Hugging Face 动态
热门模型 · 每日论文 · 热门 Spaces
Hugging Face 上正在火的模型、论文和在线 Demo，数据来自 HF 官方公开接口，严格按 HF 热度排名，每 6 小时更新一次。
2026 年 10 月 2 日的论文（UTC）· 回到最新
10 月 2 日31 篇
- 222点赞OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video Interaction
arXiv 2610.01762Xiangyu Zeng, Yuandong Yang, Zhiqiu Zhang 等 24 人代码 ↗项目主页 ↗
- 183点赞On-Policy or Off-Policy Learning? A Systematic Study of Distillation Dynamics
arXiv 2609.35259Julianna Piskorz, Antonin Berthon, Mihaela van der Schaar
- 73点赞Transformers Stop Thinking Too Early, and a Tiny LoRA Fixes It
arXiv 2609.36585Zehao Jin, Ruixuan Deng, Junran Wang代码 ↗项目主页 ↗
- 69点赞X-Tree: Tokenizing Reusable Experience for Efficient Agent Generalization
arXiv 2609.32993Sitao Cheng, Xunjian Yin, Zhiyuan Sun 等 7 人代码 ↗项目主页 ↗
- 61点赞E-MoE: Enhanced Mixture-of-Experts for Non-Factorized Diffusion Language Models
arXiv 2609.37533Arseny Ivanov, Alexander Kolesov, Alexander Korotin 等 5 人
- 56点赞Video Generation Models: A Survey of Post-Training and Alignment
arXiv 2610.00812Chaoyu Li, Xiaoyi Gu, Yogesh Kulkarni 等 13 人代码 ↗
- 51点赞Persona Dosing: Calibrated Activation Steering for Graded Trait Control
arXiv 2609.36388Zehao Jin, Junran Wang, Ruixuan Deng 等 7 人
- 46点赞Fewer Tokens, Better Action: GPT-6 Astra Robot Agents with 14% Higher Success Rate but 65% Fewer Tokens
arXiv 2610.01939Ruiyang Si, Jianxin Bi, Shunyu Yang 等 12 人代码 ↗项目主页 ↗
- 45点赞Architect-Ant: Editable Automatic Furnishing of Architectural Floor Plans
arXiv 2606.10953Fedor Rodionov, Aleksandar Cvejic, Michael Birsak 等 5 人项目主页 ↗
- 43点赞Decoding Looped Transformers Better for (Almost) Free
arXiv 2610.02185Weihao Liu, Huangjie Zheng, Tianrong Chen 等 8 人
- 33点赞CorrGRPO: Correlation-Normalized GRPO for Multi-Reward Learning
arXiv 2609.36820Wenbin Hu, Huihao Jing, Haochen Shi 等 6 人代码 ↗项目主页 ↗
- 27点赞Ego2Act: Evaluating Goal-Directed Manipulation in Egocentric Video Generation
arXiv 2610.01092Patrick Amadeus Irawan, Iskandar Muda Rizky Parlambang, Rava Maulana 等 13 人代码 ↗项目主页 ↗
- 23点赞PhysVista: Benchmarking Physical Intelligence in VLMs via a Perception-Reasoning-Assessment Loop
arXiv 2610.00559Xinge Peng, Yiting Lu, Tianwu Zhi 等 7 人代码 ↗
- 17点赞Where-OPD: Spatially Guided On-Policy Self-Distillation of MLLMs with Synthetic Scenes
arXiv 2610.02117Sophia Sirko-Galouchenko, Monika Wysoczanska, Andrei Bursuc 等 5 人
- 14点赞OpenTumorBoard: A Real-World Benchmark of Multidisciplinary Tumor Board Discussion Trajectories
arXiv 2609.32810Anqi Li, Zhixuan Ge, Yixuan Duan 等 12 人代码 ↗项目主页 ↗
- 13点赞Removing the NEEDLE in the Haystack: Backdoor Removal in LLMs via Weight Orthogonalisation
arXiv 2610.00348Minoo Kim, Vasileios Lampos, George Drayson代码 ↗
- 12点赞RLE-Bench: A Qualifying Exam for Coding Agents as Robot Learning Engineers
arXiv 2609.34210Haitong Ma, Chenxiao Gao, Rushi Qiang 等 5 人代码 ↗项目主页 ↗
- 12点赞ScholarCatalyst: A Benchmark for Retrieving Papers That Inspire New Research
arXiv 2610.02202Sohyeon Kim, Yoonho Lee, Bo Liu 等 14 人
- 11点赞SemanTok: Predictable Semantic Tokens for Efficient Autoregressive Video Generation
arXiv 2610.00686Mikhail Dereviannykh, Vikram Voleti, Simon Donne 等 6 人代码 ↗项目主页 ↗
- 11点赞VTR-Bench: A Systematic Benchmark for Evaluating Visual Text Rendering in Video Generation
arXiv 2610.01499Yu Huang, Jungang Li, Zhiyuan Wang 等 11 人代码 ↗
- 11点赞Generalization Is Stability, Not Accuracy: Multi-Axis Evaluation of LLMs
arXiv 2610.01428Nagham Omar, Mahmoud Jabarin, Maya Rozenshtein 等 6 人
- 11点赞LOCI: Spatial Linear Memory for Streaming World Models
arXiv 2609.40222Ji Xia, Tingting Liao, Xuezhi Liang 等 5 人代码 ↗项目主页 ↗
- 8点赞Latent-Foresight: End-to-End Learning Predictable Representations for Latent World Models
arXiv 2610.01942Efstathios Karypidis, Spyros Gidaris, Nikos Komodakis代码 ↗
- 7点赞MemFold: Learning Compact Soft Memory for Long-Context Personalization via On-Policy Optimization
arXiv 2609.36435Jingxuan Wu, Yuzhe Yang, Yiqiao Huang 等 9 人代码 ↗项目主页 ↗
- 7点赞OTRetarget: Joint Robot and Object Motion Retargeting via Optimal Transport
arXiv 2609.36602Guillaume Besset, Erwann Carn, Timothée Carecchio 等 8 人项目主页 ↗
- 7点赞
- 6点赞When Does Correction Become Repair? Mechanistic Auditing of Internal Interventions in Tool-Using LLMs
arXiv 2609.36138Jiayi Li, Ruizhe Li代码 ↗
- 5点赞Keyword Harnesses Fail Open: A Cheap Diagnostic Ladder for Tool-Use Claims in Small Language Models
arXiv 2610.02142Juan S. Santillana项目主页 ↗
- 5点赞Replacing Large Language Models with Jev Decision Models for Low-Latency Edge Service Orchestration
arXiv 2609.22753Delong Li, Xu Wang, Haochen Gong 等 5 人代码 ↗
- 5点赞Honeycomb: Constant-Size Scene Memory Representation for Video World Models
arXiv 2609.37690Jack Wei Lun Shi, Kaichen Zhou, Haoyu Chen 等 9 人代码 ↗项目主页 ↗
- 5点赞Before It Fades: Reinforcing Temporal Representations at Inference Time in VideoLLMs
arXiv 2610.01595Youngwoo Shin, Yusung Ro, Minseo Kim 等 4 人代码 ↗
数据说明
数据来源
热门模型、热门 Spaces、每日论文分别来自 Hugging Face 官方公开接口，每 6 小时抓取一次，按 UTC 日期每天存一份快照，可以按日期回看。论文标题保留英文原文。
排名口径
排名完全按 HF 返回的热度（trendingScore），本站不做任何筛选或调整，看到的就是 HF 热榜本身。同一天多次抓取时，以当天最后一次为准。按任务类型或 SDK 筛选只是缩小列表，名次仍是 HF 热榜上的原名次。
点赞增量与新上榜
「7 天增量」是所看那一天快照的点赞数减去它之前 7 天内最早一份快照的点赞数；历史不满 7 天时括号里标出实际天数，没有更早的快照则显示「—」。「新上榜」指那一天在榜、之前的快照里从没出现过。
免责声明
本页不是 Hugging Face 的官方页面。模型和 Space 的许可、用途与内容以 Hugging Face 原页面为准。