 2025

Contributed by Lukas

本期的 13 篇论文如下：[00:27] 🧠 Mindscape-Aware Retrieval Augmented Generation for Improved Long Context Understanding（面向提升长文...

【周末特辑】12月第5周最火AI论文 | DataFlow炼数工厂上线；AI科学家跑不完闭环

27 Dec 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:42] TOP1(🔥188) | ⚙ DataFlow: An LLM-Driven Framework for Unified Data Preparation and Workflow Automation in th...

2025.12.26 | 暗号token涨点视觉推理；3D便签本让视频长脑子

26 Dec 2025

Contributed by Lukas

本期的 6 篇论文如下：[00:19] 🧠 Latent Implicit Visual Reasoning（潜在隐式视觉推理）[00:56] 🎬 Spatia: Video Generation with Up...

2025.12.25 | 四维动态理解刷新VLM；单卡200倍速生成高清视频

25 Dec 2025

Contributed by Lukas

本期的 14 篇论文如下：[00:20] 🧠 Learning to Reason in 4D: Dynamic Spatial Understanding for Vision Language Models（学习在四维空间...

2025.12.24 | 语义蓝图提速视频生成；逐层剖析炼出强策略

24 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:19] 🎬 SemanticGen: Video Generation in Semantic Space（SemanticGen：在语义空间中的视频生成）[01:01...

2025.12.23 | 数据工厂提效；棱镜假说统合

23 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:22] ⚙ DataFlow: An LLM-Driven Framework for Unified Data Preparation and Workflow Automation in the Era of Data-C...

2025.12.22 | PhysBrain用第一人称视频让AI学会动手；大模型离科学家AI还差得远

22 Dec 2025

Contributed by Lukas

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（PhysBr...

【周末特辑】12月第4周最火AI论文 | 全能生成Kling-Omni秒出4K影片；Step-GUI让手机代理本地跑

20 Dec 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:37] TOP1(🔥117) | 🎬 Kling-Omni Technical Report（Kling-Omni技术报告）[02:55] TOP2(🔥116) | 🤖 Step-GU...

2025.12.19 | Kling-Omni一统视频生成；LLaDA2.0百亿扩散模型

19 Dec 2025

Contributed by Lukas

本期的 14 篇论文如下：[00:26] 🎬 Kling-Omni Technical Report（Kling-Omni技术报告）[01:02] 🚀 LLaDA2.0: Scaling Up Diffusion Languag...

2025.12.18 | 校准步长奖励砍成本；扩散草稿自回归验证提速

18 Dec 2025

Contributed by Lukas

本期的 14 篇论文如下：[00:25] 🤖 Step-GUI Technical Report（Step-GUI技术报告）[00:59] ⚡ DEER: Draft with Diffusion, Verify with Aut...

2025.12.17 | MMGR揭多模态推理短板；WorldPlay保几何一致实时建模

17 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:23] 🧠 MMGR: Multi-Modal Generative Reasoning（MMGR：多模态生成式推理评估与基准）[01:14] 🎮 Wor...

2025.12.16 | 代理记忆三维框架；VTP刷新生成纪录

16 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:20] 🧠 Memory in the Age of AI Agents（人工智能代理时代下的记忆）[00:57] 🚀 Towards Scalable Pre-...

2025.12.15 | 牙科小模型逆袭；扩散模型弃VAE

15 Dec 2025

Contributed by Lukas

本期的 14 篇论文如下：[00:22] 🦷 DentalGPT: Incentivizing Multimodal Complex Reasoning in Dentistry（DentalGPT：激励牙科领域多模态...

【周末特辑】12月第3周最火AI论文 | 潜轨迹制导视频运动；并行自蒸馏提速推理

13 Dec 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:30] TOP1(🔥117) | 🎬 Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance（Wan-Move：...

2025.12.12 | RL捏3D新纪录；AI奥赛摘银牌

12 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:25] 🤖 Are We Ready for RL in Text-to-3D Generation? A Progressive Investigation（我们准备好将强化学习...

2025.12.11 | StereoWorld单目秒变立体大片；BiCo跨域拼贴新概念

11 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:22] 🎥 StereoWorld: Geometry-Aware Monocular-to-Stereo Video Generation（StereoWorld：几何感知的单目到立...

2025.12.10 | 潜在轨迹控运动；WebGPU实时溅射

10 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:24] 🎬 Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance（Wan-Move：通过潜在轨...

2025.12.09 | 并行自蒸馏提速4.6倍；虚部RoPE++长文本双优化

09 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:20] ⚡ Native Parallel Reasoner: Reasoning in Parallelism via Self-Distilled Reinforcement Learning（原生并行...

2025.12.08 | 自对抗一步生成；外挂评审迭代编辑

08 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:19] ⚡ TwinFlow: Realizing One-step Generation on Large Models with Self-adversarial Flows（TwinFlow：基于自对...

【周末特辑】12月第2周最火AI论文 | 代码智能全链路拆解；开源DeepSeek-V3.2登顶

07 Dec 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:32] TOP1(🔥239) | 🧠 From Code Foundation Models to Agents and Applications: A Practical Guide to Code Intellige...

2025.12.05 | DAComp立Agent新靶；流式化身无限实时

05 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:22] 📊 DAComp: Benchmarking Data Agents across the Full Data Intelligence Lifecycle（DAComp：跨全数据智能...

2025.12.04 | Qwen3-VL多模态超长上下文；PretrainZero强化主动预训练

04 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:24] 🧠 Qwen3-VL Technical Report（Qwen3-VL 技术报告）[00:57] 🧠 PretrainZero: Reinforcement Active Pretra...

【月末特辑】11月最火AI论文 | Kandinsky 5.0全家桶开源；视频生成让模型边播边想

03 Dec 2025

Contributed by Lukas

本期的 10 篇论文如下：[00:35] TOP1(🔥219) | 🎨 Kandinsky 5.0: A Family of Foundation Models for Image and Video Generation（Kandinsky 5....

2025.12.02 | 代码智能四步落地；LongVT长视频精准理解

02 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:20] 🧠 From Code Foundation Models to Agents and Applications: A Practical Guide to Code Intelligence（从代码...

2025.12.01 | Z-Image小参高效夺冠；REASONEDIT先思后画登顶

01 Dec 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:26] 🚀 Z-Image: An Efficient Image Generation Foundation Model with Single-Stream Diffusion Transformer（Z-Image...

【周末特辑】11月第5周最火AI论文 | 自适应正交稳训练；GAM代理即搜忆

29 Nov 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:51] TOP1(🔥161) | ⚡ ROOT: Robust Orthogonalized Optimizer for Neural Network Training（ROOT：面向神经网络...

2025.11.28 | 潜在奖励模型提速降显存；画布多模态生成碾压SOTA

28 Nov 2025

Contributed by Lukas

本期的 6 篇论文如下：[00:19] 🎬 Video Generation Models Are Good Latent Reward Models（视频生成模型是优秀的潜在奖励模型）...

2025.11.27 | 俄语多模态评测补空白；潜协作提速14%

27 Nov 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:22] 🔍 Multimodal Evaluation of Russian-language Architectures（俄语多模态架构的评估框架）[01:15] 🧠...

2025.11.26 | 大模型育种进化框架开源；MedSAM-3听懂临床精准分割

26 Nov 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:17] 🧬 GigaEvo: An Open Source Optimization Framework Powered By LLMs And Evolution Algorithms（GigaEvo：基于...

2025.11.25 | 即时编译让记忆无损；AutoEnv自动挑环境提两成

25 Nov 2025

Contributed by Lukas

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:...

2025.11.24 | 开源7B模型刷新多模态推理；GeoVista小模型精准地理定位

24 Nov 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:21] 🧠 OpenMMReasoner: Pushing the Frontiers for Multimodal Reasoning with an Open and General Recipe（OpenMMRea...

【周末特辑】11月第4周最火AI论文 | Kandinsky 5.0开源全家桶；MiroThinker开源智能体

22 Nov 2025

Contributed by Lukas

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...

2025.11.21 | V-ReasonBench考视频模型推理；Step-Audio-R1让语音越“想”越强

21 Nov 2025

Contributed by Lukas

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...

2025.11.20 | 视频模型拍推理链，迷宫百发百中；无标注左右互搏，视觉模型自学跃升

20 Nov 2025

Contributed by Lukas

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（...

2025.11.19 | 像素演员难推理；视觉误导测真章

19 Nov 2025

Contributed by Lukas

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...

2025.11.18 | RL奥赛夺金；Uni-MoE 2.0全能跃升

18 Nov 2025

Contributed by Lukas

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] ...

2025.11.17 | RoPE去噪救长文本；AI速筛离子液体

17 Nov 2025

Contributed by Lukas

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] 🧪 ...

【周末特辑】11月第3周最火AI论文 | 3D游戏智能体开源方案；桌面AI少样本精准操控

15 Nov 2025

Contributed by Lukas

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...

2025.11.14 | UniVA四合一开源视频通才；Depth Anything 3单ViT通吃3D

14 Nov 2025

Contributed by Lukas

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...

2025.11.13 | 原神数据炼成7B通用AI；零训练轨迹秒变视频遥控器

13 Nov 2025

Contributed by Lukas

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...

2025.11.12 | 1.5B小模型反超671B大模型；多智能体质检聊天机器人

12 Nov 2025

Contributed by Lukas

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...

2025.11.11 | 小窗口勤总结刷新深度研究；先广撒网再啃难题激活代码竞赛

11 Nov 2025

Contributed by Lukas

本期的 13 篇论文如下：[00:25] 🧩 IterResearch: Rethinking Long-Horizon Agents via Markovian State Reconstruction（IterResearch：基于马...

2025.11.10 | DeepEyesV2小模型边看图边写代码；纯数据让AI长出立体眼

10 Nov 2025

Contributed by Lukas

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] 🧭 Vi...

【周末特辑】11月第2周最火AI论文 | 视频生成即推理；SVG草图变代码

08 Nov 2025

Contributed by Lukas

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...

2025.11.07 | 视频推理新范式；图像互动促思维

07 Nov 2025

Contributed by Lukas

本期的 12 篇论文如下：[00:21] 🎬 Thinking with Video: Video Generation as a Promising Multimodal Reasoning Paradigm（用视频思考：视...

2025.11.06 | 扩散模型省数据；音视频对口型

06 Nov 2025

Contributed by Lukas

本期的 9 篇论文如下：[00:17] 🚀 Diffusion Language Models are Super Data Learners（扩散语言模型是超级数据学习者）[01:06] 🎬...

2025.11.05 | 向量草图测代码；先画后想补视觉

05 Nov 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:21] 🖼 VCode: a Multimodal Coding Benchmark with SVG as Symbolic Visual Representation（VCode：以SVG为符号视...

2025.11.04 | 超稀疏MoE激活万亿参数；视觉模型看图胜GNN

04 Nov 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:23] 🧠 Every Activation Boosted: Scaling General Reasoner to 1 Trillion Open Language Foundation（全激活赋能...

2025.11.03 | OS-Sentinel实时守护手机操作安全；ThinkMorph让小模型边想边画

03 Nov 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:21] 🛡 OS-Sentinel: Towards Safety-Enhanced Mobile GUI Agents via Hybrid Validation in Realistic Workflows（OS-S...

【月末特辑】10月最火AI论文 | 幼龙BDH稀疏可解释；迷你递归7兆碾压大模型

02 Nov 2025

Contributed by Lukas

本期的 10 篇论文如下：[00:30] TOP1(🔥522) | 🐣 The Dragon Hatchling: The Missing Link between the Transformer and Models of the Brain（幼...

【周末特辑】11月第1周最火AI论文 | 循环模型省参强推理；Concerto 2D-3D自监督涨点

01 Nov 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:35] TOP1(🔥174) | 🔄 Scaling Latent Reasoning via Looped Language Models（通过循环语言模型扩展潜在推...

2025.10.31 | Emu3.5统一预测时空；扩散提示驱动机器人

31 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:26] 🌍 Emu3.5: Native Multimodal Models are World Learners（Emu3.5：原生多模态世界模型让AI看懂并预...

2025.10.30 | 看图写码7B逆袭；视频思维RL破局

30 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:22] 👁 JanusCoder: Towards a Foundational Visual-Programmatic Interface for Code Intelligence（JanusCoder：面向...

2025.10.29 | 通义深度研究报告；小模型折记忆胜671B巨模型

29 Oct 2025

Contributed by Lukas

本期的 10 篇论文如下：[00:23] 🔍 Tongyi DeepResearch Technical Report（通义深度研究报告：面向长程深度信息检索任务的智...

2025.10.28 | Point Transformer无标对齐长空间；代码递归统一粗细粒度

28 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:23] 🎼 Concerto: Joint 2D-3D Self-Supervised Learning Emerges Spatial Representations（Concerto：2D-3D联合自...

2025.10.27 | DeepAgent一步推理+ToolPO；视频即提示DiT秒控百种语义

27 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:27] 🧠 DeepAgent: A General Reasoning Agent with Scalable Toolsets（DeepAgent：具备可扩展工具集的通用...

【周末特辑】10月第4周最火AI论文 | 内部概率+投票剪尾，RPC省样本提精度

26 Oct 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:29] TOP1(🔥135) | 🧠 A Theoretical Study on Bridging Internal Probability and Self-Consistency for LLM Reasoning...

2025.10.24 | AdaSPEC挑40% token提速两成；AutoPage 10美分生成交互网页

24 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:23] 🎯 AdaSPEC: Selective Knowledge Distillation for Efficient Speculative Decoders（AdaSPEC：面向高效推测...

2025.10.23 | 线性注意力显存降十倍；动态裁剪PPO稳提分

23 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:19] 🧠 Every Attention Matters: An Efficient Hybrid Architecture for Long-Context Reasoning（每一种注意力都...

2025.10.22 | LightMem压缩记忆千倍提速12倍；闭环世界模型微调8万数据反超巨兽

22 Oct 2025

Contributed by Lukas

本期的 14 篇论文如下：[00:19] 🧠 LightMem: Lightweight and Efficient Memory-Augmented Generation（LightMem：轻量高效的记忆增强生...

2025.10.21 | 模型不懂光影折射；小模型也能写报告

21 Oct 2025

Contributed by Lukas

本期的 13 篇论文如下：[00:21] 🪞 PICABench: How Far Are We from Physically Realistic Image Editing?（PICABench：我们离物理真实的图...

2025.10.20 | RPC剪枝提速保准；OmniVinci小数据跨模态称王

20 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:20] 🧠 A Theoretical Study on Bridging Internal Probability and Self-Consistency for LLM Reasoning（大模型推...

【周末特辑】10月第3周最火AI论文 | 量化噪声变探索，单卡跑RL；冻结编码器放语义，DiT生成新纪录

18 Oct 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:40] TOP1(🔥154) | 🚀 QeRL: Beyond Efficiency -- Quantization-enhanced Reinforcement Learning for LLMs（QeRL：超...

2025.10.17 | AI眼镜预判式服务；视频生成补想象力

17 Oct 2025

Contributed by Lukas

本期的 11 篇论文如下：[00:25] 👓 AI for Service: Proactive Assistance with AI Glasses（AI服务：AI眼镜的主动式协助）[01:06] 🎬...

2025.10.16 | UniMoE一统语音音乐；注意力图点亮大模型推理

16 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:21] 🎧 UniMoE-Audio: Unified Speech and Music Generation with Dynamic-Capacity MoE（UniMoE-Audio：基于动态容...

2025.10.15 | 像素级自监督ViT刷新生成基准；多智能体评测网文翻译新标尺

15 Oct 2025

Contributed by Lukas

本期的 14 篇论文如下：[00:20] 🖼 Advancing End-to-End Pixel Space Generative Modeling via Self-supervised Pre-training（通过自监督预...

2025.10.14 | 量化误差变奖励，单卡训32B；面向多模态大模型的音视频评测基准

14 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:23] 🚀 QeRL: Beyond Efficiency -- Quantization-enhanced Reinforcement Learning for LLMs（QeRL：超越效率——...

2025.10.13 | 桌面交互预训练解锁机器人潜能；统一模型赋予相机空间想象力

13 Oct 2025

Contributed by Lukas

本期的 14 篇论文如下：[00:20] 🖥 D2E: Scaling Vision-Action Pretraining on Desktop Data for Transfer to Embodied AI（D2E：利用桌面数...

【周末特辑】10月第2周最火AI论文 | 递归小模型刷爆推理榜；未来经验点亮零奖励学习

12 Oct 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:33] TOP1(🔥300) | 🧠 Less is More: Recursive Reasoning with Tiny Networks（小而精：用微型网络递归推...

2025.10.10 | 早期经验的Agent Learning；图文交错反思链跃升至24.9%

10 Oct 2025

Contributed by Lukas

本期的 14 篇论文如下：[00:16] 🌱 Agent Learning via Early Experience（基于早期经验的主体学习）[00:50] 🧠 MM-HELIX: Boosting ...

2025.10.09 | Ming-UniVision统一视觉词表；KV-Cache直连让大模型秒聊

09 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:21] 🔄 Ming-UniVision: Joint Image Understanding and Generation with a Unified Continuous Tokenizer（Ming-UniVis...

2025.10.08 | TaTToo用外挂代码干翻大模型；4B小模型32步逼近闭源巨头

08 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:24] 📊 TaTToo: Tool-Grounded Thinking PRM for Test-Time Scaling in Tabular Reasoning（TaTToo：面向表格推理...

2025.10.07 | 论文秒变演讲；Video-LMM后训练突破

07 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:21] 🎬 Paper2Video: Automatic Video Generation from Scientific Papers（论文自动生成学术演讲视频）[0...

2025.10.06 | 15B小模型追平DeepSeek-R1；渐进蒸馏128 token省八成算力

06 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:28] 🧠 Apriel-1.5-15b-Thinker（Apriel-1.5-15B-Thinker：以小博大实现前沿多模态推理的15B开源模型...

【周末特辑】10月第1周最火AI论文 | Transformer长出大脑的壳；LongLive把长视频做成直播

05 Oct 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:43] TOP1(🔥323) | 🐣 The Dragon Hatchling: The Missing Link between the Transformer and Models of the Brain（幼...

2025.10.03 | LongCodeZip删得快准；迈向分钟级高质量视频生成

03 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:22] 🗜 LongCodeZip: Compress Long Context for Code Language Models（LongCodeZip：面向代码大模型的长上...

2025.10.02 | MCTS破局RLVR瓶颈；GEM开源智能体训练场

02 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:19] 🧠 DeepSearch: Overcome the Bottleneck of Reinforcement Learning with Verifiable Rewards via Monte Carlo Tree...

【月末特辑】9月最火AI论文 | 群体RL共享降本；SAPO让旧机也能训大模型

02 Oct 2025

Contributed by Lukas

本期的 10 篇论文如下：[00:29] TOP1(🔥640) | 🤝 Sharing is Caring: Efficient LM Post-Training with Collective RL Experience Sharing（共享...

2025.10.01 | 自对弈零标注训练；MCP代理深度评测

01 Oct 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:20] 🎮 Vision-Zero: Scalable VLM Self-Improvement via Strategic Gamified Self-Play（Vision-Zero：基于策略化...

2025.09.30 | SLA稀疏注意力砍算力；StableToken抗噪不训模

30 Sep 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:22] ⚡ SLA: Beyond Sparsity in Diffusion Transformers via Fine-Tunable Sparse-Linear Attention（SLA：通过可微...

2025.09.29 | 实时长视频边聊边播；分位数基线稳控推理熵

29 Sep 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:20] 🎬 LongLive: Real-time Interactive Long Video Generation（LongLive：实时交互式长视频生成框架）...

【周末特辑】9月第5周最火AI论文 | Qwen3-Omni开源称王; 锁定视觉训解码，Baseer刷新阿文OCR；

27 Sep 2025

Contributed by Lukas

本期的 5 篇论文如下：[00:38] TOP1(🔥116) | 📜 Baseer: A Vision-Language Model for Arabic Document-to-Markdown OCR（Baseer：面向阿拉...

2025.09.26 | SciReasoner八项全能；MMR1模糊区炼出开源多模态

26 Sep 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:20] 🔬 SciReasoner: Laying the Scientific Reasoning Ground Across Disciplines（SciReasoner：跨学科夯实科学...

2025.09.25 | 视频模型零样本全能；隐式思维链省token提效

25 Sep 2025

Contributed by Lukas

本期的 10 篇论文如下：[00:22] 🎥 Video models are zero-shot learners and reasoners（视频模型是零样本学习者与推理者）[01:09...

2025.09.24 | 阿语OCR刷新指标；无标注RL涨分

24 Sep 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:24] 📜 Baseer: A Vision-Language Model for Arabic Document-to-Markdown OCR（Baseer：面向阿拉伯文档OCR的...

2025.09.23 | 少78条示范让AI飙73.5%；免掩膜视频插主体超Pika

23 Sep 2025

Contributed by Lukas

本期的 15 篇论文如下：[00:21] 🚀 LIMI: Less is More for Agency（LIMI：少即是多，打造AI智能体）[00:55] 🎬 OmniInsert: Mask-Fr...