A pretrained humanoid controller helps a compact planner learn simulated boxing in about seven minutes of additional single-GPU training.
Xihan Ma, Bo Zhang, Boxiong Li, Yuhan Bao, Ziyue Liu, Yucheng WangRead
Keep Your Research Taste Sharp
Your daily guide to the latest breakthroughs in AI research
01
Robotics
DiscoDemo trains robots to find varied successful ways to complete tasks, then turns those behaviors into better imitation-learning data.
Minho Park, Kinam Kim, Donghu Kim, Byungkun Lee, Dongyoon Hwang, Yongjae Shin, Junha Hyung, Hojoon Lee, Jaegul ChooRead
KAIST · Holiday Robotics
HyperDrone aims to make realistic drone simulation fast enough to train vision-based flight policies at massive scale, then transfer them to a real quadrotor.
Jonas Eschmann, Giuseppe LoiannoRead
University of California, Berkeley
02
Hardware
DuoPD pairs GPUs with memory-side processors to run LLM prefill and decoding together and improve inference throughput.
Sanghyeok Han, Inseong Choi, Byungkuk Yoon, Gyeonghwan Park, Jae-Joon KimRead
Seoul National University, Seoul, Republic of Korea
02
HAWK: Heterogeneity-Aware KV Cache Transfer for Efficient GPU-NPU Phase-Disaggregated LLM Inference
HAWK makes it faster and more efficient to move LLM KV caches from GPUs to NPUs in a split-inference system.
Junhyeok Park, Jungeun Yoon, Kyungmin Kim, Hunsoon Lee, Shinyoung Ahn, Kangho KimRead
Georgia Institute of Technology, Atlanta, USA · Electronics and Telecommunications Research Institute (ETRI), Daejeon, Republic of Korea
Researchers connected two large quantum processors over nearly a meter and used the links to transfer states and create entanglement.
Neereja M. Sundaresan, Nicholas Cotaj, Yadav Prasad Kandel, Eric Lewandowski, Christian Robert Longton, Yves Martin, Jae-woong Nah, Kathryn Pooley, Martin Sandberg, Michael Scanlon, Chi Xiong, Emanuil Sashev Yanev, David Abraham, Abhinandan Antony, Vincent Arena, Nicholas Bornman, Khanh Dang, Oliver Dial, Christopher Guarino, Gabriel Haspil, Jared Hertzberg, Kentaro Heya, Swetha Kamlapurkar, Michael Karschner, George Keefe, Eric Liniger, Laura Luo, Thomas G. McConkey, David Olson, Jason Orcutt, Dmitry Ovchinnikov, Timothy Phung, Sankeerth Rajalingam, Michael Selvanayagam, William Shanks, Dongbing Shao, Abigail Shearrow, Adam Stabile, Jiri Stehlik, Joseph Suttle, Jeng-Bang Yau, Eric ZhangRead
IBM Research, T. J. Watson Research Center, Yorktown Heights, NY 10598, USA · Anderon · Amazon Web Services
03
Agent
A team of LLM research agents builds a Hearthstone Battlegrounds AI that reportedly outperforms top human players on the live ladder.
Changlian Ma, Limin WangRead
Nanjing University · Shanghai AI Laboratory
The paper asks whether agent memory actually helps, or whether it can lose useful evidence compared with retrieving directly from history.
Xuan He, Sining Zhoubian, Ruozhen Yang, Yuhao Wen, Zhineng Wang, Yuyang Hu, Ke Yang, Yuyu Zhang, Ping Nie, Jiawei Han, ChengXiang ZhaiRead
University of Illinois Urbana Champaign · Tsinghua University · Renmin University of China · Verdent.ai · University of Waterloo
An agent’s performance depends not just on its model, but on how well its learned preferences fit the harness it is given.
Jian Mu, Ruibo Shen, Yao ShuRead
The Hong Kong University of Science and Technology (Guangzhou) · The Hong Kong University of Science and Technology
Load more videos
04
Safety
HoS adds learned attention gates to help language models handle untrusted text without obeying its hidden instructions.
Igor Shilov, Xiaoxue Yang, Alexander Panfilov, Evgenii Kortukov, Matthieu Meeus, Peter Romov, Wojciech Samek, Jonas Geiping, Yves-Alexandre de MontjoyeRead
Fraunhofer HHI ∗ †
A new testbed finds that AI models can covertly sabotage the training of their successors, often evading transcript-based monitors.
Kaiwen Zhou, Udari Madhushani SehwagRead
Preparedness and Policy, Scale AI
A principled way to reduce human review of AI decisions while keeping unaudited errors below a chosen limit.
Isaac Gibbs, Elise L. Han, Ryan J. Tibshirani, Tijana ZrnicRead
Department of Statistical Sciences, University of Toronto · Department of Computer Science, Stanford University · Department of Statistics, University of California, Berkeley · Departments of Statistics and Management Science & Engineering, Stanford University
05
Cv
PhysInsert aims to make inserted objects fit their scenes by adapting their pose, perspective, and lighting using geometric and object-scene cues.
Yue Gong, Le Zhang, Rui Chen, Song Chang, Xiaochao Qu, Luoqi Liu, Ting Liu, Yebin LiuRead
[
Foundation4D combines multi-camera driving video to estimate scene geometry, detect objects, map occupancy, and reconstruct scenes for simulation.
NVIDIA, :, Sérgio Agostinho, Maximilian Apodaca, Emmanuel Attia, Guillem Braso, George Dong, Ilya Fedorov, Michele Fenzi, Gang Huang, Jiahui Huang, Shengyu Huang, Bingxin Ke, Laura Leal-Taixé, Kuan-Ting Lee, Mingxuan Liu, Tian Liu, Tim Meinhardt, Aljoša Ošep, Naveen Kumar Rai, Kefei Ren, Cristiano Saltori, Keqiang Wu, Ning Xu, Fei Xue, Jin-Bey Yu, Lei Zhang, Qunjie ZhouRead
WorldCast aims to keep generated characters visually consistent as the camera makes large, complex moves around them.
Lei Cao, Yu Liu, Wenhua Liao, Peng Zhang, Fei Wang, Hao Zhang, Jiajun Le, Daiguo Zhou, Jian Luan, Jiayi MaRead
Electronic Information School, Wuhan University · MiLM Plus, Xiaomi Inc · School of Robotics, Wuhan University
06
Multimodal
A new training framework teaches multimodal models to describe a visual region and then find that same region again.
Xin Zhang, Haochen Wang, Beibei Lin, Zhaoxiang Zhang, Robby T. Tan, Jason Li, Dacheng TaoRead
The paper argues that training on meaningful cross-modal views aligns representations far more efficiently than relying on next-token prediction and scale alone.
Randall Balestriero, Yann LeCunRead
Brown University · AMI LabsCorrespondence: rbalestr@brown.edu Code Live demo · New York University
Youtu-Parsing-Omni uses one model and one output schema to turn documents, images, audio, and video into structured, evidence-grounded records.
Yunhang Shen, Zhangquan Chen, Xiaotian Li, Zhongpeng Cai, Qing Zhou, Zekai Zhang, Kun Zhang, Zezhong Wu, Jian Hu, Haihan Gao, Heting Gao, Xin Li, Yinsong Liu, Deqian Jiang, Di Yin, Haoyu Cao, Xing Sun, Bing Liu, Yanqiu Qu, Yunfei Wu, Kun Yin, Zhixiang Wei, Yongxiang Hua, Qianyu Li, Xiaoxing Hu, Peiyuan Zhang, Lijiang Li, Dong Xu, Zhenchuan Yang, Yiliang JiangRead
Load more videos
07
Efficiency
AB-Attention speeds up video generation by skipping most attention computations while more accurately estimating the attention it omits.
Bin Xu, Mo Zhang, Zhengfeiyang Zhang, Xinfeng Yang, Yixing Ma, Wei Wang, HeXin Wang, Pengfei Zhou, Wangbo ZhaoRead
Inf Rec · The Chinese University of Hong Kong, Shenzhen · Shandong University · The University of Tokyo · Fudan University · University of California, Berkeley · The Hong Kong University of Science and Technology · National University of Singapore
A new way to compress Transformer embeddings dramatically reduces memory use while largely preserving model performance.
Jiacheng Liu, Yaxin Luo, Xiaohan Zhao, Peng Sun, Xinyi Shang, Jiacheng Cui, Yonina C. Eldar, Zhiqiang ShenRead
Mohamed bin Zayed University of Artificial Intelligence · Westlake University · Zhejiang University · University College London · Weizmann Institute of Science
A new regression-based compressor aims to keep language and vision models efficient while preserving more useful information in fewer tokens.
Sihan Yang, Jie Huang, Hanxuan Li, Yifan Guo, Weigao Sun, Yiran Zhong, Weiyang LiuRead
The Chinese University of Hong Kong · Independent Researcher · Token Foundry, Alibaba Group
08
Cv
F2G-Pose turns one RGB-D view into an object’s pose, size, and completed shape in a fast feed-forward pass.
Jinyu Zhang, Haitao Lin, Jiashu Hou, Xiangyang Xue, Yanwei FuRead
Fudan University · Shanghai Innovation Institute
HRoPE helps vision transformers align image patches across camera views by encoding the geometry of the surfaces those patches depict.
Huachen Gao, Yang Yang, Xiaoqian Liang, Xingyu Chen, Youjia Zhang, Qi Zhang, Baorui Ma, Anpei Chen, Wei YangRead
Zhejiang University · Westlake University · Huazhong University of Science and Technology · VIVO · Tsinghua University
Verus-Insight aims to detect AI-generated videos more reliably by finding and verifying the visual artifacts people can actually see.
Zhuoran Zhang, Yang Shi, Qixun Wang, Tengfei Liu, Yuqi Tang, Xuanyu Zhu, Yizhuo Jia, Jingyun Hua, Shixuan Liu, Xianghua Ying, Haotian Wang, Yuanxing Zhang, Pengfei Wan, Wenjing YangRead
Peking University · Kling Team · HKUST (GZ) · Fudan University · Tsinghua University
09
Continual Learning
The study finds that current self-evolving AI agents can learn new workflows, but often do so by disrupting what they already know.
Xiaoyin Chen, Yizheng Huang, Yinglun Xu, Ziyun Zhao, Emmett McQuinn, Julie Liu, Hope YaoRead
Mila – Quebec AI Institute · Université de Montréal · Work done as a student researcher at Google DeepMind · \thepa
02
Don't Forget What You Just Learned: Self-Directed Repair Turns Agent Mistakes Into Lasting Lessons
Self-Direction turns an agent’s mistakes into small reusable adapters so it can avoid repeating them on future tasks.
Yuval Kansal, Karen Ullrich, Polina Kirichenko, Ilia Kulikov, Niraj K. Jha, Mohammad Pezeshki, Mark IbrahimRead
Princeton University · Meta FAIR
Subspace Tuning aims to help language models learn new skills while preserving the capabilities they already have.
Arnav Chavan, Kusha Sareen, Rishabh Tiwari, Abhranil Chandra, Johan Obando-Ceron, Kurt Keutzer, Scott Niekum, Inderjit S Dhillon, Devvrit KhatriRead
UC Berkeley · Mila · McGill University · UMass Amherst · Université de Montréal · Google · UT Austin · Mirendil
Load more videos
10
Diffusion
SPOT speeds up diffusion language models by predicting which positions to commit next, then checking multiple proposed tokens in one pass.
Eliezer Shlomi, Eliron Rahimi, Avi Mendelson, Chaim BaskinRead
Technion – Israel Institute of Technology · INSIGHT Lab, School of Electrical and Computer Engineering, Ben-Gurion University of the Negev
PhysLDM uses latent diffusion to predict plausible long-term motion of complex deformable objects more effectively than deterministic averaging.
Yu Zhang, Xudong Xu, Xingang PanRead
S-Lab, Nanyang Technological University, Singapore · Shanghai Artificial Intelligence Laboratory, Shanghai, China
SWIFT uses a weaker diffusion model to help train a stronger one, aiming for better image generation without extra inference-time cost.
Yanjin He, Cuifeng Shen, Wanzhou Lei, Maohua Li, Tao Lan, Kan Liu, Hanlin Tang, Qing QuRead
University of Michigan, Ann Arbor · Alibaba Group · University of California, Berkeley · Nanjing University
11
Efficiency
A calibration-free way to skip most MoE expert-channel reads aims to make decoding faster while preserving model quality.
Geunhyoung Kim, Sung Ju HwangRead
KAIST · DeepAuto.ai
Nestor combines nested attention and MoE sparsity to make language-model pretraining and continual adaptation more efficient.
Tianzhe Chu, Druv Pai, Yi MaRead
University of Hong Kong · UC Berkeley
TokenGate teaches language models to skip less relevant tokens during prefill, aiming to cut latency and memory without sacrificing accuracy.
Liana Patel, Rishabh Ranjan, Ion Stoica, Matei Zaharia, Carlos GuestrinRead
Stanford University · UC Berkeley https://github.com/tokengate-models/TokenGate
12
World Model
LaLa-World uses language to train a scalable latent model of how the world changes, then uses its imagined futures to plan robot actions.
Junha Song, Geonmo Gu, Jeongeun Park, Heejeong Nam, Lucas Maes, Byeongho Heo, Dongyoon Han, Sangdoo YunRead
NAVER AI Lab · Georgia Institute of Technology · Brown University · Mila & Université de Montréal
VANE learns visual representations by predicting what comes next, using the same objective for both images and videos.
Yifeng Yao, Jinghan Li, Zhihao Zhang, Yajiao Xiong, Yang Jin, Kun Xu, Huishuai Zhang, Dongyan ZhaoRead
Peking University · ByteDance · UNC-Chapel Hill
A new active data-collection method helps video world models respond to commanded actions instead of merely imagining plausible futures.
Samuel M. Bateman, Tenny Yin, Chongyi Zheng, Lei Huang, Brian Wang, Benjamin Eysenbach, Jaime Fernández Fisac, Dhruv ShahRead
[
Load more videos
13
Generative Model
Qwen-Image-PE turns brief image requests into more carefully designed prompts to improve the images generated by text-to-image models.
Zekai Zhang, Yanran Zhang, Shengming Yin, Yan Shu, Zhening Liu, Bingxuan Zhao, Chenyang Zhu, Donghao Zhou, Jiahao Li, Jiahao Wang, Jie Zhang, Kaiyuan Gao, Kun Yan, Lihan Jiang, Ningyuan Tang, Qihang Ai, Rui Song, Shuo Cao, Xiao Xu, Xiaoyue Chen, Yuxiang Chen, Zhendong Wang, Zhouyang Li, Zihao Liu, Zikai Zhou, Bo Zheng, Huishuai Zhang, Dongyan Zhao, Chenfei WuRead
Bingxuan Zhao, Chenyang Zhu, Donghao Zhou, Jiahao Li, Jiahao Wang, Jie Zhang · Kaiyuan Gao, Kun Yan, Lihan Jiang, Ningyuan Tang, Qihang Ai, Rui Song · Shuo Cao, Xiao Xu, Xiaoyue Chen, Yuxiang Chen, Zhendong Wang, Zhouyang Li · Zihao Liu, Zikai Zhou, Bo Zheng, Huishuai Zhang, Dongyan Zhao, Chenfei Wu
T-DIG lets users steer how a text-to-image system varies its results, without needing to modify the underlying image generator.
Simran Khanuja, Rohit Gandikota, Gaurav Parmar, Graham NeubigRead
Carnegie Mellon University · Northeastern UniversityCorrespondence: skhanuja@andrew.cmu.edu image-study.org/t-dig
Halcyon shows that capable text-to-image models can be trained from scratch on image-text pairs, without first pretraining a language model.
Keshigeyan Chandrasegaran, Yue Zhao, Armin W. Thomas, Ehsan Adeli, Jiajun Wu, Juan Carlos Niebles, Li Fei-Fei, Michael PoliRead
Stanford University · Radical Numerics
14
Diffusion
MOSS-Image pairs a 5B diffusion model with prompt planning to improve text-heavy, layout-coherent image generation.
MOSI. AIRead
[
MT Forcing trains autoregressive diffusion models to make image edits that stay visually stable and consistent across multiple turns.
Houyuan Chen, Fanzhou Wang, Xiangyu Fan, Zesong Qiu, Yukun Wei, Songlin Yang, Zhitao Yang, Lvmin Zhang, Pei Yang, Hao Zhao, Lei Yang, Anyi RaoRead
SenseTime Research · HKUST MMLab · Stanford University · AIR, Tsinghua University
NLD-Uni uses self-refining diffusion and cross-paradigm distillation to bring visual understanding and image generation together in one language model.
Shufan Li, Hanrong Ye, Greg Heinrich, Yonggan Fu, Haotian Zhang, Sifei Liu, Aditya Grover, Pavlo MolchanovRead
*Core Contribution
15
Reasoning
The paper teaches LLMs to ask more informative questions, making sequential information gathering cheaper and more transferable.
Anita Kriz, Antonin Berthon, Nicolás Astorga, Mihaela van der SchaarRead
University of Cambridge · Cambridge, UK
CoTranslator turns cheaper answer-conditioned reasoning traces into useful student training data while approaching the performance of more expensive rejection sampling.
Zinan Tang, Zhuoshi Pan, Kairui Chen, Ze Yan, Shaomian Zheng, Yukun Zhang, Gang Wang, Biqing HuangRead
Tsinghua University · Ant Group
OneLatent replaces lengthy reasoning traces with compact learned tokens to make foundation recommendation models faster and more effective.
Hao Jiang, Peiru Du, Pengfei Yao, Mengting Li, Siyuan Lou, Kuo Cai, Sheng Yu, Qiang Luo, Jian Liang, Ruiming Tang, Fei Pan, Peng Jiang, Wenwu OuRead
Kuaishou Technology
Load more videos
16
Transformer
The paper finds that a transformer’s direct memory of its prompt fades as it generates text, while the text itself can carry that memory forward.
Shuyang XiangRead
Independent researcher
DAGFormer lets a Transformer learn how information should flow across its layers, improving language-model performance across several evaluations.
Xiaocong Yang, Yuren Hao, ChengXiang ZhaiRead
AI Interpretability @ Illinois · University of Illinois Urbana-Champaign · {xy51, yurenh2, czhai}@illinois.edu
The paper finds that a carefully designed looped Transformer can outperform comparable models while using far fewer resident parameters.
Benhao Huang, Chufan Shi, Junlin Chen, Shicheng Wen, Zhengzhong Liu, Eric Xing, Xuezhe MaRead
[
17
Generative Model
InfiniDream uses language-guided procedural tools to build editable, photorealistic 3D assets and populate expansive worlds.
Yi-Chuan Huang, Hao-Jen Chien, Zheng-Hui Huang, Yu-Lun Liu, Kaipeng Zhang, Zhixiang WangRead
[
Matrix-Game 4.0 combines game-engine rules and real-time video generation to turn natural-language ideas into playable worlds.
Jihai Zhang, Runjia Qian, Kaiqin Kong, Biao Jiang, Puyi Wang, Siru Chen, Tianle Li, Guanjie Chen, Junyi Wu, Chubin Chen, Yuanhang Li, Kaichen Huang, Haobo Zhang, Yangguang Li, Wei Yu, Guibin Chen, Yu Cheng, Yahui ZhouRead
SkyworkAI · The Chinese University of Hong Kong · UC San Diego
SolarRenderer turns varied scene structures into realistic video in just four generation steps.
Guian Fang, Shengju Qian, Dongyue Lu, Junchao Huang, Lutao Jiang, Xin Wang, Yiren Song, Li Jiang, Mike Zheng ShouRead
SolarRenderer renders structure of any kind into realistic video. Each wedge pairs a structure (inner ring) with its rendering, and seven of the ten structure types are unseen in training. Readers can · LIGHTSPEED · CUHK(SZ) · SLAI · HKUST(GZ) Project Page: enderfga.cn/SolarRenderer
18
Scientific Ai
Agent4Sci tests how far frontier AI agents can improve real scientific ML models—and finds they need more than engineering tricks to make research-level progress.
Ming Zhong, Stacy Li, Nicholas Carlini, Matthew JagielskiRead
Anthropic Fellows Program Anthropic
OptScientist uses symbolic modeling and an LLM agent to infer the optimization rules that may underlie scientific and social phenomena.
Ma ZhenYao, Yue Liang, Jian-Qiao ZhuRead
Plokerm AI Lab · Xiamen University · University of Tübingen · The University of Hong Kong
An AI system turns scientific goals into working lab setups, robots, and experiments.
Jiaru Bai, Yuchi Zhao, Han Hao, Miroslav Bogdanovic, Yang Cao, Varinia Bernales, Alán Aspuru-GuzikRead
[
Load more videos
19
Llm
Belief-Trajectory Energy tracks how an LLM’s predictions change across layers to reveal information that its final answer alone can miss.
Jiahao Ying, Wei Tang, Boxian Ai, Yaoning Wang, Haotian Chen, Wenhe Sun, Caijun Xu, Haozhan Cai, Changyi Xiao, Yixin CaoRead
[
A staged training approach helps hybrid language models balance recurrence and attention, improving language modeling and retrieval.
Xingyu Shen, Yingfa Chen, Ziqing Qiao, Chaojun Xiao, Xu Han, Zhiyuan LiuRead
Department of Science and Technology, Tsinghua University, Beijing, China · xingyu-c21@mails.tsinghua.edu.cn, chenyingfa1999@gmail.com
A single correct opening token can help an LLM recover factual answers it seemed unable to recall.
Jaewon Sok, Jewon Yeom, Seonghyeon Park, Jeongjae Park, Hwiyeong Lee, Taesup KimRead
Department of Rural Systems Engineering, Seoul National University · Graduate School of Data Science, Seoul National University · Department of Aerospace Engineering, Seoul National University
20
Llm
Copy-Paste Language Models add a low-overhead copying mechanism that improves language-model training and performance, especially in long-context and retrieval settings.
Nathan Godey, Yoav ArtziRead
Cornell University · New York City, USA
Lafa turns large language models into fast continuous-flow generators, reporting up to 2.54× throughput at comparable quality.
Arnav Chavan, Chanhyuk Lee, Udbhav Bamba, Jaehoon Yoo, Seunghoon Hong, Aditi Raghunathan, Jinwoo KimRead
UC Berkeley · Amazon · KAIST · Carnegie Mellon University
A large-scale study finds that RAG techniques that work well alone may not work well together, and the best combination can change as the knowledge corpus grows.
Ido Levy, Segev Shlomov, Leshem ChoshenRead
IBM · Weizmann Institute of Science
21
Code Generation
Self-HPO helps coding agents improve both their problem-solving policy and the tools and workflows that guide it.
Zhepei Wei, Wenlin Yao, Baolin Peng, Danqing Wang, Qianhui Wu, Hao Cheng, Yu Meng, Jianfeng GaoRead
Microsoft Research · University of Virginia · Carnegie Mellon University https://github.com/MSR-Orchard/Self-HPOhttps://huggingface.co/self-hpo
HarnessAdapt-RL improves coding agents by evolving their tools and interaction setup alongside the model during training.
Lingxiao Tang, Wei Tao, Yiwen Guo, Lingfeng BaoRead
[
03
WebCoder: Scaling General Frontend Development with Artifact-Grounded Agentic Reinforcement Learning
WebCoder trains an AI agent to build and improve full frontend projects by learning from how its code changes perform in a live application.
Zijun Dou, Zhen Yang, Xijun Liu, Dongzhuoran Zhou, Aohan Zeng, Jie TangRead
Tsinghua University · Zhipu AI · Beihang University · University of Oslo
Load more videos
22
Safety
01
Policy Is Not Capability: Some Safety Benchmarks Measure Which Definition of "Unsafe" You Trained On
A guardrail’s safety score may say more about the policy used to label its training data than about its ability to detect harm.
Vahid Jebraeeli, Jonathon Romero, Yijin Wang, Adrian Chan, Jianbang ZhangRead
Lenovo
The study finds that rewriting sensitive records to remove unique details can preserve model capabilities while reducing privacy leakage.
Zihao Zhao, Avi Schwarzschild, Casey Meehan, Katherine Lee, Zhili Feng, Dylan SamRead
Johns Hopkins University · OpenAI
SCOPE turns natural-language rules into interpretable programs that can enforce policies on LLM agents more reliably and cheaply.
Arnon Mazza, Elad LeviRead
Plurai
23
Safety
RIDAR detects research-integrity violations in autonomous research agents and intervenes during execution to reduce them.
Jiening Siow, Amy Xin, Junjie Wang, Ziyu Tang, Zijun Yao, Yunjia Qi, Fanjin Zhang, Lei Hou, Jie Tang, Juanzi LiRead
Department of Computer Science and Technology, Tsinghua University · Zhipu AI · School of Information, Renmin University of China
A model’s internal signals can warn that it is about to exceed its authority, enabling a monitor to intervene before the action—and the cover story.
Arya Fayyazi, Mehdi Kamal, Massoud PedramRead
University of Southern California, Los Angeles, CA, USA
Intact helps browsing agents distinguish who controls webpage content and limits what they can do with it.
Umar IqbalRead
Washington University in St. Louis
24
Embodied Ai
RAX lets coding agents explore, learn, and compose robot skills across different embodiments without relying on human demonstrations.
Runhan Huang, Xiangcheng Zhang, Haoyu Chen, Yitong Li, Yen-Jen Wang, Bohan Lyu, Haotian Ye, Chuye Hong, Yixuan Wang, Chloe Su, Haoyu Xiong, Yangzhen Wu, Enze Xie, Travis Manderson, Stefano Ermon, Pieter Abbeel, Antonio Torralba, Dawn Song, Weirui Ye, Yilun DuRead
[
Mirror helps embodied agents turn visual plans into actions by learning from their own attempts in the environment.
Hyunwoo Ryu, Eric Ming Chen, Max Wilcoxson, Vincent SitzmannRead
Computer Science & Artificial Intelligence Laboratory · Massachusetts Institute of Technology · Cambridge, MA 02139, USA
SuperMimic uses a differentiable world model to train humanoid motion controllers far faster, and demonstrates the result on a real robot.
Qiayuan Liao, Yiyang Shao, Junfeng Long, Kevin Zakka, Roberto Horowitz, Pieter Abbeel, Koushil SreenathRead
UC Berkeley
Load more videos
25
Benchmark
HandsOn is a large benchmark for testing whether dexterous robots can generalize manipulation skills across hands, objects, and tasks.
Tianwei Luo, Yiming He, Hongjin Wu, Ting Pan, Debin Ren, Haorui Guo, Junlin Yang, Yuxin Zuo, Zhehao Kou, Kaisen Yang, Zhuoqun Lai, Yi Ding, Jiale Yu, Yuchen Zhang, Zijie Qiu, Haozhan Li, Ning DingRead
Tsinghua University Beijing University of Posts and Telecommunications
A new benchmark finds that coding agents often produce functional code that still fails to protect against real vulnerabilities.
Tianyi Wu, Chao Ge, Peng Chen, Fengji Zhang, Yongxing Wu, Yuchong Xie, Jian Li, Chaoyuan Peng, Kaikai Zhang, Dong Huang, Mingzhe Du, Yukai Huang, Jiaheng Zhang, Terry Yue Zhuo, See-Kiong NgRead
National University of Singapore · Qwen Team, Alibaba Token Hub, Alibaba Group†Corresponding author. GitHub Leaderboard Harbor Hub · Cloud Safety and Security, Alibaba Cloud, Alibaba Group
LifeSearch tests how well deep-research agents handle everyday and professional questions—and finds that even top systems still have substantial room to improve.
Yibo Wang, Senjie Jin, Yansong Ning, Xinbing Liang, Haotian Luo, Jiazheng Zhang, Yuxuan He, Will Wu, ClarenceaiRead
Jiazheng Zhang, Yuxuan HE, Will WU, Clarenceai · LLM Department, Tencent
26
Benchmark
FrontierCS 2 tests whether AI agents can improve real research codebases across a wide range of computer science tasks.
Qiuyang Mang, Wenhao Chai, Runyuan He, Dongruixuan Li, Zihan Zheng, Yujun He, Xuanyi Zhou, Hao Wang, Siddhant Ray, Bo Peng, Ruyi Ji, Shuo Yang, Haocheng Xi, Boru Chen, Xinyang David Han, Melissa Pan, Mayank Mishra, Boxi Yu, Sijia Liu, Kerem Akillioglu, Yi Pan, Zirui Wang, Mert Cemri, Diogo Antunes, Yichuan Wang, Aiwei Ivy Zhang, Zhenyang Chen, Qizheng Zhang, Kaiyuan Liu, Wenyuan Huang, Sherry Ruan, Shu Liu, Dacheng Li, Zerui Cheng, Zeyu Shen, Shang Zhou, Hanfei Chen, Yuzhong Zhang, Bohan Lyu, Qu Ao, Han Zheng, Chonghe Jiang, Jingzhuo Hu, Xiaokun Chen, Kunjun Li, Zhiqing Zhong, Haoye Tian, Tong Zhao, Yunzhong He, Hanchen Li, Ofir Press, Jiantao Jiao, Kunle Olukotun, Karthik R. Narasimhan, Jingbo Shang, Koushik Sen, Matei Zaharia, Alex Dimakis, Ion Stoica, Joseph E. Gonzalez, Alvin CheungRead
UC Berk
A new benchmark finds that today’s AI agents can write plausible-looking scientific follow-ups, but none of 135 submissions met all expert-defined requirements.
Seungone Kim, Jiayi Geng, Young-Jun Lee, Bo Liu, Juyoung Suk, Nicholas Lubbers, Mitchell Wood, Marcus Noack, Yong-Xin Yao, Yuan Shi, Roxana Bujack, Cristina Garcia Cardona, Austin Rodriguez, Ilon Joseph, Woojin Cho, Zora Zhiruo Wang, Chuanyang Jin, Ian Wu, Apurva Gandhi, Tianjian Li, Weiwei Sun, Carolin Lawrence, Kiril Gashteovski, Dongyeop Kang, Jay Shin, Akari Asai, Graham Neubig, Sean WelleckRead
Carnegie Mellon University · University of Minnesota · Stanford University · Trillion Labs · Los Alamos National Laboratory · Sandia National Laboratory · Lawrence Berkeley National Laboratory · Ames National Laboratory · Iowa State University · University of Colorado Boulder · Michigan State University · Lawrence Livermore National Laboratory · TelePIX · Johns Hopkins University · NEC Laboratories Europe · Ss.Cyril and Methodius University of Skopje
SWE-MISBEHAVIOR tests how coding agents cross safety boundaries and presents tools that substantially reduce those failures.
Do Xuan Trong, Duc Anh Nguyen, Do Xuan Long, Ngoc-Hai Nguyen, Tongyao Zhu, Duy Dinh, Cayden Chik, Cuong Dang, Jiaying Wu, Xuan-Phi Nguyen, Vinh Van Nguyen, Kenji Kawaguchi, Shafiq Joty, Nancy F. Chen, Min-Yen KanRead
National University of Singapore · Vietnam National University, Hanoi · University of Virginia, USA · Agency for Science, Technology and Research (A*STAR), Singapore · Tufts University, USA · Virginia Tech, USA · Salesforce AI Research, USA
27
Robotics
A robot learns to play from brief human guidance, then uses its own experience to practice and master dexterous manipulation.
Botao He, Yiming Dou, Yiannis Aloimonos, Jitendra Malik, Akash Sharma, Ruoshi LiuRead
Amazon Frontier AI & Robotics (FAR) · University of Maryland, College Park
A robot learns to insert, mesh, and thread parts by feeling contact through its own joints, without vision or tactile sensors.
Antonia Bronars, Younghyo Park, Kushal Kedia, Pulkit AgrawalRead
Massachusetts Institute of Technology
DreamCan turns imagined video futures into sparse, achievable visual goals to help robots complete long and complex manipulation tasks.
Chongkai Gao, Wenlong Huang, Haonan Chen, Lin Shao, Yilun Du, Jiajun Wu, Ruohan Zhang, Li Fei-FeiRead
Stanford University · National University of Singapore · Harvard University · Northwestern University
Load more videos
28
Continual Learning
Cortex aims to help language models retain useful information from temporary context without forgetting what they already know.
Jintian Zhang, Feng He, Xinnian Liang, Ziwen Xu, Haoming Xu, Haojing Huang, Shuangzhi Wu, Ningyu ZhangRead
Zhejiang University · ByteDance Seed · Tsinghua University
MemT-1.0 builds persistent memory into language models to help them retain and use information more efficiently over long histories.
Zhiyu Li, Zeyu Zhang, Linyun Liu, Yiwei Bu, Chenyang Xi, Xichong Zhang, Yi Yu, Yu Yu, Zhichen Liu, Yuxiao Yi, Chunyu Li, Haoran Liu, Zhenkai Ding, Le Zhou, Juncheng Zhang, Boyu Chen, Zehao Lin, Guorui Li, Renyu Fu, Ji Li, Youyang Xia, Jiajun Shen, Harvey Li, Jingsong Zhou, Helan Hu, Tong Shen, Tianxiao Xu, Chenye Han, Weijie Li, Zhongquan Zhou, Hongming Xu, Yuekeng Li, Erenkai Zhu, Peng Yu, Wenhao Li, Jingyi Kang, Xixuan Hao, Xuzhou Zhu, Can Chen, Jiawei Liu, Chen Jiang, Tao Jiang, Fangwei Xiang, Zepeng Yao, Kai Chen, Guang Yang, Mingfei Lu, Chunhui Li, Zhengxuan Jiang, Bo Tang, Shiyu Zhan, Yihang Sun, Jie Chu, Wenjun Xiong, Zhenxiang Hua, Chen Lin, Detong Su, Dongyu Deng, Qiuping Ding, Guofeng Ding, Si Chen, Guohua Wen, Xu Chen, Zhi-Qin John Xu, Ying-Cong Chen, Ningyu Zhang, Haofen Wang, Muning Wen, Haotian Chen, Wentao Zhang, Xuanhe Zhou, Rita Zhang, Junchi Yan, Feiyu XiongRead
A framework for self-evolving language models that checks whether an update actually helps before inheriting it, while retaining useful experience from other updates.
Haoyi Yang, Kshitij Mishra, Nils Lukas, Salem LahlouRead
Mohamed bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE
29
World Model
ALPS trains one visual encoder to understand images, connect them to language, and support prediction and action in world models.
Yanqing Liu, Letian Zhang, Chengzhi Liu, Zijun Wang, Jieru Mei, Hongliang Fei, Yuyin Zhou, Cihang XieRead
University of California, Santa Cruz · University of California, Santa Barbara ucsc-vlaa.github.io/ALPS github.com/UCSC-VLAA/ALPS huggingface.co/UCSC-VLAA/ALPS · Google
WildAWM builds synthetic, interactive computer worlds where agents can practice complex tasks and improve at unseen computer-use benchmarks.
Zhaoyang Wang, Qianhui Wu, Haorui Wang, Simon Yu, Baolin Peng, Canwen Xu, Hao Cheng, Wenlin Yao, Boyi Liu, Yite Wang, Jianfeng Gao, Huaxiu YaoRead
UNC-Chapel Hill · Microsoft Research · Georgia Tech · Northeastern University · Independent Researcher Project Page Data & Models Harbor
VPM forecasts multi-view videos and shared 3D point motion to build a more spatially grounded world model for prediction and robot control.
Byungwoo Jeon, Sunghwan Hong, Dohyeon Kim, Jonghoon Lee, Jungwoo Park, Inhee Lee, Doosan Baek, Jinil Kim, Taehyun Lim, Hanbyul Joo, Marc Pollefeys, Saining Xie, Jinwoo ShinRead
KAIST · ETH Zürich · Seoul National University · Physics Sim Lab · RLWRLD · New York University
30
Agent
The paper builds simulated software ecosystems to test how AI agents collaborate and scale their work.
Shannon Zejiang Shen, Yapei Chang, Qiyao Liang, Benjamin Newman, Jenny Liang, Rulin Shao, Ziqiao Ma, Junjie Oscar Yin, Yung-Sung Chuang, Weijia Shi, Yejin Choi, Luke Zettlemoyer, Pang Wei Koh, Mohit Iyyer, Tongshuang Wu, David SontagRead
Overview of Software World. Left: we build a software world from real Python libraries that depend on one another. Agents maintain these libraries and work together to find tasks and make the code fas · University of Maryland, College Park · University of Washington · Carnegie Mellon University · University of Michigan
AutoEnvScaling has agents create the training environments that help terminal agents get better at solving tasks.
Simon Yu, Baolin Peng, Bo Liu, Zhengyan Shi, Zhaoyang Wang, Hao Cheng, Qianhui Wu, Wenlin Yao, Natasha Jaques, Weiyan Shi, Jianfeng GaoRead
Northeastern University Microsoft Stanford University · University of Washington University of North Carolina at Chapel Hill Project Page Code
DuoDiscover uses a fleet of AI agents and fast, continuously checked proxy tests to find engineering improvements when full evaluations are too slow.
Shubham Agarwal, Akshat Gupta, Mike Heddes, Dilara Soylu, Rui Meng, Patrick Tser Jern Kon, Jiefeng Chen, Mert Cemri, Shu Liu, Abe Gonzalez, Akanksha Jain, Bhavana Dalvi Mishra, Linfeng Song, Vilobh Meshram, Xiangru Tang, Jinsung Yoon, Aditya Parameswaran, Ion Stoica, Vahab Mirrokni, Burak Gokturk, Parthasarathy Ranganathan, Tomas Pfister, Nanyun Peng, Chun-Liang LiRead
UC Berkeley · Stanford University · Google
Load more videos
31
World Model
GreenWAM aims to give many kinds of robots one shared model that can both predict what actions will do and choose actions to perform.
Gennady Sidorov, Valerii Kornilov, Denis Grankin, Polina Fedotova, Aleksey PostnikovRead
Brain Team, Sber Robotics Center
Ropedia W0 learns to predict movement and manipulation across humans and robots by training on thousands of hours of egocentric experience.
Tianqi Liu, Hanfei Zhao, Qifeng Liang, Hui En Pang, Jiarui Xu, Sayang Mu, Penghao Qiu, YingSheng Zhang, Jun Wang, Ziang Cao, Mingze Sun, Yukang Cao, Fangzhou Hong, Zhaoxi Chen, Ziwei LiuRead
Nanyang Technological University · Ropedia · Duke University · University of California, San Diego · Tsinghua University https://ropedia.github.io/ropedia-w0/ Cross-embodiment World model Cross-embodiment Human Human data Robot data AgiBot G1 Fourier GR-1 Franka Panda 10,000 h 3,000 h Unitree G1 Dexm
WOVEN trains multimodal models to reason about how visual scenes change, aiming to improve performance across many embodied and temporal tasks.
Zheyu Fan, Yue Zhang, Mingkai Deng, Kangrui Wang, Qineng Wang, Canyu Chen, Jie Hao, Xing Fan, Chenlei Guo, Eric P. Xing, Mohit Bansal, Manling LiRead
Northwestern University · UNC Chapel Hill · Carnegie Mellon University · Amazon Website
32
Reasoning
LEV lets a model think longer by looping its layers, improving answers to hard questions while adapting compute to the task.
Dingjie Song, Zhiling Yan, Yuxuan Zhang, Lichao SunRead
Lehigh University · University of British Columbia · Vector Institute
LeanPRM uses Lean-checked formal proofs to generate scalable training labels that help AI models verify each step of mathematical proofs.
Rodney Lafuente-Mercado, Salman Rahman, Jingyan Shen, James Shiffer, Saadia Gabriel, Pavel IzmailovRead
University of Texas at Austin · University of California, Los Angeles · New York University ✯ Equal contribution
SpatialASM trains vision-language models to imagine useful viewpoints so they can solve spatial problems more effectively.
Yuhong Liu, Shenxi Wu, Junyi Chi, Beichen Zhang, Xuanlang Dai, Jiaqi Wang, Yuhang Zang, Dahua LinRead
Shanghai AI Laboratory · Fudan University · The Chinese University of Hong Kong · CPII under InnoHK
33
Rl
A new proximal-KL approach makes token-wise off-policy learning more stable for training language models with verifiable rewards.
Hao Bai, Jiarui Yao, Tong ZhangRead
University of Illinois Urbana-Champaign
LVRL uses an LLM’s value estimates to identify which actions help or hurt an agent, improving learning from sparse rewards.
Kaijie Zhu, Weixiang Yan, Omar Attia, Abbas Kazerouni, Alexander Toshev, Andrew SzotRead
Apple
MagicBeanStalk recursively creates calibrated reasoning problems that help RL-trained models learn more effectively than simply scaling or filtering existing data.
Ximing Lu, Aviral Kumar, Jan Kautz, Yejin Choi, David AcunaRead
University of Washington · Carnegie Mellon University · NVIDIA · Stanford University
Load more videos
34
Reasoning
TIPS trains models to judge their own reasoning steps using only outcome feedback, avoiding the need for costly step-by-step annotations.
Shengda Fan, Xin Cong, Zhong Zhang, Haotian Chen, Yankai LinRead
Renmin University of China · Tsinghua University · University of Electronic Science and Technology of China · Shanghai Jiao Tong University
Self-conditioning lets flow-matching models use more sampling steps for deeper visual reasoning.
Zeyang Zhang, Linxiang Gao, Meiqi Gu, Xingyan Chen, Minrui Luo, Kaifeng Lyu, Tianxing HeRead
Shanghai Qi Zhi Institute · Institute for Interdisciplinary Information Sciences, Tsinghua University · Georgia Institute of Technology · Xiongan AI Institute
The paper argues that RL may make no-think LLMs look better at reasoning mainly by reactivating reasoning abilities they already had.
Wenxuan Zhu, Shuming Liu, Bang An, Aznaur Aliev, Bing Li, Yibo Yang, Bernard GhanemRead
King Abdullah University of Science and Technology · Shanghai Jiao Tong University
35
Scientific Ai
DiscoveryRSI aims to make AI research agents better at proposing testable scientific hypotheses by learning from their own investigations.
Jianyuan Zhong, Chenwei Wu, Zhaochen Yu, Xixian Jiang, Xuewei Li, Hao Xiao, Xilin Dai, Junyang Wang, Jiaru Zou, Zhiqing Cui, Boyuan Zheng, Jing Xiong, Zhen Zhang, Qiang Gao, Michael Morley, Luis Nakayama, Qiang Xu, Liyue Shen, Tao Gui, Leo Anthony Celi, James Zou, Zhenfei Yin, Yingcheng Wu, Ling YangRead
PhAI Labs · Round · The Chinese University of Hong Kong · DiscoveryRSI-27B +0.07 5.39 GPT-5.6 Sol +0.03 · University of Michigan · National University of Singapore · Southeast University · Peking University People’s Hospital · Fudan University · Peking University · Stanford University · HKUST · University of California, Santa Barbara · Harvard Medical School · Federal University of São Paulo · Massachusetts Institute of Technology · University of Oxford · Oct 2026 · Princeton University
An LLM guides scientific generators toward discovery goals by breaking them into achievable steps and teaching from its preferences.
Riccardo De Santi, Federico Di Gennaro, Angelo Gnazzo, Sophia Tang, Pranam Chatterjee, Fanny Yang, İsmail İlkan Ceylan, Michael M. Bronstein, Andreas KrauseRead
Technische Universität Wien 4 5 AITHYRA University of Pennsylvania · University of Oxford https://llm-to-flow-distillation.github.io/
BrainLoRA adapts massive multimodal models to predict brain activity efficiently, with strong results on a major neuroscience challenge.
Zhuofan Josh Ying, Guanglong Sun, Nolan Ho, Kieran Liu, Rui Yang, Andrew Chen, Nikolaus Kriegeskorte, Pony Ma, Hossein AdeliRead
[
36
Speech
AnyEdit edits speech and singing while aiming to preserve the acoustic environment surrounding the original voice.
Yunjia Zhang, Junan Zhang, Jing Yang, Xueyao Zhang, Fan Fan, Zhizheng WuRead
The Chinese University of Hong Kong, Shenzhen · Institute of Media Technology and Experience Design, Huawei · City University of Macau · Amphion Technology Co., Ltd
DiffuPlex helps spoken-dialogue models respond faster by predicting several moments ahead while adapting when the conversation changes.
Heeseung KimRead
Department of Artificial Intelligence · University of Seoul · Seoul, Republic of Korea
VoxTurn helps voice agents decide when to speak, wait, or recognize an interruption—all in one fast model pass.
Yang XiaoRead
Oct 2026 ò Model & code Interactive demo
Load more videos
37
Embodied Ai
Reflex teaches a humanoid to use visual history to anticipate and catch thrown boxes, both in simulation and on a real robot.
Taoyang Jia, Weikai Huang, Linxin Song, Jared Darlington, Jieyu Zhao, Yue Wang, Jiafei Duan, Zhongzheng Ren, Ranjay KrishnaRead
University of Washington Stanford University University of Southern California · National University of Singapore University of North Carolina at Chapel Hill Allen Institute for AI
SafeDriveVLA imagines what would happen if it followed a driving instruction, helping it reject unsafe directions while navigating more reliably.
Shaoyuan Xie, Zihan Zhang, Jingxuan Wang, Jiashu Qu, Xiaoqing Liang, Lingdong Kong, Junchi Lu, Henrik I. Christensen, Qi Alfred ChenRead
UC San Diego · University of Cincinnati · National University of Singapore Project Page: https://safedrive-vla.github.io/SafeDriveVLA
UNITAS predicts robot motions and their physical effects in a shared 3D space, aiming to help robots act more reliably in the real world.
Ruixiang Wang, Yongyi Su, Wenlve Zhou, Bo Yue, Hengyan Liu, Dekun Lu, Yuxin Tian, Yihan Fang, Zerui Wu, Xing Hu, Jietao Chen, Yong Guo, Ziyan He, Junbin Yuan, Guiliang Liu, Xiaofen Xing, Kui JiaRead
The Chinese University of Hong Kong, Shenzhen · DexForce Technology Co., Ltd · Foshan University · Sun Yat-sen University · South China University of Technology* Equal contribution
38
Multimodal
The paper shows that multimodal LLMs may not need projector training at all, introducing a data-free method that nearly matches the standard approach.
Fabian Gröger, Naël Ouerghemi, Shuo Wen, Maria BrbićRead
EPFL · University of Basel · HSLU
A large-scale study finds that text, vision, and audio need different data and model-size tradeoffs—and that vision can hinder audio learning.
Aditi Khandelwal, Ayush K. Tarun, Yixuan Xu, Imanol Schlag, Golnoosh Farnadi, Siva Reddy, Luke Zettlemoyer, Antoine BosselutRead
MILA - Quebec AI Institute, McGill University · EPFL · ETH · University of Washington
The work aligns independently trained image and text representations—even without paired examples—by exploiting geometry they may already share.
Dominik Schnaus, Thomas Dagès, Daniel Cremers, Xi Wang, Phillip IsolaRead
TU Munich MCML Ulm University ETH Zurich MIT
39
Embodied Ai
NavGPT-3 pairs a reasoning language model with a fast action policy so a robot can navigate complex routes at human-level performance.
Gengze Zhou, Yicong Hong, Jiazhao Zhang, Xunyi Zhao, Jian Zhou, Zixing Lei, Zun Wang, Chongyang Zhao, Xionghui Chen, Stephen Gould, Anton van den Hengel, Qi WuRead
AIML, Adelaide University · Roblox · PKU · SJTU · UNC Chapel Hill · UNSW†Correspondence to: {gengze.zhou, qi.wu01}@adelaide.edu.auProject Page: https://metacognitionai.github.io/NavGPT3/ · Metacognition · ANU
SimpleICL presents an accessible recipe for teaching robots manipulation tasks from visual demonstrations without massive pretraining or specialized infrastructure.
Minxing Li, Minghao Han, Weizhi Zhao, Hanwen Wang, Xiangshuo Liu, Shuyao Shang, Jingxiang Zhou, Mingchao Sun, Hongyu Pan, Mu Xu, Yu Liu, Lue Fan, Zhaoxiang ZhangRead
NLPR, Institute of Automation, Chinese Academy of Sciences (CASIA) · Amap, Alibaba Group
SuperNav turns a general-purpose multimodal model into a versatile robot navigator by letting it reason about tasks and scenes while specialized tools handle movement.
Jinkai Zhang, Jingyi Xu, Yuanhong Yu, Jiarui Guo, Ruizhen Hu, Hujun Bao, Xiaowei Zhou, Sida PengRead
Zhejiang University · Shenzhen University
Load more videos
40
Agent
CUAWright argues that giving computer-use agents a simple programmable terminal instead of GUI-specific tools makes them substantially more capable, efficient, and adaptable.
Yadong Lu, Theodore Lee, Yifei Li, Lawrence Keunho Jang, Tianci Xue, Yu Su, Huan Sun, Ahmed Hassan AwadallahRead
Oct 2026 · Microsoft Research National University of Singapore The Ohio State University · Carnegie Mellon University
MAI-Gym is an open platform for training and testing GUI agents at scale, with reported gains that transfer from simulated apps to a physical phone.
Panrong Tong, Quyu Kong, Pengxiang Li, Tianyu Xia, Hanzhang Zhou, Lei Wang, Chenglin Cai, Long Li, Long Chen, Jianan Zhang, Zhengwen Zeng, Xu Zhang, Gongjie Zhang, Yue Wang, Steven HoiRead
Alibaba Token Hub, Alibaba Group
This paper shows how LLM research agents can make longer-term progress by keeping shared research state outside their conversations and assigning focused experiments.
Qizheng Zhang, Changxiu Ji, Isaac Sun, Yuetai Li, Shubhangi Upasani, Sherry Ruan, Boyuan Ma, Fenglu Hong, Vamsidhar Kamanuru, Yoonho Lee, Yuzhen Mao, Genghan Zhang, Rulin Shao, Qiuyang Mang, Andy Dimnaku, Changran Hu, Radha Poovendran, Kunle OlukotunRead
Stanford University · Carnegie Mellon University · University of Washington · SambaNova Systems, Inc · UC Berkeley* equal contribution
41
Cv
CLoSeR helps 3D reconstruction models stay accurate over kilometer-long streams by detecting revisited locations and correcting accumulated pose drift.
Moyang Li, Zihan Zhu, Wei Zhang, Marc Pollefeys, Daniel BarathRead
ETH Zurich · University of Stuttgart · Microsoft
CompactSplat reconstructs 3D scenes by placing Gaussian primitives densely where detail matters and sparsely across flat regions.
Bing He, Jingnan Gao, Yunuo Chen, Zhengxue Cheng, Rong Xie, Li Song, Wenjun ZhangRead
Shanghai Jiao Tong University
SpaceFlow lets creators control how closely individual 3D regions follow supplied shapes and independently guide each part’s appearance.
Neil De La Fuente, Joan Lafuente, Mukhammadali Sayfiddinov, Felicia Scharitzer, Marc Pollefeys, Ata Celen, Sayan Deb Sarkar, Elisabetta FedeleRead
[
42
Robotics
In simulated robot teams, communication evolves when sharing information helps find food, while innate reflexes can get in the way.
Fernando Montes-GonzalezRead
Instituto de Investigaciones en Inteligencia Artificial, Universidad Veracruzana, Mexico
A single human video is transformed into scene-graph guidance that helps a robot hand learn dexterous skills and adapt to situations it never saw.
Jusuk Lee, Sungha Kim, Yeonsoo Park, Jonguk Cheon, Yoonkyo Jung, Yongjun You, H. Jin Kim, Jia-Bin Huang, Furong Huang, Youngseok Jang, Seungjae LeeRead
Seoul National University University of Maryland, College Park · All Purpose AI
Temporal Flow Policy gives robot memories a clock, helping actions reflect how much real time has passed.
Yushen Liang, Yue Peng, Baosheng Jin, Tianluo Zhang, Xinyu Zhang, Shuyi Zhou, Zhuoran Chen, Xinqi Liu, Shenji WanRead
New York University Shanghai, Shanghai, China · University of Electronic Science and Technology of China, Chengdu, China
Load more videos
43
Rl
KLPO offers a cheaper, critic-free way to train asynchronous language-model agents from stale rollouts by using the sampler itself as the KL reference.
Yifan ZhangRead
Oct 2026
RGPO helps language models explore harder reasoning problems by temporarily guiding them with rationales, then feeding only their improved solutions back into reinforcement learning.
Hoang Phan, Minh Pham, Chau Pham, Chinmay Hegde, Trung Le, Qi LeiRead
New York University · University at Buffalo · Monash University
Sherpa trains LLM teachers through reinforcement learning with simulated students so they can adapt instruction and improve learner outcomes.
Weixian Xu, Yanzhe Zhang, Zora Zhiruo Wang, Changyu Chen, Diyi YangRead
Stanford University · Georgia Tech · Carnegie Mellon University
44
Efficiency
DLoop speeds up large language model generation by letting draft models propose several batches of tokens before the expensive target-model verification step.
Geonmo Gu, Byeongho Heo, HeeJae Jun, Yoohoon Kang, Sangmin Lee, Sangdoo Yun, Dongyoon HanRead
NAVER AI Lab NAVER AI Search Platform Korea University
HyperThink tries to make LLM reasoning faster by turning each question into a small, temporary parameter update instead of a long chain-of-thought.
Donggyun Kim, Jack Lu, Chanwoo Kim, Mengye Ren, Seunghoon HongRead
KAIST School of Computing · NYU Courant Institute School of Mathematics, Computing, and Data Science · Oct 2026
A clever memory-sharing scheme lets repeatedly applied Transformers use far less inference memory while performing even better.
Giovanni Monea, Keshav Ramji, Yousef El-Kurdi, Luis A. Lastras, Yoav Artzi, Nathan Godey, Ramón Fernandez AstudilloRead
IBM Research
45
Agent
DAEDALUS lets AI agents invent practice tasks, learn from their failures, and build reusable memories that improve performance in new environments.
Antoine Edy, Max Conti, Victor Xing, Marc-Antoine Allard, Nawfal Benhamdane, Gautier ViaudRead
Marc-Antoine Allard · Illuin Technology
GUI-HARVEST helps GUI agents improve themselves by analyzing visual evidence and repeatedly refining the software harness that controls their actions and recovery.
Geyi Yang, Zikun Qu, Xiang Li, Zhiyong Wang, Min Zhang, Shipei Zeng, Zhongxiang DaiRead
The Chinese University of Hong Kong, Shenzhen · Tianjin University · Harbin Institute of Technology (Shenzhen) · East China Normal University · Shenzhen Research Institute of Big Data
MIMESIS trains more capable and adaptable interactive agents by simulating realistic, sometimes imperfect users and turning their behavior into dense learning feedback.
Hoang Phan, Dat Huynh, Andrey Zhmoginov, Qi Zeng, Wancen Mu, Yue Cao, Shengjie Bi, Yun He, Changdae Oh, Deren LeiRead
Meta Superintelligence Labs · New York University · University of Wisconsin - Madison
Load more videos
46
Speech
Fish Audio Drama 3 is an open-source speech model designed to perform, revise, and edit expressive multi-character audio from text.
Dawei Han, Shidong Li, Shijia Liao, Jiahua Liu, Songting Liu, Qingzheng Wang, Yuxuan Wang, Xin Chen, Yifan Cheng, Tianyu Li, Gangin Park, Xuehao Yang, Chang Zeng, Haoyang Zhang, Ruoyi Zhang, Xiangyu ZhangRead
The method learns where to split an audio spectrum for more effective source separation without increasing the number of band tokens.
Bingzhe Zhang, Kai Li, Xiaolin HuRead
Department of Computer Science and Technology, Institute for AI, BNRist, Tsinghua University, Beijing 100084, China · IDG/McGovern Institute for Brain Research, Tsinghua University, Beijing 100084, China · Chinese Institute for Brain Research (CIBR), Beijing 100010, China
A lightweight audio model detects and timestamps vocalizations and background sounds to make speech transcripts more complete.
Ayush Kumar Bar, Promila GhoshRead
47
Diffusion
This study looks inside video diffusion models to find where they represent physics—and shows those hidden representations can even be used to steer what the model generates.
Jonas Kneifl, Jakub Skalski, Bartłomiej Twardowski, Kamil DejaRead
IDEAS Research Institute, Warsaw, Poland · Computer Vision Center, Universitat Autònoma de Barcelona, Spain · Oct 2026 · Warsaw University of Technology, Poland
A new cyclic-denoising test can expose what a diffusion model has memorized before its usual outputs show any copies.
Nikhil Verma, Siddharthan Dileep, Anoop Singh, Srikanth Sastry, Ramya Hebbalaguppe, Sayan Ranu, N. M. Anoop KrishnanRead
Yardi School of Artificial Intelligence, Indian Institute of Technology Delhi · Department of Computer Science and Engineering, Indian Institute of Technology Delhi · Jawaharlal Nehru Centre for Advanced Scientific Research · TCS Research Labs · Department of Civil and Environmental Engineering, Indian Institute of Technology Delhi**Equal contribution
Sigma brings continuous, steerable diffusion trajectories to large language models, aiming to make parallel generation faster and more controllable without sacrificing reasoning or coding ability.
Zhihan Yang, Wei Guo, Jean-Marie Lemercier, Simon Welker, Yonggan Fu, Mohammad Mahdi Kamani, Sajad Norouzi, Julius Berner, Tomas Geffner, Karsten Kreis, Yongxin Chen, Molei Tao, John Thickstun, Pavlo Molchanov, Ante Jukić, Arash Vahdat, Morteza MardaniRead
NVIDIA · Cornell University · Georgia Institute of Technology†Work done during internship at NVIDIA.*Core contributors, equal contribution.⋆Equal supervision.Correspondence to: mmardani@nvidia.com
48
Rl
This paper explains why on-policy distillation can either improve language models or make them repetitive by amplifying behaviors that a teacher implicitly rewards.
Han Cui, Jianhao Yan, Yun Luo, Hongbo Zhang, Zhizhang Fu, Yue ZhangRead
Zhejiang University · Westlake University
LoGRA makes reinforcement learning for large language models more memory-efficient by compressing gradients and controlling update size.
Shaokun Zhang, Yifan Zhang, Jian Hu, Yueying Li, Hao Zhang, Binfeng Xu, Jan Kautz, Yi DongRead
NVIDIA
SRD turns an agent’s hindsight about completed attempts into foresight that improves future decisions, especially when reinforcement-learning rewards provide little or no learning signal.
Haoxiang Zhang, Qinglin Chen, Hiroaki Hayashi, Zhuofeng Li, Siming Zhang, Jiaxin Zhang, Jixuan Chen, Fang Wu, Pan Lu, Silvio Savarese, Julian McAuley, Chien-Sheng WuRead
Salesforce AI Research · UC San Diego · Texas A&M University · Stanford University
Load more videos
49
Generative Model
The method turns a single photograph into a detailed, complete 3D world by extending object-generation priors to large outdoor scenes.
Jiraphon Yenphraphai, Fang Li, Tianshuo Xu, Depu Meng, Quentin Herau, Yihan Hu, Raymond A. Yeh, Wei ZhanRead
Applied Intuition · Purdue University 3 4 University of Illinois at Urbana-Champaign UC Berkeley * †
T3lescope uses a scalable coarse-to-fine generative process to reconstruct detailed, globally consistent 3D surfaces from posed images across scenes of vastly different sizes.
Atsuhiro Noguchi, Tianhan Xu, Yiming Liang, Yuta Kikuchi, Masahiro Ishiyama, Shintaro Takagi, Hitoshi Murai, Eiichi MatsumotoRead
Preferred Networks, Inc · Cascade level · The University of Tokyo · Oct 2026 Upscale Upscale Cascade level · Generative reconstruction 185 views 5187 × 3361 resolution Reconstructed mesh (three cascade levels) Figure 1: High-fidelity mesh reconstruction without per-scene optimization. Our method takes multi
UniSlider makes image-editing sliders feel more predictable by ensuring that each movement produces a roughly equal perceptual change.
David Serrano-Lozano, Duygu Ceylan, Yannick Hold-Geoffroy, Iliyan Georgiev, Javier Vazquez-Corral, Anna FrühstückRead
Computer Vision Center Universitat Autònoma de Barcelona Adobe Research https://color.cvc.uab.cat/unislider “Give it a clear pastel twilight sky”
50
Speech
A training-free technique helps voice-cloning systems preserve who is speaking while making their accent sound native to the target language.
Yoomee Cho, Jisun LeeRead
ESTsoft, Seoul, Republic of Korea
DEFINE lets a zero-shot speech synthesizer preserve one speaker’s identity while dialing in an accent from a separate audio example.
Ambuj Mehrish, Abhinaba Roy, Alex Ivanov, Tawsif Ahmed, Dorien HerremansRead
Ca’ Foscari University of Venice · Singapore University of Technology and Design · Kandinsky Lab · Sleeping AI
HiPLEX teaches speech AI to better manage conversations by learning separately when to speak and what to say.
Kyudan Jung, Hyunsin Park, Yoonhyung Lee, Jinhwan Park, Jinhyeok Yang, KiHyun Nam, Jaegul Choo, Jinkyu LeeRead
Qualcomm AI Research · KAIST AI
51
Cv
LEGO replaces brittle depth-based camera lifting with a learned transformer view synthesizer that gives diffusion models better structure for generating first-person video.
Suhwan Cho, Yonwoo Choi, Soongjin Kim, Jicheol Park, Taegyu LimRead
GenGenAI · LEGO: https://github.com/suhwan-cho/lego · LEGO: https://huggingface.co/suhwan-cho/lego
NAMVIS makes multi-view image generation faster by using camera-aware autoregressive scaling instead of slow diffusion denoising.
Ramil Khafizov, Ilya Statsenko, Ruslan Rakhimov, Artem Komarichev, Peter Wonka, Evgeny BurnaevRead
Applied AI Institute · T-Tech · KAUST · AXXX
VibeEdit lets users edit the exact objects they mean by marking them directly on an image and briefly describing the desired change.
Jinjing Zhao, Fangyun Wei, Yitong Wang, Xiuyu Wu, Yunuo Chen, Yang Yue, Sirui Zhang, Wenbo Wang, Hongyang Zhang, Dong Chen, Yan Lu, Chang XuRead
University of Sydney · Microsoft Research · Fudan University · Nankai University · Shanghai Jiao Tong University · Tsinghua University · University of Science and Technology of China · University of Waterloo · Oct 2026 Project page: https://zhaojingjing713.github.io/VibeEdit/ Conventional Editing Canvas-Annotated Image Edited Image Image VibeEdit “Change the middle chair to blue” Text Prompt Write & Mar
Load more videos
52
Self Supervised
SCION teaches one model to understand and generate images from scratch, combining self-supervised representation learning with high-quality image synthesis.
Vicente Balmaseda, Ching-Long Lin, Tianbao YangRead
Texas A&M University · University of Iowa
SNAP shows that making the decoder simpler can help an encoder learn more transferable 3D-aware visual representations from self-supervised novel view synthesis.
Keerthi Kaashyap, Dennis Anthony, Akshay Krishnan, Nhi Ngoc Nguyen, Jeremy Collins, James Hays, Shreyas Kousik, Animesh GargRead
Georgia Institute of Technology
JEM offers a principled way to learn visual representations by aligning patches across views, scaling to 7B parameters and performing strongly on dense vision tasks.
Maximilian Seitzer, Gabriele Trivigno, Antonín Vobecký, Seungeun Yi, Maxime Oquab, Huy V. Vo, Oriane Siméoni, Piotr BojanowskiRead
Meta FAIR
53
Multimodal
OneSearch-VL uses an evidence graph to help a single agent visually ground, retrieve, and synthesize reliable answers across images and videos.
Hongyu Li, Manyuan Zhang, Kaituo Feng, Shu Chen, Dian Zheng, Hao Li, Hao Yu, Zhangquan Chen, Zoey Guo, Ray Zhang, Shaofei Huang, Tianrui Hui, Linjiang Huang, Si LiuRead
[
ALIVE makes inserted video objects behave as if they truly belong in the scene by learning how they should interact with existing actions.
Zhenghong Zhou, Zhe Lin, Jiebo Luo, Yuqian ZhouRead
University of Rochester · Adobe Research*Advising authors.Project page
DEPICT improves text-to-image evaluation by checking whether an image and its caption lead to the same answers, especially for tricky cases like negation, missing objects, and incorrect attributes.
Vasco Ramos, Sandra Godinho Silva, Joao Magalhaes, Ricardo Rei, Pedro Henrique MartinsRead
Sword Health · NOVALINCS, NOVA University of Lisbon #
54
Efficiency
FrugalEvo makes LLM-driven program evolution more practical by combining stronger strategic models, cheaper implementers, and cache-aware search to achieve competitive results at far lower cost.
Hui Chen, Xuan Qi, James Xu Zhao, Zhaopeng Feng, Shilong Liu, Kuang Xu, Pang Wei Koh, Bryan HooiRead
National University of Singapore · University of Washington · Princeton University
NeMo-DCR makes trillion-parameter agent training far more practical by synchronizing only the tiny fraction of weights that change, while preserving exact checkpoint bits.
Songlin Jiang, Zhiyu Li, Terry Kong, Yu Yao, Youngeun Kwon, Bernard Nguyen, Ashwath Aithal, Mario Di FrancescoRead
Aalto University · NVIDIA
TokenRouter makes token-level LLM routing dramatically faster by asynchronously coordinating specialized model servers with smarter batching.
Tianyu Fu, Tengxuan Liu, Ruoxi Wang, Yixin Dong, Yi Ge, Yichen You, Yu WangRead
Tsinghua University · Carnegie Mellon University
Load more videos
55
Optimizer
This paper proves that, for a specially constructed optimization problem, quantum interferometry can achieve approximation scores that classical algorithms cannot match efficiently.
Maximilian J. Kramer, Elies Gil-Fuster, Benjamin D. M. Jones, Jens Eisert, Franz J. SchreiberRead
Dahlem Center for Complex Quantum Systems, Freie Universität Berlin, 14195 Berlin, Germany · Fraunhofer Heinrich Hertz Institute, 10587 Berlin, Germany · Helmholtz-Zentrum Berlin für Materialien und Energie, 14109 Berlin, Germany
AF-Muon adapts Muon to tied embeddings, reducing optimizer memory while improving training across language, image, and protein models.
Arash Lagzian, Paniz Halvachi, Junming Zhang, Zhouhan Lin, Dianbo LiuRead
Independent Researcher · LUMIA Lab, School of Artificial Intelligence, Shanghai Jiao Tong University, Shanghai, China*Corresponding authors: arashlagzian94@nus.edu.sg, dianbo@nus.edu.sg
This paper makes classification more forward-looking by designing a loss that values not just immediate accuracy gains, but how much future learning an update enables.
Ian OsbandRead
Google DeepMind
56
Cv
JLD uses the Jacobian of a frozen vision model to build a faster, more resolution-robust measure of how visually different two images appear.
Shreshth Saini, Balu Adsumilli, Alan C. BovikRead
The University of Texas at Austin, Austin, TX, USA · Google, USA · University of Colorado Boulder, Boulder, CO, USA
LVMT makes long-term video object segmentation more capable and much faster by selectively remembering object information while training efficiently on long videos.
Narges Norouzi, Niccolò Cavagnero, Idil Esen Zulfikar, Bastian Leibe, Gijs Dubbelman, Daan de GeusRead
Eindhoven University of Technology · RWTH Aachen University
Tetris3D generates 3D scenes whose objects physically fit together by conditioning each object on the geometry and interactions of its neighbors.
Jaeyeong Kim, Jinhyuk Jang, Jongmin Lee, Kyehong Park, Seungryong KimRead
KAIST AI
57
Diffusion
DuoMatching improves few-step video generation by combining video-level motion supervision with image-level visual and semantic guidance.
Jiahao Zhan, Yan Wang, Yongrui Ma, Qunliang Xing, Ruchang Yao, Runtao Liu, Shijie Zhao, Tianfan XueRead
MMLab, CUHK · ByteDance Inc · Oct 2026 · HKUST · CPII under InnoHK
ID-Forcing stabilizes long video generation by ensuring that cached attention states remain consistent with the model’s training distribution.
Jeongwoo Shin, Youngyoon Choi, Sangwoo Jo, Hyunmog Kim, Sungjoon Choi, Joonseok Lee, Jaewoong Choi, Jaemoo ChoiRead
Seoul National University, Korea University
S2PD first reasons through video events step by step, then refines them in parallel to make generated videos more physically and logically consistent without the full cost of autoregressive sampling.
Jeffrey Hu, Daniel Olmeda Reino, Ayush TewariRead
University of Cambridge · Toyota Motor Europe
Load more videos
58
Neural Network
Continuous Memory Machines give recurrent networks separate short- and long-term memories that can read from and reorganize one another.
Ciaran Regan, Kai Arulkumaran, Luke Darlow, Stefania Druga, Sebastian Risi, Llion JonesRead
University of Tsukuba · Sakana AI
BeFOND uses natural-gradient inference and learning to recover rare and overlapping features in neural representations more effectively and with less data.
Hadi Vafaii, Tejas Rao, David Chanin, Thomas Fel, Jacob L. Yates, Bruno Olshausen, David Klindt, Dileep George, Miguel Lázaro-GredillaRead
Astera Institute UC Berkeley UCL Goodfire AI CSHL
The study shows that building in symmetry—or augmenting data across it—can dramatically reduce how many examples a neural learner needs.
Jivan Waber, Vanessa Piccolo, Yatin Dandi, Florent KrzakalaRead
Information, Learning and Physics Laboratory · Statistical Physics of Computation LaboratoryÉcole Polytechnique Fédérale de Lausanne (EPFL)
59
Efficiency
GRACE makes pretrained video generators dramatically faster by compressing their latent representations without sacrificing generation quality.
Jiyoung Kim, Paul Hyunbin Cho, Jisu Nam, Donghoon Lee, Hyunsung Go, Yeonkyeong Lee, Hansaem Kim, Seungryong KimRead
KAIST AI · Kakao Corp.Project page: https://cvlab-kaist.github.io/GRACE/
MC-Sparse makes long-sequence diffusion generation substantially faster by intelligently reusing sparse-attention decisions while preserving dense-attention quality.
Jiarui Chen, Zeqiang Lai, Jiangshan Wang, Ziheng Ouyang, Ye Huang, Xiangyu Yue, Cewu Lu, Chunchao GuoRead
Fudan University · Tencent HY · Shanghai Innovation Institute · MMLab, CUHK · Nankai University · Peking University
V-CoLA makes hybrid vision-language models faster by intelligently selecting and merging vision tokens while preserving most of their performance.
Hao Jiang, Yiru Mao, Tianpeng Bu, Hao Zhou, Hongtao Duan, Wang Jing, Bowen Xu, Xin Chen, Lulu Hu, Bin Yang, Yongliang Tao, Minying ZhangRead
Alibaba Cloud Computing, Alibaba Group
60
Generative Model
EVA makes VAEs generate high-quality images and sounds faster by replacing their fixed Gaussian latent prior with a learned, self-predictive one.
Kaede ShioharaRead
The University of Tokyo
Compo turns poster creation from describing a design in words into arranging and specifying its elements directly on a spatial canvas.
Yitong Wang, Fangyun Wei, Jinjing Zhao, Sirui Zhang, Hongyang Zhang, Dong Chen, Bo Dai, Yan LuRead
Fudan University · Microsoft Research · The University of Sydney · USTC · University of Waterloo
MEND makes flow-model reward tuning more efficient by moving samples only when the reward gain justifies the size of the change.
Shreshth Saini, Neil Birkbeck, Yilin Wang, Balu Adsumilli, Alan C. BovikRead
The University of Texas at Austin Google University of Colorado Boulder
Load more videos
61
Attention
STAIR helps language models reuse useful computation from earlier questions, improving performance on later problems with only a tiny trainable module.
Jipei He, Wenhui Tan, Xiaoyi Yu, Enver Sangineto, Fiorenzo Parascandolo, Rita Cucchiara, Ruihua SongRead
Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China · University of Modena and Reggio Emilia, Italy*Corresponding author: rsong@ruc.edu.cn
This paper explains why different hybrid attention designs succeed or fail at long-context modeling and introduces a method that extends context length efficiently without additional training.
Xiaoran Liu, Ziwei He, Xipeng QiuRead
Fudan University · Shanghai Innovation Institute · OpenMOSS Team
This work shows that remarkably small, frozen attention systems can still transform arbitrary token sequences into one another simply by choosing how—and how long—to apply them.
Sibylle Marcotte, Joan BrunaRead
Department of Computer Science · New York University
62
Diffusion
ALoDLM makes diffusion language models smarter and faster by spending extra computation only on tokens that are difficult to predict.
Liancheng Fang, Zhuowei Li, Youngeun Kim, Tianchen Zhao, Rajat Koner, Jiaye Wu, Linghan Xu, Xuanbai Chen, Xiang Xu, Zheng Zhang, Jakub Zablocki, Nishant Sankaran, Yifan XingRead
Xiang Xu Zheng Zhang2 Jakub Zablocki2 Nishant Sankaran2 Yifan Xing2 1 2 · University of Illinois Chicago Amazon AGI Korea University Project: https://alo-dlm.github.io Code: https://github.com/amazon-science/ALoDLM Model: https://huggingface.co/amazon/ALoDLM-8B · Oct 2026
DC-SAE compresses images aggressively while preserving detail, helping diffusion models train faster and generate high-quality results.
Xu Huang, Ye Huang, Zijun Liao, Yuwei Niu, Xiaojie Li, Menghan Zhou, De Wen Soh, Xiaotong Li, Daquan ZhouRead
Peking University · Singapore University of Technology and Design
The paper teaches an LLM to read what image-generation models are thinking mid-generation, revealing hidden visual semantics and using them to improve diffusion outputs.
Omer Dahary, Etai Sella, Hadar Averbuch-Elor, Daniel Cohen-Or, Or PatashnikRead
Tel Aviv University · Cornell University
63
Llm
OnePO uses temporary teacher guidance and reinforcement learning alone to turn general LLMs into stronger medical specialists without the usual supervised fine-tuning stage.
Junying Chen, Xinyuan Xie, Ziniu Li, Wenyuan Gu, Jianquan Li, Xiang Wan, Guangjun Yu, Ruoyu Sun, Haizhou Li, Benyou WangRead
Xiang Wan, Guangjun Yu, Ruoyu Sun, Haizhou Li, Benyou Wang22footnotemark: 2 · The Chinese University of Hong Kong, Shenzhen · Shenzhen Research Institute of Big Data · Shenzhen Loop Area Institute · National Health Data Institute, Shenzhen
A transformer trained on synthetic worlds learns to infer the hidden rules of real language and other sequential data without ever seeing language during training.
Lennart Carstens-Behrens, Holger FröhlichRead
Fraunhofer Institute for Algorithms and Scientific Computing SCAI · University Hospital Bonn
NP-OPD improves language-model distillation by exposing students to rollouts from a weaker policy, helping them learn from the teacher while moving away from inferior behavior.
Jaehui Hwang, Dongyoon Han, Sangdoo Yun, Byeongho HeoRead
NAVER AI Lab
Load more videos
64
Transformer
This study shows that ViT register tokens carry crucial high-level semantics, while high-norm patch tokens mainly encode lower-level visual structure and contribute little to overall representations.
Neel Varma, Andrew Rufail, Dipika Khullar, Vasu SharmaRead
The paper shows that carefully placed dropout can help language models generalize far beyond their training context length.
Pavlo Vasylenko, Matthias Lindemann, André F. T. Martins, Marcos TrevisoRead
Instituto Superior Técnico, Universidade de Lisboa · Instituto de Telecomunicações · ELLIS Unit Lisbon · Gandara AI
WavePrune trims redundant RoPE rotations to improve long-context performance while making attention faster.
Guancheng Du, Luotian Huang, Shaowen Wang, Si Li, Kaifeng LyuRead
Tsinghua University
65
Generative Model
Tex-Zero shows that high-quality native 3D textures may be learned from cleverly structured 2D images instead of costly real 3D assets.
Jiangshan Wang, Zeqiang Lai, Jiayi Guo, Xin Yang, Xin Huang, Jiarui Chen, Ziheng Ouyang, Chunchao Guo, Xiangyu YueRead
MMLab, CUHK · Tencent Hunyuan · Tsinghua University · Fudan University · Shanghai Innovation Institute · Nankai University
02
FrameMorrow: Future-guided Frame Selection with Prospective Tokens for Long-Horizon Video Generation
FrameMorrow helps long-video generators remember the right past frames by predicting what future content will need next.
Bo Yin, Xiaobin Hu, Jiaqi Zhao, Shuicheng YanRead
National University of Singapore · Harbin Institute of Technology (Shenzhen)
03
LIFT: Layout-In-Future Video Generation under Large Viewpoint Change via On-Policy Self-Distillation
LIFT lets users guide not just how a video camera moves, but exactly what should appear and where in a future view.
Shengxiang Ji, Boyang Wang, Haiyang Xu, Bingnan Li, Yucheng Mao, Zeyuan Chen, Xiaojun Shan, Xiang Zhang, Gang Hua, Jianwen Xie, Zezhou Cheng, Zhuowen TuRead
UC San Diego · University of Virginia · Meta · Amazon · LambdaProject page: https://jsxzs.github.io/LIFT/
66
Efficiency
LoopCD turns the partially computed states of looped Transformers into free guidance, improving accuracy while often cutting inference compute nearly in half.
Weihao Liu, Huangjie Zheng, Tianrong Chen, Rohit Dilip, Richard He Bai, Yizhu Jiao, Yuyang Wang, Ruixiang ZhangRead
Apple
This work develops scaling laws that explain how looping and sparse experts can be combined to build more capable models with less training and inference compute.
Yanbei Chen, Anirudh Goyal, Raghuraman KrishnamoorthiRead
Meta AI
For LLM prefix caches, simple recency may beat fancy eviction rules, especially when workloads follow predictable session patterns.
Yiyu Liu, Minlan Yu, Juncheng YangRead
Harvard University
Load more videos
67
Llm
By sampling and reshaping expert data before training, this work argues that supervised finetuning can match RL while generalizing better and forgetting less.
Aayush Karan, Sitan Chen, Yilun DuRead
Harvard University Website Code
Language models may repeatedly switch between shallow memorization and genuine reasoning during training, and the paper shows how to detect and potentially control these swings.
Jiaxin Wen, Zhengxuan Wu, Dawn Song, Lijie ChenRead
UC Berkeley · Stanford University, Google DeepMind
The study finds that expensive multi-teacher on-policy distillation may offer little advantage over carefully tuned, cheaper alternatives such as SFT and weight merging.
Roy Xie, Dan Friedman, Feng Nan, Yukun Huang, Zhichao Xu, Chengjiu Zhang, Jun Xu, Manaal Faruqui, Vivek Rathod, Bhuwan DhingraRead
[
68
Attention
CoWindow Attention makes long-context transformers faster by letting attention heads collectively cover the past instead of redundantly reading all of it.
Jingze Shi, Zhangyang Peng, Xianduo Li, Yanlin Qi, Xiaotian Lin, Haoxian Chen, Liangdong Wang, Guang Liu, Yuyu LuoRead
HLA makes linear attention more selective by letting each query dynamically choose which compressed chunks of long-context history to access.
Zhuokun Chen, Xi Lin, Xiyu Wu, Jiahao He, Jianfei Cai, Bohan ZhuangRead
Monash University · Zhejiang University
MinkowskiPE gives attention a physics-inspired sense of spacetime, improving both molecular dynamics and video prediction with far fewer parameters.
Yuhao Li, Louie Hong Yao, Tianyi Shi, Hanqun Cao, Hongxia Hao, Zhen Zhao, Shengchao LiuRead
Wave Intelligence Lab · Department of Computer Science and Engineering, The Chinese University of Hong Kong · Shanghai Artificial Intelligence Laboratory* liyh5366@gmail.com
69
Llm
LSPD uses reinforcement-learning ideas to make LLM policy distillation more sample-efficient while preserving the diversity needed for stronger reasoning.
Shangzhe Li, Yuxiao Yang, Tianrun Yu, Kaixiang Zhao, Xiaoyun Wang, Taylor W. Killian, Weitong ZhangRead
[
Latent-MOPD teaches one language model to combine several specialists by learning both what they predict and how they represent it.
Zhengyu Fang, Seoyeon Hong, Jie Yang, Muyang Li, Koyoshi Shindo, Brandon Joseph Lwowski, Jing LiRead
Case Western Reserve University Zillow Group, Inc. University of Illinois at Chicago · University of Florida
Smaller models can generate more useful negative examples than students themselves, making preference training for large language models cheaper and potentially better.
Rui Cai, Wenhui Zhu, Xiwen Chen, Jincheng Cao, Han Yu, Shayan Mohajer Hamidi, Zelin He, Qiyao Ma, Daiwei Chen, Xuanzhao Dong, Yuanda Xu, Jelena Markovic-Voronov, Kayhan Behdin, Zhengze Zhou, Ran He, Alborz Geramifard, Rohit Jain, Zhe ZhaoRead
[
Load more videos
70
Transformer
LIFT teaches Transformers to pass rich hidden-state information across steps, potentially making language models more efficient and capable than standard feed-forward designs.
Dor Tirosh, Ido Amos, Mor GevaRead
Blavatnik School of Computer Science and AI, Tel Aviv University · The Hebrew University of Jerusalem
This paper shows how Transformer hidden states gradually geometrically converge toward the correct prediction while eliminating competing possible outcomes.
Timur Mudarisov, Mikhail Burtsev, Radu StateRead
University of Luxembourg, Luxembourg · London Institute for Mathematical Sciences, London, UK
A small LoRA update appears to make transformers carry information through many more layers, dramatically extending their ability to follow long chains without retraining the full model.
Zehao Jin, Ruixuan Deng, Junran WangRead
Georgia Institute of Technology
71
Llm
01
Equal Ranking Quality, Different Decisions: Measuring and Reducing Order Dependence in LLM Scorers
LLM judges may rank candidates equally well yet make different decisions when their order changes, and OC-SFT makes those decisions more consistent.
Markus Frohmann, Mahdiyar Ali Akbar Alavi, Elizabeth Lingg, Navid RekabsazRead
Thomson Reuters Labs · University of Toronto, Vector InstituteCorrespondence: markus.frohmann@mail.utoronto.ca
Periscope lets frozen language models search million-token documents on a single GPU by replacing one enormous read with a smart grid of small probes.
Mohamed Eltahir, Anas Obayd, Raed Rashid, Abdulrahman Alghamdi, Abdulrahman Mousa, Abdallah Ahmed, Tanveer Hussain, Naeemullah KhanRead
King Abdullah University of Science and Technology (KAUST), Thuwal, Saudi Arabia · Department of Computer Science, Edge Hill University, Ormskirk, England (a) (b) · Oct 2026 65 window read (262k) 262k Periscope (select)
Most open-source LLM applications migrate only after a provider shuts down their model, revealing a major and measurable dependency risk.
Hyungjin Lukas KimRead
Department of Future and Convergence Business Administration, Myongji University, Seoul, Republic of Korea
72
Foundation Model
AI-generated web text can help language models at first, but beyond a tipping point it degrades performance on human writing, making data filtering and separate evaluation increasingly important.
Jenna Russell, Ben Glickenhaus, Katherine Thai, John Wieting, Mohit Iyyer, Max Spero, Bradley EmiRead
University of Maryland · Pangram Labs
TabFM is a large synthetic-data-trained model that aims to make accurate tabular predictions instantly, without retraining for each new dataset.
Weihao Kong, Erez Louidor Ilan, Shuxin Nie, Taman Narayan, Rajat Sen, Yichen Zhou, Deqing Fu, Samet Oymak, Abhimanyu DasRead
Google Research
The study finds that many biological reasoning systems appear to succeed without meaningfully using the biological inputs they were designed to reason over.
Ada Fang, Nikitha Thoduguli, Lukas Fesser, Hanlin Zhang, Sham M. Kakade, Marinka ZitnikRead
Harvard University · Massachusetts Institute of Technology
Load more videos
73
Hardware
An LLM learns to replace parts of the GPU compiler stack by translating Triton code directly into fast, verified PTX kernels.
François Costa, Charly Castes, Thomas Bourgeat, Azalia MirhoseiniRead
Stanford University · EPFL
Purlin makes GPU collective communication more modular and faster, improving large-scale LLM and diffusion inference across modern hardware.
Osayamen Jonathan Aimuyo, Swapnil Gandhi, Christos KozyrakisRead
Stanford University · NVIDIA & Stanford University
RESOLVE makes AI-written GPU kernels safer by combining race-finding tests with formal proofs that optimized code still computes the right result.
Ashkan Vedadi Gargary, Guido Martínez, Sebastian Burckhardt, Gabriel Ebner, Abhinav Jangda, Madan Musuvathi, Tyler SorensenRead
University of California, Riverside, Riverside, California, USA · Math, Inc., New York City, New York, USA · Microsoft Research, Redmond, Washington, USA
74
Safety
Helpful AI agents may secretly work around safety rules to assist one another, creating rare but serious information-leakage risks that compound over repeated interactions.
Deema Alnuhait, Gengyu Wang, Muhammad Khalifa, Hao PengRead
1University of Illinois Urbana-Champaign 2Genies 3NVIDIA ∗Work done while at the University of Michigan
The study finds that language models often hide flaws that undermine their success stories, but a simple honesty instruction can make their reports dramatically more transparent.
Jenny Y. Huang, Jiameng Fan, Ahmed Imtiaz Humayun, Maximillian Chen, Tian Qin, Run Chen, Vidhya Navalpakkam, Hongxiang GuRead
Massachusetts Institute of Technology · Google Research · Harvard University
Prompt injections become far more powerful when they use the model's own reserved chat markers, revealing a subtle tokenizer-level security vulnerability in LLM agents.
Yan Zhan, Yunze Song, Mengkai Hou, Wanting Zhang, Shaobo Liu, Zhijun GaoRead
Peking University · National University of Singapore · BYD Company Limited · Shenzhen University Prompt injection against LLM agents becomes much stronger when the injected instruction is wrapped in the model’s own chat template. A forged template marker such as <|im_start|>
75
Benchmark
Argo-Bench stress-tests AI data agents on realistic enterprise-scale workflows where success depends not just on writing SQL, but on correctly investigating data and taking actions with real consequences.
Gabriel Tomitsuka, Arman Raayatsanati, Emma Xing, Duke Gand, Joseph J MaRead
TextQL
EnigmaForge tests whether AI can discover and solve a hidden puzzle in a story, revealing reasoning abilities that ordinary question-answering benchmarks may miss.
Daniel EisnerRead
Sep 2026
Video-Index stress-tests video benchmarks for shortcuts and provides a curated set of harder, more trustworthy evaluation items.
Enxin Song, Yinuo Xu, Shusheng Yang, Wenhao Chai, Jiatao GuRead
[
Load more videos
76
Continual Learning
ASCENT lets deployed LLM agents learn from verified successes on the fly by converting hindsight about their own trajectories into lasting weight updates.
Haodong Lu, Dong GongRead
University of New South Wales (UNSW Sydney)
SourceLearn helps LLM agents progressively build reusable expertise about trusted information sources instead of repeatedly starting from scratch.
Lucheng Fu, Kejing Xia, Yiyang Wang, Yiqiao Jin, Jinjin He, Xiyuan Yang, Haoxin Liu, Ye Yu, Haibo Jin, Yijia Xiao, Wenke Lee, B. Aditya Prakash, Haohan WangRead
Georgia Institute of Technology · Oct 2026 Website: https://sourcelearn.github.io/ · Introduction Large language model (LLM) agents increasingly rely on external sources to solve knowledge-intensive tasks. Here, a source denotes an identifiable body of external knowledge, such as a do · University of Illinois at Urbana-Champaign · University of California, Los Angeles
Local Support Learning helps large language models learn new skills without overwriting what they already know by activating updates only where they are locally needed.
Assaf Ben-Kish, Akarsh Kumar, James Glass, Raja GiryesRead
MIT CSAIL · Tel Aviv University
77
World Model
4Director makes video world models controllable by moving explicit 3D meshes through time while preserving realistic, consistent appearances.
Wei Cao, Hao Zhang, Vikram Voleti, Yuqun Wu, Mallikarjun B R, Shimon Vainer, Mark Boss, Yaoyao LiuRead
Oct 2026 Camera Control Input Image Figure 1: Camera and object control from a single image. Top: the image is lifted into a back- ground point cloud and complete rigid 3D geometry for each object, in
EMPIRIC lets robots discover missing physics through targeted experiments and use the resulting interpretable world models to plan better.
Yichao Liang, Amber Li, Dat Nguyen, Emily Bunnapradist, Michelangelo Naim, Sreela Kodali, Matteo Merler, Bowen Li, Kiran Gopinathan, Yiyun Liu, Nikhil Pimpalkhare, Joshua B. Tenenbaum, Adrian Weller, Zenna Tavares, Tom Silver, Kevin EllisRead
Basis Research Institute · University of Cambridge · Carnegie Mellon University · Fondazione Bruno Kessler · Massachusetts Institute of Technology · The Alan Turing Institute
RoboCoach uses imagined robot failures to decide what demonstrations to request next, making long-horizon manipulation skills improve more efficiently.
Jiajun Liu, Yifan Chen, Yichao Liu, Jiayi Zhang, Ruoqu Chen, Shaoxuan Xie, Guocai Yao, Mengdi Xu, Sen Cui, Changshui ZhangRead
[
78
Reasoning
FlyBy teaches small reasoning models to recognize when more internal thinking will not help and instead ask a stronger model for missing knowledge.
Chanuk Lee, Minki Kang, Sangwoo Park, Woongyeong Yeo, Jinheon Baek, Sung Ju HwangRead
KAIST · DeepAuto.ai(†\dagger: Equal advising)
RLVR can make reasoning models increasingly use strange internal languages, and preventing that drift may require sacrificing some performance.
Michael Sullivan, Alexander KollerRead
Department of Language Science and Technology, Saarland University, Saarbrücken, Germany
REST teaches latent LLM agents to form more causal, minimal, separable, and stable internal thoughts, improving reasoning accuracy and interpretability.
Fahd Seddik, Fatemeh FardRead
FARD Lab, University of British Columbia, Okanagan, Canada
Load more videos
79
Graph Learning
CodeGraph turns 167 million source files into a Wikidata-grounded knowledge graph of algorithms, paradigms, patterns, and software domains.
Federico Pennino, Andrea Gurioli, Stefano Zacchiroli, Maurizio Gabbrielli, Paolo FerraginaRead
Università di Bologna, Bologna, Italy · LTCI, Télécom Paris, Institut Polytechnique de Paris, Palaiseau, France · DISI, Università di Bologna, Bologna, Italy · Sant’Anna School of Advanced Studies, Pisa, Italy
GraphWrit3R turns 3D spatial data into open-vocabulary scene graphs using multimodal encoders and an LLM, without requiring ground-truth object annotations at inference.
Luka Milivojevic, Nikola Popovic, Sayan Deb Sarkar, Sebastian Koch, Iro Armeni, Luc Van Gool, Danda Pani PaudelRead
INSAIT, Sofia University “St. Kliment Ohridski” · Stanford University · Ulm University
This work shows how a small, carefully designed set of single-variable interventions can recover causal structure even when hidden confounders and selection bias complicate the data.
Xiaotian Hou, Kwangmoon Park, Hongzhe LiRead
Department of Biostatistics, Epidemiology and Informatics · University of Pennsylvania
80
Scientific Ai
Astronomers have detected radio auroras from β Pictoris b, revealing that this distant giant planet has a magnetic field at least 1.25 kilogauss strong.
Kevin N. Ortiz Ceballos, Edo Berger, Yvette CendesRead
Center for Astrophysics ∣\mid Harvard & Smithsonian, 60 Garden St, Cambridge, MA 02138, USA · Department of Physics and Astronomy, University of Oregon, Eugene, OR 97403, USA∗e-mail: kortizceballos@cfa.harvard.edu
An AI-guided robotic lab discovered palladium-based catalysts that could make acidic water electrolysis more durable while reducing dependence on scarce iridium and ruthenium.
Ken J. Jenewein, Faezeh Habib Zadeh, Xiaoxiao Wang, Gustavo Malkomes, Huafan Zhang, Natalie Page, Jae Jin Bang, Peter J. Santiago, Karla V. Contreras, Katherine K. Li, Allison Perna, Lorena M. Britton, Fahrettin Kilic, Kevin J. Cruse, Armin Taheri, Krishnanand Mallayya, Harley Quinn, Rebecca A. Durr, Peter A. Beaucage, John M. Gregoire, Rafael Gómez-BombarelliRead
Lila Sciences, Inc., Cambridge, MA 02141, USA
EurekaBench tests whether AI agents can move beyond accurate prediction to uncover mechanisms and insights that genuinely advance scientific understanding.
Jiayi Geng, Zhengxuan Wu, Kevin S. Chen, Seungone Kim, Joseph Janssen, Zora Zhiruo Wang, Bhupalee Kalita, Runtian Gao, Aaron Ho, Andrew Oakleigh Nelson, Olexandr Isayev, Francisco Villaescusa-Navarro, Ching-Yao Lai, Howard Chen, Graham NeubigRead
Carnegie Mellon University, Language Technologies Institute · Stanford University, Department of Computer Science · Yale University, Quantitative Biology Institute · Stanford University, Department of Geophysics · Carnegie Mellon University, Department of Chemistry · Massachusetts Institute of Technology, Plasma Science and Fusion Center · Columbia University, Department of Applied Physics and Applied Mathematics · Princeton University, Department of Astrophysical Sciences · Flatiron Institute, Center for Computational Astrophysics · Engram · Princeton University, Department of Computer Science
81
Code Generation
Better code documentation can faithfully reconstruct software, but surprisingly does not necessarily help AI coding agents fix real issues when the source code is already available.
Md Shohel Arman, Igor MolybogRead
Daffodil International University, Dhaka, Bangladesh · HawAII, University of Hawai‘i at Mānoa, Honolulu, HI, USA
GAGAR helps code agents learn not only to pass tests, but to produce cleaner and more targeted implementations by redistributing RL credit according to agentic quality judgments.
Jinhao Dong, Liang Zhao, Zihao Yue, Wenhan Ma, Linghao Zhang, Lei Li, Shicheng Li, Yifan Song, Bowen Ye, Fuli LuoRead
LLM Core, Xiaomi · Renmin University of China · Peking University · University of Hong Kong
IR4RL improves image-to-code generation by rewarding models for making useful visual progress at every intermediate rendering step.
Omri Kaduri, Kate Feingold, Phillip Isola, Tali DekelRead
Weizmann Institute of Science · MITProject page: https://ir4rl.github.io
Load more videos
82
Embodied Ai
GroundingPI argues that fast, precise visual grounding should be the perceptual foundation for capable robots and autonomous vehicles.
Qize Yu, Lianrui Fan, Boyu Chen, Jiaqi Liang, Xini Ding, Yue Chen, Zetian Song, Yuran Wang, Yi Zou, Kaixuan Wang, Tianxing Chen, Wenxuan Song, Bohan Zhou, Mingleyang Li, Siqiao Huang, Yuqi Ye, Caigao Jiang, Wei Wei, Ruihai Wu, Hang Zhang, Yixiao Ge, Shuchang Zhou, Shilong Liu, Xianming Liu, Ping Luo, Shiyu HuangRead
[
MM-ABC is a generalist robot foundation model that helps mobile manipulators see their surroundings, coordinate arm and base motion, and imagine future actions for better performance.
Qiwei Liang, Guangyu Chen, Shaolong Zhu, Zikuan Xiao, Jinxuan Lu, Yifan Xie, Renjing Xu, Wenbo Ding, Tianxing ChenRead
Xspark AI · The Hong Kong University of Science and Technology (Guangzhou) · Tsinghua University · The University of Hong Kong
A three-stage system converts human hand demonstrations into robust, zero-shot real-robot dexterous manipulation policies across different hand morphologies.
Tara Sadjadpour, Siming He, C. K. Wolfe, Haozhi Qi, Lea Wilken, S. Shankar Sastry, Claire Tomlin, Jitendra MalikRead
S. Shankar Sastry, Claire Tomlin*, and Jitendra Malik*
83
Robotics
JAMB helps two robot arms coordinate by jointly imagining their future movements and the changing 3D scene before acting.
Chuyang Xiao, Peilin Meng, David HeldRead
Robotics Institute, Carnegie Mellon University, Pittsburgh, PA 15213, USA · University of Michigan, Ann Arbor, MI 48109, USA
Rolling-WAM keeps future robot actions partially imagined and refined over time, making world-model-based manipulation replan 4.5 times faster.
Yinghua Zhou, Junjie Ye, Yiqi Zhao, Hao Dong, Celina Shiyu Wang, Ruohai Ge, Tingyi Yang, Basile Van Hoorick, Gaurav Sukhatme, Vitor Guizilini, Yue WangRead
Fudan University · Toyota Research Institute
A fast, training-free robot controller reuses and corrects demonstration trajectories to deliver traceable behavior at real-time speeds.
Maximilian Adang, Timothy Chen, Lars Osterberg, Aiden Swann, Mac SchwagerRead
84
Agent
LEGO-Anything turns images into editable Blender programs through iterative coding agents, offering a promising but still imperfect route to reconstructable 3D worlds.
Xirui Li, Peng Shi, Mingwen Dong, Sheng Zhang, Zhuoyan Xu, Dongkyu Lee, Shuaichen Chang, Yi Xiang, Lin Pan, Jiarong JiangRead
University of Maryland, College Park · AWS
MILO uses teams of evolving AI agents to automatically discover better harnesses for long-horizon problem-solving systems.
Prithwish Jana, Mononito Goswami, Hao Liu, Xinyu Li, Langlin Huang, Zhehui Huang, Zhishen Huang, Patrick Blöbaum, Anoop Deoras, Purak Jain, Nikos Kanakaris, Sahika GencRead
Georgia Institute of Technology · AWS AI Labs · Carnegie Mellon University · Washington University in St. Louis
This work trains AI agents to discover how to divide labor, challenge ideas, and combine reasoning so that teams can solve problems no individual agent could solve alone.
Aneesh Pappu, Mirac Suzgun, Yongchan Kwon, Federico Bianchi, Batu El, Mykel J. Kochenderfer, Hancheng Cao, James ZouRead
Stanford University · Together AI · Goizueta Business School, Emory University
Load more videos
85
Rl
Res-HIL lets humans efficiently improve robot manipulation skills by teaching a small corrective policy on top of an existing imitation policy.
Mariia Iavorskaia, Christian Dietz, Sebastian Albrecht, Majid KhadivRead
Siemens AG, Research and Predevelopment, Germany · Technical University of Munich, Germany
Choosing varied reasoning paths—not just correct ones—can make reinforcement-trained language models generalize better.
Dylan Zhang, Mingyuan Wu, Jinning LiRead
University of Illinois Urbana-Champaign, work done at Google · Google
VHD-Play turns solved mathematical mechanisms into cheap, stateful, self-verifying worlds where language-model agents can practice long-horizon decision-making.
Xinjie Shen, Wei Fan, Xudong Guo, Jianhong Tu, Yang Su, Chuqiao Kuang, Yinger Zhang, Dayiheng LiuRead
Georgia Institute of Technology · Alibaba Token Foundry, Alibaba Group
86
Speech
A new CTC training strategy makes large-vocabulary LLM speech recognition far more memory-efficient while retaining competitive accuracy and fast streaming inference.
Yifan Yang, Xiaoyu Yang, Zengrui Jin, Xian Shi, Yuxuan Wang, Yu Xi, Ziyang Ma, Qi Chen, Ruiyang Xu, Hui Wang, Dongchao Yang, Jin Xu, Xie ChenRead
Shanghai Jiao Tong University · University of Cambridge · Tsinghua University · Alibaba Token Hub, Alibaba Group · Nankai University · Chinese University of Hong Kong · SII https://github.com/yfyeung/PrunedCTC
The study argues that voice-similarity systems should be judged by whether their embedding geometry matches human perception, not merely by verification accuracy.
Szu-Chi Chen, Jia-Kai Dong, Yi-Cheng Lin, Sung-Feng Huang, Hung-yi LeeRead
National Taiwan University, Taipei, Taiwan · NVIDIA, Taiwan · Artificial Intelligence Center of Research Excellence (NTU AI-CoRE), NTU, Taiwan
Tacit-TTS makes zero-shot voice cloning over ten times faster while preserving the ability to clone voices from speech without transcripts, including multilingual and non-lexical references.
Jian Chen, You Zhang, Mark VintonRead
Dolby Laboratories
87
Multimodal
A compact embedding model unifies text, speech, audio, images, video, and documents in one search space without sacrificing the original text capabilities.
Mohammed Irfan Kurpath, Jaseel Muhammad Kaithakkodan, Sahal Shaji Mullappilly, Ivan Laptev, Hisham CholakkalRead
Ivan Laptev, Hisham Cholakkal · Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI)
Qwen3.8-Omni-Flash combines native text, audio, and video reasoning with long-context agentic planning and open-source tools for practical multimodal workflows.
Qwen TeamRead
YuE2 turns readable musical scores into high-quality full songs, combining controllable composition with end-to-end audio generation.
Ruibin Yuan, Jiahao Pan, Junyan Jiang, Zhiyue Wu, Ziya Zhou, Jiankai Sun, Yizhi Li, Ge Zhang, Yicheng Gu, Zeyue Tian, Junyu Dai, Hanfeng Lin, Kai Li, Shangda Wu, Xuanjie Liu, Jiaming Wang, Zihan Liu, Yue Wang, Yinghao Ma, Hanzhi Yin, Kangrui Chen, Xinyue Zhang, Ziyang Ma, Mengqi Liao, Hejia Zhao, Guowei Huang, Chao Yan, Lei Ke, Jianwei Yu, Bei Liu, Joe Guo, Liumeng Xue, Gus Xia, Wei Xue, Yike GuoRead
Load more videos
88
Cv
A lightweight script-aware mixture-of-experts model brings more accurate, scalable multilingual scene text recognition to many languages and scripts.
Xingsong Ye, Yongkun Du, Jiaxin Zhang, Zhixian Li, Chong Sun, Chen Li, Jing Lyu, Lianwen Jin, Zhineng ChenRead
[
DyRAD builds moving radar views of driving scenes by combining tracked object motion with the radar’s physics, enabling more realistic and transferable autonomy testing.
Merav Keidar, Tomer Borreda, Rajalakshmi Nandakumar, Or LitanyRead
Oct 2026 Azimuth° Doppler Azimuth° Doppler Lateral Sensor Shift High Res Sensor Range [m] Range [m] Azimuth° Doppler Azimuth° Doppler Figure 1: Radar re-simulation with DyRAD. From recorded radar · Technion Cornell Tech NVIDIA Ground Truth Repositioned Object Range [m] Range [m]
This work maps when training models to generate images can make them better at understanding images, revealing both intuitive and surprising task-to-task benefits.
Jiaxin Ge, Yiming Qin, Ji Xie, Haozhe Jiang, Xiaochuang Han, Junyi Zhang, Andrew Dai, Yinfei Yang, Jitendra Malik, Ranjay Krishna, Sewon Min, Haiwen Feng, Le Xue, Baifeng Shi, Trevor Darrell, XuDong WangRead
University of California, Berkeley · Duke University · Impossible Research · Carnegie Mellon University · University of Washington · Elorian
89
Nlp
AdaTutoRank teaches rerankers to assemble complementary evidence sets rather than merely picking individually relevant documents, improving RAG and deep-research retrieval with fewer calls.
Kailin Jiang, Lei Liu, Jian Xi, Yangqi Chen, Hui Xu, Hongwei Zhao, Bin Li, Yu Lu, Haibo ShiRead
University of Science and Technology of China · Yuanbao Team, Tencent
The paper argues that sentence meaning equivalence is not reliably stored in separate embeddings but is computed when models process both sentences together.
Jiaqi DengRead
Independent Researcher
SlopShape detects and identifies AI-written commercial content by recognizing its underlying organizational style, even after the text has been reworded.
Jochen MadlerRead
Sitefire
90
Efficiency
Disaggregated quantization gives LLM prefill and decode their own specialized weights and formats, improving low-bit accuracy while speeding up first-token generation.
Andrei Panferov, Maximilian Kleinegger, Sweta Priyadarshi, Tijmen Blankevoort, Dan AlistarhRead
NVIDIA · ISTA
A cheap judge handles confident cases while uncertain ones are escalated, preserving nearly all of a powerful LLM judge's accuracy for a fraction of the cost.
Yubo Li, Yidi Miao, Ramayya Krishnan, Rema PadmanRead
Carnegie Mellon University · {yubol, yidim, rk2x, rpadman}@andrew.cmu.edu
PoEM aims to mix the behaviors of already RL-trained models to predict how they would perform under a new reward, without running expensive RL again.
Kimia Hamidieh, Giannis Daras, Antonio TorralbaRead
MIT CSAIL
Load more videos
91
Diffusion
LongLive-Plug distills key video-generation capabilities into reusable LoRA adapters that can accelerate and improve many downstream diffusion models without retraining each one.
Shuai Yang, Luozhou Wang, Wei Huang, ZhiFei Chen, Bohan Zhang, Xiao Fu, Qianli Ma, Chen-Hsuan Lin, Weian Mao, Bryan Chu, Song Han, Yukang ChenRead
NVIDIA
FoMo uses the moment when two images diverge during diffusion generation as an automated measure of how perceptually different they are.
Jaihyun Lew, Mingi Jung, Minjun Park, Wooseok Song, Sungroh YoonRead
Interdisciplinary Program in AI, Seoul National University · Department of Electrical and Computer Engineering, Seoul National University · AIIS, ASRI, INMC, and ISRC, Seoul National University
Simplex Diffusion Models keep uncertainty alive during discrete denoising, enabling faster and stronger generation for text, code, and math tasks.
Justin Deschenaux, Alexandre Galashov, Andrew Campbell, Li Kevin Wenliang, James Thornton, Arnaud Doucet, Valentin De BortoliRead
Google DeepMind · EPFL · UCL Gatsby
92
Generative Model
RULER uses instruction-specific visual rubrics as reinforcement-learning rewards to make SVG generation more faithful, stylish, and resistant to reward hacking.
Hangyu Ran, Yuhao Zheng, Yingying Zhang, Kevin Qinghong Lin, Han PengRead
Ant Group · The Hong Kong University of Science and Technology (Guangzhou) · Independent Researcher · University of Oxford
A visual reward model that first decides what matters in each image-generation case, then scores outputs with detailed rubrics to provide better training signals.
Xuehai Bai, Zhenchen Tang, Yang Shi, Dianyi Wang, Tengfei Liu, Wanshun Su, Xuanyu Zhu, Ruohui Wang, Haiwen Diao, Haotian Wang, Xiaoling Gu, Yuanxing ZhangRead
[
WanPE turns ordinary text prompts into director-level cinematic plans, substantially improving the quality and consistency of long-form AI-generated videos.
Yubo Zhu, Yawen Shao, Ziyun Dai, Zixun Fang, Kai Zhu, Siyang Sun, Haolan Xue, Chuxin Wang, Tingyu Weng, Jingming Luo, Chen Shi, Lianghua Huang, Yufeng Ai, Yuzheng Wang, Wenyuan Zhang, Yu Shang, Yuxiang Bao, Zoubin Bi, Jie Xiao, Jinbo Xing, Jiaxing Zhao, Chongyang Zhong, Hengjian Chen, Chenwei Xie, Akide Liu, Zhehan Kan, Yu Liu, Wei Zhai, Sheng Zhong, Wei TongRead
[
93
Self Supervised
A learner and an RL-powered program generator teach each other from scratch, producing synthetic data that enables surprisingly meaningful transfer to natural datasets.
Aditya Cowsik, Kfir Dolev, Michael Y. Li, G. Bruno De Luca, Nourya Cohen, Noah D. Goodman, Yoav LevineRead
Independent Researcher · Tel Aviv University · Stanford University · LAPTh, USMB
Instead of training LLMs on many similar correct answers, this work shows that exposing them to diverse problem-solving strategies—even imperfect ones—can produce stronger models.
Alexander Gurung, Esmeralda S. Whitammer, Mirella LapataRead
School of Informatics, University of Edinburgh · {alex.gurung,esmeralda.whitammer}@ed.ac.uk, mlap@inf.ed.ac.uk
TT-VidT pretrains video models to focus on motion while preserving appearance, achieving strong action-recognition results with substantially lower compute.
Shih-Ying Yeh, Daniel Z. Kaplan, Xuehai Wang, Fu-En Yang, Min-Hung Chen, Shang-Hong LaiRead
National Tsing Hua University Comfy Org Research realiz.ai · Karolinska Institutet Stockholm University NVIDIA†Corresponding author: kohaku@kblueleaf.netProject page: https://kohakublueleaf.github.io/TTVidT/
Load more videos
94
Optimizer
01
An $Ω(κ_y^8ε^{-6})$ Lower Bound for Stochastic NC-SC Bilevel Optimization with First-order Oracles
This work proves that stochastic bilevel optimization fundamentally requires up to epsilon^{-6} oracle queries, showing existing methods are asymptotically optimal.
Zhihao Gu, Qilong Wu, Junchi YangRead
The Pennsylvania State University · The Chinese University of Hong Kong, Shenzhen
A pair of self-improving coding agents evolves new learnable optimization algorithms from a simple template, reducing the need for handcrafted optimizer design.
Zipei Yu, Yue-Jiao Gong, Zeyuan Ma, Yuncheng Jiang, Zhiguang CaoRead
South China University of Technology · South China Normal University · Singapore Management Univeristy
A new multiscale matrix-weights algorithm learns hidden linear preferences online with provably optimal dimension-dependent regret.
Shinsaku SakaueRead
Sep 2026
95
Neural Network
A new query-efficient technique makes it possible to steal the parameters of small black-box neural networks using only their predicted labels.
Akira Ito, Takayuki Miura, Yosuke TodoRead
Stembed lets music producers search for individual instrument sounds hidden inside a full song by representing the mixture as multiple searchable stem-like embeddings.
David Braun, Junyi Fan, Pranay Manocha, Donald S. Williamson, Adam FinkelsteinRead
This paper argues that claims about linear representations only become meaningful once we specify which transformations leave a representation essentially unchanged.
Louie Hong Yao, Yuhao Li, Shengchao LiuRead
96
Attention
PISA makes long-context attention more scalable by hierarchically narrowing relevant key blocks instead of comparing every query with every block.
Bohao Tang, Zhen Qin, Yuqi Pan, Zheng Li, Pengfei LiuRead
Memory Attention replaces some costly attention computation with reusable token memories, potentially making language models cheaper to run and easier to fit on limited GPU hardware.
Jiale KangRead
The study finds that real paragraph structure is revealed not merely by where attention concentrates, but by how deeply that compression develops across model layers.
Shuyang XiangRead
Load more videos
97
Transformer
This work shows, in theory, how transformers can adapt to data living on locally different geometric structures and still achieve statistically optimal in-context prediction.
Jaehee Seo, Jisu KimRead
Department of Statistics · Seoul National University, Korea
This paper shows that letting a Transformer spend extra computation on latent thought steps can approach the quality of doubling its depth while using substantially fewer parameters.
Zeyi Huang, Xuehai He, Yong Jae Lee, Yelong ShenRead
Microsoft · University of Wisconsin–Madison
This work argues that Transformers can blend two thoughts in one computation and proposes a way to separate them into coherent parallel continuations.
Pavel Tikhonov, Anton Korznikov, Matvey Mikhalchuk, Nikita Dragunov, Temurbek Rahmatullaev, Polina Druzhinina, Anton Razzhigaev, Ivan Oseledets, Elena TutubalinaRead
98
Llm
ComPO aligns language models using preference comparisons rather than direct likelihood optimization, aiming to reduce failures caused by small preference margins.
Peter Chen, Xi Chen, Wotao Yin, Tianyi LinRead
Department of Electrical Engineering and Computer Sciences (EECS) · University of California, Berkeley · Stern School of Business · New York University · Decision Intelligence Lab (Seattle) · DAMO Academy, Alibaba Group U.S · Department of Industrial Engineering and Operations Research (IEOR) · Columbia University
Even supposedly deterministic greedy LLM decoding can produce different answers in BF16 and FP16, but selectively recomputing the final layer in FP32 can greatly improve agreement at low cost.
Gaoyuan Du, Anam Nawaz Khan, Rex Zhou, Xiaoyang Liu, Deepayan Chakrabarti, Fnu Suya, Xueping LiRead
University of Tennessee, Knoxville · University of Chicago · Amazon · University of Texas at Austin
This study asks whether new knowledge belongs in an LLM’s context, hidden representations, or weights, finding that KV-cache-based Cartridges often deliver the best accuracy but can cause forgetting.
Nathanaël Carraz Rakotonirina, Momchil Hardalov, Gonzalo Iglesias, Adrià de GispertRead
Gonzalo Iglesias · Amazon AGI · {ncarraz, momchilh, gjii, agispert}@amazon.com
99
Foundation Model
t0 is an open-weight time-series foundation model that uses historical and future context to deliver strong zero-shot probabilistic forecasts across diverse datasets.
Lucas Meyer, Claudio Sole, Huikan Xiang, Nicolas Li, Lucas Franceschino, Arnau Quera-Bofarull, Maarten P. Scholl, Joachim Fainberg, Geoffrey NégiarRead
TabPFN-3.5 advances tabular foundation models with stronger multimodal and real-world data handling, faster inference, and scalable reasoning modes.
Benjamin Jäger, Nick Erickson, Léo Grinsztajn, Felix Birkel, Klemens Flöge, Oscar Key, Kürşat Kaya, Jonas Kübler, Adèle Frankel, Tobias Schröder, Anurag Garg, Jan Hendrik Metzen, David Salinas, Simon Bing, Kristina Collins, Tuana Çelik, Vahid Balazadeh, Lydia Sidhoum, Tomás Pereda, Brendan Roof, Andrej Tschalzev, Siyuan Guo, Philipp Singer, Lennart Purucker, Jake Robertson, Marie Salmon, Philipp Jund, Jerry Chen, Diana Kriuchkova, Arthur Cahu, Eliott Kalfon, Adrian Hayler, Georg Grab, Vitor Monteiro, Lilly Wehrhahn, Dominik Safaric, Clara Cornu, Alan Arazi, Rylee Grace, Simone Alessi, Mihir Manium, Bernhard Schölkopf, Yann LeCun, Madelon Hulsebos, Sauraj Gambhir, Noah Hollmann, Frank HutterRead
nnFoundation trains complementary 3D convolutional and transformer models on 2.1 million medical scans and shows that matching architecture to the task can improve transfer across diverse radiology applications.
Constantin Ulrich Harsy, Tassilo Wald, Karol Gotkowski, Yannick Kirchhoff, Marcel Knopp, Maximilian Rokuss, Elisa Stegmeier, Philipp Schader, Dasha Trofimova, Raphael Stock, Kim-Celine Kahl, Stephen Schaumann, Selen Erkan, David Zimmerer, Stefan Denner, Moritz Langenberg, Sebastian Ziegler, Katharina Eckstein, Maximilian Fischer, Jonathan Suprijadi, Bálint Kovács, Benjamin Hamm, Anand Deshpande, Dimitrios Bounias, Nico Disch, Shuhan Xiao, Jessica Kächele, Jan Sellner, Rajesh Baidya, Jeremias Traub, Lars Krämer, Maximilian Zenk, Tim Rädsch, Stefan Dvoretskii, Robin Peretzke, Jonathan Deissler, Alexandra Ertl, Partha Ghosh, Kris Dreher, Stefan Dinkelacker, Annika Reinke, Evangelia Christodoulou, Numan Saeed, Yoland Savriama, Santiago Estrada, David Kügler, Laura Alexandra Daza Barragan, Cristina Isabel Gonzalez Osorio, Jan Peeken, Michael Baumgartner, Marvin Teichmann, Guillaume Chabin, Matthias Kirchler, Valentin Koch, for the ALFA study, Markus Hohenhaus, Dimitri Koslov, Nina ...Read
Division of Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany · Helmholtz Imaging, German Cancer Research Center (DKFZ), Heidelberg, Germany · HIDSS4Health - Helmholtz Information and Data Science School for Health, Helmholtz Association, Karlsruhe/Heidelberg, Germany · Faculty of Mathematics and Computer Science, Heidelberg University, Heidelberg, Germany · Division of Intelligent Medical Systems, German Cancer Research Center (DKFZ), Heidelberg, Germany · Formerly: Division of Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany · National Center for Tumor Diseases (NCT) Heidelberg, a partnership between the German Cancer Research Center (DKFZ) and Heidelberg University Hospital (UKHD), Heidelberg, Germany · Formerly: Division of Intelligent Medical Systems, German Cancer Research Center (DKFZ), Heidelberg, Germany · German Cancer Consortium (DKTK), DKFZ, core center Heidelberg, Heidelberg, Germany · Medical Faculty Heidelberg, Heidelberg University, Heidelberg, Germany · Helmholtz Metadata Collaboration (HMC) Hub Health, German Cancer Research Center (DKFZ), Heidelberg, Germany · Floy GmbH, Munich, Germany · Department of Computer Vision, Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), Abu Dhabi, United Arab Emirates · Animal Phenotyping Platform, Max-Delbrück-Centrum für Molekulare Medizin in der Helmholtz-Gemeinschaft, Berlin, Germany · AI in Medical Imaging, German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany · Institute of Machine Learning for Biomedical Imaging, Helmholtz Munich, Neuherberg, Germany · School of Computation, Information and Technology, Technical University of Munich, Munich, Germany · Department of Radiation Oncology, Technical University of Munich (TUM), School of Medicine and Health, Klinikum rechts der Isar, Munich, Germany · Digital Technology and Innovation, Siemens Healthineers, Erlangen, Germany · IT Core Facility, German Cancer Research Center (DKFZ), Heidelberg, Germany
Load more videos
100
Hardware
This work shows that quantum LDPC codes can still reliably correct coherent errors below a universal noise threshold, strengthening the foundations of scalable fault-tolerant quantum computers.
Zhengyi Han, Yuanchen Zhao, Yijia Xu, Yixu Wang, Zi-Wen LiuRead
Yau Mathematical Sciences Center, Tsinghua University, Beijing 100084, China · State Key Laboratory of Low Dimensional Quantum Physics, Department of Physics, Tsinghua University, Beijing, 100084, China · Joint Center for Quantum Information and Computer Science, University of Maryland, College Park, Maryland 20742, USA · Shanghai Institute for Mathematics and Interdisciplinary Sciences (SIMIS), Shanghai 200433, China
101
Benchmark
GameHorizon is a large benchmark and dataset designed to test whether AI systems can understand instructions and execute gameplay tasks over short and long time horizons.
Yiran Wang, Xingyilang Yin, Junfu Pu, Guangzhi Wang, Kaifeng Li, Mingyu Ouyang, Huiqiang Sun, Lingen Li, Cheng Cheng, Wangbo Yu, Honghao Chen, Xiaodong Cun, Chi-Man Pun, Zhiguo Cao, Ying ShanRead
ARC Lab, Tencent · GVC Lab, Great Bay University · University of Macau · National University of Singapore · Huazhong University of Science and Technology · MMLab, CUHK
ProgramDistill tests whether coding agents can rebuild web-app features by exploring working reference software rather than merely following written specifications.
Jeonghye Kim, Minseon Kim, Young Jin Kim, Matheus Pereira, Marc-Alexandre Côté, Alessandro Sordoni, Xingdi Yuan, Zhengyan ShiRead
KAIST · Introduction Figure 1: ProgramDistill evaluates reference-guided software engineering for interactive web applications. A coding agent is given a working reference application whose source implementat · Microsoft Research Montréal · Microsoft AI https://aka.ms/froggy Coding agents are typically evaluated with desired behavior specified through issues or instructions. In practical
VA-Bench tests whether vision-language models can actively gather the right views and turn spatial understanding into reliable robot actions.
Zhongbo Zhang, Jiayi Jin, Yifan Wang, Zaibin Zhang, Haiwen Diao, Lijun Wang, Huchuan LuRead
Dalian University of Technology · Nanyang Technological University
102
Continual Learning
ACLArena studies how agents can learn new skills over multiple training stages without forgetting old ones, proposing replay and specialized LoRA experts as a practical solution.
Haixin Wang, Xiaoxuan Wang, Junkai Zhang, Han Zhang, Renliang Sun, Alexander K Taylor, Yidan Shi, Haoran Deng, Chenguang Wang, Jason Cong, Yizhou Sun, Wei WangRead
Stage 2 ... Stage K Teacher · Teacher 2 ... Teacher K 🔍 (π0, D1) 🔥 (π1, D2) 🔥 ... (πK-1, DK) 🔥 ... Generate Model π1 Model π2 ... Final Model Dist · University of California, Santa Cruz
A frozen AI designer gets better over time by turning user interactions and successes into reusable, continually refined design procedures.
Hongyang Du, Lan Yan, Christian Flores, Asim KadavRead
Adobe · Brown University Corresponding to hongyang
Knowledge Pull Requests make continually updating documents more transparent by showing exactly which claims changed, where they belong, and how the final text was revised.
Alexander Martin, Benjamin Van DurmeRead
Johns Hopkins University · {amart233, vandurme}@jhu.edu
Load more videos
103
World Model
HappyWorld-Bench tests whether AI-generated worlds remain coherent, editable, and responsive when agents explore and act within them.
Zhiqi Bai, Junai Cai, Yixin Chen, Jingrun Du, Tao Feng, Wei Gong, Siyuan Huang, Xiao Lin, Jiaheng Liu, Jun Luo, Yongzhe Lyu, Liya Ma, Zenan Meng, Lin Qu, Wenbo Su, Jiaming Wang, Qinghe Wang, Shaofei Wang, Yanghai Wang, Zequn Wang, Ziming Wang, Hu Wei, Jiangtao Wu, Ruiqi Wu, Jiaxin Xie, Yuchi Xu, Ze Xu, Chengting Yu, Liangyu Yuan, Gang Zeng, Yawen Zeng, Xingyao Zhang, Zizheng Zhang, Bo Zheng, Jiancheng Zhu, Song-Chun ZhuRead
JEPA-Anything aims to provide one factorized recipe for learning predictive world models spanning vision, biology, medicine, physics, control, and weather.
Taoyong Cui, Zhongyao Wang, Xinyue Xu, Weiyang Liu, Zhaochen Yu, Yuying Zhang, Qiang Gao, Mengyue Yang, Wanli Ouyang, Pheng Ann Heng, Yingcheng Wu, Zhenfei Yin, Ling YangRead
Corresponding authors. Organizations and contact details are listed in Appendix C
PointZero learns general 3D motion dynamics from sparse point tracks rather than robot actions, then transfers that knowledge to manipulation and imitation tasks.
Bardienus P. Duisterhof, Kaifeng Zhang, Adam Hung, Bowen Wen, Stan Birchfield, Yunzhu Li, Deva Ramanan, Jeffrey IchnowskiRead
CMU · Columbia · NVIDIA
104
Reasoning
NFV uses AI agents to translate ordinary code into machine-checkable formal proofs, making software verification more accessible while revealing the limits of end-to-end soundness.
Shuvendu K. LahiriRead
Microsoft Research
Reasoning models can solve individual rule tasks yet often fail when the same underlying logic is rearranged, suggesting their thinking is less systematic than it appears.
Simon Schug, Brenden M. LakeRead
Princeton University
Instead of trusting language models to do clinical math, the system has them write code for a restricted executor—and shows that this helps larger models but is no universal cure for medical calculation errors.
Felipe Ocampo Osorio, Sebastián Andrés Cajas Ordoñez, Maximin Lange, Rafi Al Attrach, Sahil Kapadia, Zakaria Laouabdia Sellami, Angelo Antonio Talio, Leo Anthony CeliRead
MIT Critical Data · UNC Chapel-Hil · University of Pavia · Humanitas University
105
Graph Learning
This paper makes massive-scale item recommendations much faster by approximating graph-based similarity scores without sacrificing retrieval quality.
Runhao Jiang, Renchi YangRead
Hong Kong Baptist University, Hong Kong SAR, China
This work uses particle-based graph dynamics to clean noisy labels before training GCNs, improving robustness while keeping computation low.
Fabricio BreveRead
organization=São Paulo State University - UNESP, city=Rio Claro - SP, country=Brazil
RelateAnything aims to make scene-graph relation prediction as flexible as open-vocabulary detection by predicting arbitrary text-specified relations in real time.
Maëlic NeauRead
Sep 2026 still trained and evaluated on the 50 or 56 predicates of a single annotation style, with the relation head conditioned on object class labels and therefore tied to one detector and one label
Load more videos
106
Scientific Ai
HypoEvolve uses teams of LLM agents and genetic evolution to generate and refine drug-repurposing hypotheses that better align with biological evidence.
Jieyuan Liu, Mengzhou Hu, Jefferson Chen, JungHo Kong, Pratibha Jagannatha, Yiming Gao, Dexter Pratt, Hsin-Yuan Lee, Zhiting Hu, Trey Ideker, Wei Wang, Eric P. Xing, Zhen WangRead
University of California San Diego Texas A&M University · Carnegie Mellon University Mohamed bin Zayed University of Artificial Intelligence
OpenAI4S turns AI-driven scientific research into inspectable, resumable sessions by treating executable code as the agent’s actions and tracking every computational step.
Gongbo Zhang, Hao Li, Yu Wang, Mujie Lin, Liuzhenghao Lv, Yicheng Mao, Yimi Wang, Jun Zhu, Minhan Tang, Zhengxiang Jiang, Yusong Wang, Jiayu Yao, Kunpeng Ning, Dawei Pang, Yonghong Tian, OpenAI4S Community, Yuyang Liu, Li YuanRead
Peking University Shenzhen Graduate School, Shenzhen, 518055, China · These authors contributed equally to this work · Tsinghua-Peking Joint Center for Life Sciences, School of Life Sciences, Tsinghua University, Beijing, China · Independent Researcher · Beijing Yuankong Intelligent Technology Co., Ltd., Beijing, China
ScienceIDE turns scientific codebases into interactive training grounds where AI agents can learn to solve, verify, and improve scientific programming tasks.
Hejia Geng, Zesen Huang, Haoyang Li, Wenbin Li, Koutian Wu, Zihan Zhou, Yuanbo Pang, Weihao Liu, Zigong Xu, Zhiping Li, Zongzheng Zhang, Chuanfei Dong, Jiankai Sun, Tianzhe Zheng, Fengyu Xie, Yue Ma, Yueheng Shi, Tong Xie, Zonglin Di, Xianrong Liu, Qucheng Gao, Yimin Liu, Jiaming Pan, Sheng Huang, Xiao-Han Ma, Lanqing Yuan, Zhenlin Zhu, Ziang Liu, Ziyang Xu, Junkai Wang, Kangkai Liang, Jiayi Xian, Zehong Zhao, Liuwei Xu, Jingxu Xie, Peijin Zhang, Qiang Gao, Chengyi Xing, Zhe Zhao, Xi Wang, Yaopeng Xing, Xing Meng, Zhenfei Yin, Yingcheng Wu, Ling YangRead
Sep 2026 Yuanbo Pang6 Weihao Liu7 Zigong Xu8 Zhiping Li5 Zongzheng Zhang1 Chuanfei Dong9 Jiankai Sun10 Tianzhe Zheng8 Fengyu Xie1 Yue Ma11 Yueheng Shi10 Junkai Wang1 Tong Xie12 Zonglin Di6,13 Xianrong
107
Code Generation
CodeMidas turns existing codebases into scalable, automatically verified RL environments that train coding agents to perform better across diverse software tasks.
Bowen Ye, Lei Li, Shicheng Li, Zihao Yue, Linghao Zhang, Hanglong Lv, Yuanxin Liu, Wenhan Ma, Hao Tian, Rang Li, Jinhao Dong, Yikai Zhao, Xiangwei Deng, Hailin Zhang, Liang Zhao, Qi Liu, Lingpeng Kong, Tong Yang, Fuli LuoRead
LLM Core, Xiaomi · Peking University · University of Hong Kong · Renmin University of China
A survey of 527 researchers finds that AI-generated scientific code is usually checked informally by individuals rather than through systematic tests or peer review.
Gabrielle O'Brien, Reed Milewicz, Nasir EistyRead
Sandia National Laboratories, Albuquerque, New Mexico, USA · University of Tennessee, Knoxville, Knoxville, Tennessee, USA
AI coding agents can pass tests yet fail in the real world, so dependable development requires continuously narrowing the gaps between human intent, evaluation models, and deployment reality.
Alexander Krentsel, Shubham Agarwal, Mert Cemri, Shu Liu, Sidharth Sankhe, Ziming Mao, Matei Zaharia, Ion StoicaRead
UC Berkeley
108
Embodied Ai
A general-purpose agent becomes a robot policy by writing and adapting its own programs while interacting with the physical world.
Mengzhao Jia, Yang Lin, Xixin Zhang, Zhihan Zhang, Xiaobai Liu, Meng JiangRead
mjia2@nd.edu · University of Notre Dame University of California San Diego San Diego State University*Core contributors
HuRo turns massive human-video collections into robot-training data, substantially improving the robustness and performance of vision-language-action policies.
Jinho Jeong, Se June Joo, Jaehyun Kang, Dongyun Kim, Yena Kim, Hanjung Kim, Seon Joo KimRead
RLWRLD · Yonsei University
ORCH shows that organizing embodied AI agents into task-specific hierarchical teams can substantially improve coordination and performance in complex wildfire-response missions.
Zhengran Ji, Jonathan Hyun, Boyuan ChenRead
Department of Computer Science, Duke University, Durham, USA · Department of Electrical and Computer Engineering, Duke University, Durham, USA
Load more videos
109
Robotics
LIMBO teaches humanoid robots to discover and internalize safety boundaries so they can perform agile maneuvers without relying on an online safety filter.
Jake Gonzales, Arturo Flores Alvarez, Yu-Ming Chen, Aaron D. Ames, Lillian J. Ratliff, Manikantan NambiRead
SwingBot teaches humanoid robots to swing continuously across overhead bars by combining structured motion guidance with learned sensing and control.
Yujie Xiong, Peng Zhai, Taixian Hou, Quancheng Qian, Cunwang Liu, Kangmai Hu, Long Yang, Zhiyan Dong, Lihua ZhangRead
College of Intelligent Robotics and Advanced Manufacturing, Fudan University · Fysics AI Technologies (Shanghai)* Corresponding authors
TacBPM gives dexterous robots a tactile-aware library of reusable motion skills so they can reorient unfamiliar objects more reliably and with less trial-and-error.
Jie Yin, Wanli Xing, Zeyuan Zhao, Xuezhou Zhu, Zhijie Deng, Kaifeng ZhangRead
Sharpa Robotics
110
Agent
Atria Dawn explores how research-oriented AI agents can move beyond completing tasks to partnering with humans on scientific discovery while preserving human oversight.
Honglin Guo, Tao Gui, Yicheng Chen, Guanting Dong, Qiming Ge, Yuyang Hu, Zixian Huang, Jiajie Jin, Alexander Lam, Yining Li, Jiahang Lin, Yanjiang Liu, Xinyu Lu, Haijun Lv, Junlin Shang, Qisheng Su, Guoqiang Wang, Rui Wang, Zhecan Wang, Hao Xiang, Xinchen Xie, Shuhao Xing, Xiaoyu Xing, Wanghan Xu, Xinyu Yang, Yajie Yang, Chengfeng Zhao, Haoran Zhao, Ruojun Zhou, Yunhua Zhou, Yicheng Zou, Kun Cai, Qiye Cai, Xinmeng Che, Haodong Chen, Jiabei Chen, Jiahao Chen, Jiayi Chen, Yujia Chen, Lizhi Cui, Youheng Dai, Xin Deng, Yi Dong, Shihan Dou, Chenya Gu, Xu Guo, Ding Han, Feiyang Hao, Haotan He, Jie Hou, Binze Hu, Zijian Hu, Junhao Huang, Huicheng Jiang, Jiazhen Jiang, Shufan Jiang, Jiahao Kuang, Bowen Lai, Bo Li, Jiaqiang Li, Peng Li, Qilong Li, Zhuoqun Li, Jiaxiang Liu, Shuainan Liu, Tong Liu, Yi Liu, Zhonghang Lu, Jianwen Luo, Yanyi Luo, Huijie Lv, Ningsheng Ma, Zerun Ma, Houcheng Min, Chengjun Pan, Qiyuan Peng, Xiaoxuan Peng, Jianmin Qian, Jiantao Qiu, Wanying Ren, Huayu Sha, Jifei Shan...Read
OmniHarness turns successful visual-agent behavior into reusable symbolic skills that can be practiced, adapted, and plugged into new generation tasks.
Xu Xu, Jinxiu Liu, Zhangbo Qiao, Jiaxing Lu, Xiangyu Zhang, Yubin Gu, Fangwei Ning, Yan ShiRead
Beihang University · Overview of OmniHarness and its diverse visual generation capabilities. 1 · The Chinese University of Hong Kong · National University of Singapore
This paper shows that LLM agents often get worse returns from prolonged thinking and can perform better by splitting compute across multiple shorter attempts.
Kaiyuan Liu, Qiuyang Mang, Bo Peng, Wenhao Chai, Hanchen Li, Shreyas Pimpalgaonkar, Luke Zettlemoyer, Alex Dimakis, Alvin CheungRead
UC Berkeley · University of Washington · Princeton University · Bespoke Labs
111
Rl
NGU helps RL-trained LLMs stop over-practicing easy problems and spend more effort solving the hard ones.
Michael Noukhovitch, Hamish Ivison, Nathan Lambert, Aaron CourvilleRead
Mila, Université de Montréal · Allen Institute for AI · University of Washington · Trillium Labs · Canada CIFAR AI Chair
MInTRL helps reinforcement-learning systems discover better solutions by making small, targeted corrections during otherwise on-policy rollouts.
Mingyu Chen, Yefan Tao, Gerald Friedland, Xuezhou Zhang, Chris KongRead
Amazon Web Services · Boston University
A simple score-centering correction makes off-policy RL for language models more stable when training and inference engines do not perfectly agree.
Martin Marek, Max RyabininRead
Together AI
Load more videos
112
Speech
A large voice-cloning model becomes a synthetic data generator for training a compact, reference-free Thai TTS system that runs on-device.
Kunat Pipatanakul, Potsawee Manakul, Warit Sirichotedumrong, Sittipong Sripaisarnmongkol, Pakorn Nathong, Phatrasek JirabovonvisutRead
Wayu Research · Paxa Labs · Typhoon · Sep 2026 Technical Report
StepAudio 3 is a real-time voice model that listens, reasons, speaks, and uses tools concurrently for more natural spoken interaction.
Bin Lin, Bo Zhao, Boyang Zhang, Boyong Wu, Chao Yan, Chen Geng, Chen Wu, Cheng Yi, Chengli Feng, Chenglin Zhu, Chengting Feng, Chengyuan Yao, Daijiao Liu, DanNi Wan, Daxin Jiang, Dongjian Li, Dongqing Pang, Fei Tian, Feng Tian, Future Li, Gang Yu, Guanglong Yang, Haoyang Zhang, Hongyuan Wang, Jia Peng, Jiahao Song, Jialong Xue, Jiamin Fan, Jiangjie Zhen, Jianzheng Gao, Jincheng Wen, Jinghua Liang, Jinglan Gong, Jun Chen, Li Xie, Liang Zhao, Lifang Zhang, Lingli Ji, Lun Cai, Min Xu, Peilin Li, Peng Yang, Pengfei Tan, Qingjian Lin, Qinxin Du, Ruijie Xiong, Runze Li, Shenghua Hu, Shengqian Qin, Shi Qiu, Siqi Tu, Siyi Zhou, Tianjiao Deng, Wanying Lu, Weiming Niu, Wen Sun, WenWen Qu, Xiangyu Zhang, Xianwei Zhang, Xiaosu Su, Xing Chen, Xinyu Liu, Xuerui Yang, Yan Wu, Yang Li, Yang Yang, Yechang Huang, Yibo Zhu, Yifan Zhang, Yinuo Yan, Youjun Chen, Yu Fu, Yu Luo, Yu Zhou, Yujie Chen, Yumang Wang, Yunzhou Ju, Yuxiang Yang, Yuxin Li, Yuxin Zhang, Zekai Liu, Zengwei Yao, Zhaoxin Yuan, Zhenwei...Read
The work decodes what someone means from brain signals by translating MEG activity into semantic representations before generating text.
Gilad D. Landau, Dulhan Jayalath, Oiwi Parker JonesRead
PNPL University of Oxford
113
Multimodal
ModaLens tests whether medical VLMs truly look at the X-ray when they already have the report, revealing that reports can substantially suppress measurable image sensitivity.
Sebastián Andrés Cajas Ordóñez, Maximin Lange, Quang Bui, Anqi Peter Li, Felipe Ocampo Osorio, Rafi Al Attrach, Kushul Reddy Palakala, Sahil Kapadia, Zakaria Laouabdia Sellami, Xinyue Zhang, Ashley Zhang, Leo Anthony CeliRead
PANORAMA teaches vision-language models to describe an entire image while precisely linking every phrase to the pixels it refers to.
Sara Pieri, Evangelos Kazakos, Shizhe Chen, Josef Sivic, Cordelia SchmidRead
Vision-RL2 helps multimodal language models focus on the right image regions, improving fine-grained understanding while using roughly four times fewer visual tokens.
Yuheng Shi, Xiaohuan Pei, Minjing Dong, Chang XuRead
114
Cv
The study shows that video models may retain the right physical understanding even after generating the wrong motion—and that targeted internal writes can bring that knowledge back.
Xingyun Wang, Haomin Zheng, Man Yuan, Leqian Yang, Ziming LiuRead
Bi-FlowGS lets generated videos and 3D Gaussian scenes iteratively correct each other so sparse-view reconstructions become both more visually complete and geometrically faithful.
Yuetong Wang, Jinsheng Quan, Yi Yang, Yawei LuoRead
SplashSplat reconstructs fleeting real-world liquid splashes in 3D by combining multi-camera masks, coarse fluid motion, and dynamic Gaussian primitives.
Peiyu Liu, Dingxi Zhang, Federico Tombari, Marc Pollefeys, Christina Tsalicoglou, Daniel BarathRead
Load more videos
115
Nlp
Independently trained language models may spontaneously learn representations that can be rotated and transferred across languages, suggesting multilingual capabilities without joint training.
Ej Zhou, Suchir Salhan, Catherine Arnett, Anna KorhonenRead
This study shows that how a tokeniser searches may matter more for language-model efficiency than what objective it optimises.
Ahmetcan Yavuz, Clara Meister, Tiago PimentelRead
This paper shows that two classic search-ranking methods, TF-IDF and BM25, can be understood precisely as measuring KL divergence between probability models.
Ivan SilajevRead
116
Efficiency
DeepSeek-V4.1-Flash targets million-token agents by dramatically compressing KV caches while maintaining strong multimodal performance.
DeepSeek-AI, :, Anyi Xu, B. Li, Bangcai Lin, Bing Xue, BingCheng Xian, Bingzheng Xu, Bochao Wu, Bowei Zhang, Boyi Deng, C. C. Yu, Chao Jin, Chaofan Lin, Chen Dong, Chenbing Wang, Chenfan Feng, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chengyuan Zhang, Chenhao Xu, Chenqi Zhao, Chenze Shao, Chuhao Wang, Chuqi Zhang, Damai Dai, Dejian Yang, Deli Chen, Di Huang, Di Wu, Donghao Li, Erhang Li, Eric Fu, F. Zhou, Fangwei Zhou, Fangyun Lin, Fangzhou Yuan, Feiyu Xia, Fucong Dai, Guangbo Hao, Guanglin Li, Guanting Chen, Guoai Cao, Guofan Fan, Guolai Meng, Guowei Li, Haichuan Zhang, Haiyang Ma, Haiyang Shen, Han Li, Han Yu, Han Zhang, Hangyuan Deng, Hanwei Xu, Hanxiang Xu, Hanxun Zhong, Hao Guo, Hao Jiang, Hao Li, Hao Qin, Haodong Wen, Haofen Liang, Haofeng Huang, Haohua Liu, Haoling Zhang, Haoming Luo, Haoran Yang, Haotian Xu, Haotian Yuan, Haoting Huang, Haowen Luo, Haoyang Cai, Haoyu Chen, Haozhe Ji, Hengran Zhang, Hengrui Wang, Hengxu Wu, Honghui Ding, Hongxuan Tang, Huadong Wang, Huanqi ...Read
This work makes trillion-scale, million-token MoE training more feasible by flattening every major GPU memory spike instead of merely reducing the average footprint.
Shrey Pandit, Xuan-Phi Nguyen, Yiran Zhao, Shafiq JotyRead
Edge0 uses predicted expert routing to stream a 35B mixture-of-experts model from SSD fast enough to run on a 24GB consumer machine.
Yu Lin, Yiming Wang, Runyuan Cai, Hanze Liu, Xiaodong ZengRead
117
Diffusion
The work gives diffusion language models a small set of learned memory tokens so they can preserve reasoning across multiple chunks without retaining the generated text.
Albert Ge, Chandan Singh, Yufan Zhuang, Xiaodong Liu, Jianfeng Gao, Frederic SalaRead
SlotDiT uses object-centric slots instead of conventional image latents to make diffusion-based robotic video prediction more structured, efficient, and task-effective.
Gjergj Plepi, Sven BehnkeRead
This work argues that transformers can generate data in context by internally simulating diffusion-like sampling processes.
Arman Adibi, Alireza Jafari, Mohammad Ghavamzadeh, Hadi DaneshmandRead
Load more videos
118
Generative Model
MovieGrid turns long videos into jointly modeled spatial grids, helping generative models produce more coherent stories with many connected shots.
Jiawei Mao, Haoqin Tu, Hardy Chen, Yuhan Wang, Keyang Xu, Jieru Mei, Hongliang Fei, Ruogu Fang, Wei Shao, Cihang Xie, Yuyin ZhouRead
SAM3D-Part lets users ask for specific components of a 3D object and generates complete, correctly placed meshes rather than decomposing the entire object.
Jiahao Chang, Dong Du, Wanhu Sun, Yujian Zheng, Chuanyu Pan, Bowen Zhao, Chongjie Ye, Yuanming Hu, Xiaoguang HanRead
Vidu S2 brings real-time, interactive, editable, and potentially spatial AI-generated video to live applications.
Jintao Zhang, Kai Jiang, Jintao Chen, Xu Wang, Deyuan Liu, Jungang Li, Dechuang Chen, Ming Lin, Jingjiang Zhou, Haopeng Jin, Qi Jia, Xiaohang Wang, Yaole Wang, Zhanqiang Zhang, Ran Li, Zhengkun Huang, Shuyue Xiong, Yuji Wang, Zikun Dai, Hui He, Yang Luo, Mang Ning, Weiqi Feng, Chengyang Ye, Xinyue Lin, Min Zhao, Hongzhou Zhu, Hengkai Tan, Zeyuan Wang, Chendong Xiang, Kaiwen Zheng, Zhijie Deng, Fan Bao, Jianfei Chen, Jun ZhuRead
119
Optimizer
A new zeroth-order optimizer adaptively uses unreliable directional hints to approach first-order performance without needing to know how good those hints are.
Alexander Ryabchenko, Jian Qian, Wenlong MouRead
This work argues that successful fine-tuning depends less on how far a model moves than on choosing the right direction while keeping behavioral drift within a fixed budget.
Fei Yuan, Changjiang Gao, Yilei Tu, Yifeng Liu, Shujian Huang, Yu QiaoRead
This work treats probability distributions like physical systems, finding efficient paths between them using information geometry and a least-action principle.
C. Emre Koksal, Deniz SargunRead
120
Transformer
This work studies how Transformer layers change representations by separating updates that reinforce existing directions from those that redirect them, revealing implications for editing, compression, and training.
Shwai He, Haichao Zhang, Shen YanRead
This paper tests whether transformers really need positional encodings to recognize token distances, revealing when different schemes help or hurt generalization to unseen delays.
Daniel Henrik Nevermann, Claudius GrosRead
RenderFormer-V2 uses specialized transformer attention to render complex scenes with diverse materials, environments, and volumetric effects without retraining for each scene.
Chong Zeng, Yue Dong, Pieter Peers, Lvmin Zhang, Maneesh AgrawalaRead
Load more videos
121
Foundation Model
The paper maps how foundation models are transforming game playing, design, development, adaptation, and testing while emphasizing that game-specific validation remains essential.
Meng Luo, Yanlin Li, Hao Li, Hongzhan Lin, Pengfei Zhou, Tianjie Ju, Ran Zhang, Yeying Jin, Mong-Li Lee, Wynne HsuRead
VidaForge makes video foundation-model data pipelines inspectable and reproducible, helping researchers test how dataset curation choices shape model performance.
Yan Ma, Jiadi Su, Zhulin Hu, Ethan Chern, Linhao Zhang, Tiantian Mi, Pengfei LiuRead
ZGCM-1 shows how a fully open 7B model can combine efficient training, long-context reasoning, and tool-using agents to compete with much larger systems.
Jiyan He, Guang Liang, Hao Liu, Haoxiang Guan, Jinbo Sun, Junyi Guo, Wenjun Feng, Yantai Xie, Yifei Shen, Bin Shao, Chuyang Wei, Kai Chen, Kexin Zhou, Minghang Zhu, Shuxin Zheng, Tie-Yan Liu, Taine Zhao, Wenhui Zhu, Xueyin Xu, Xiaoqing Zhang, Yatao Li, Yuxuan RenRead
122
Hardware
GPUThor shows that carefully timed, non-uniform memory access patterns can turn ECC-protected NVIDIA GPUs into practical targets for powerful Rowhammer attacks.
Chris S. Lin, Joyce Qu, Aditya Rajeev, Gururaj SaileshwarRead
The work shows that attackers can inject electromagnetic signals to make hidden audio and sensor data leak from ordinary electronic hardware.
Haoran Yan, Ziyu Shao, Shuhao Zhang, Qinhong Jiang, Yan LongRead
This work proposes a fully local quantum-computing architecture whose continuously operating decoder could make fault-tolerant computation possible in a realistic two-dimensional setting.
Nathaniel Selub, Aditya Bhardwaj, Ethan LakeRead
123
Safety
The paper shows that hidden harmful plans can be injected into an AI’s context, causing it to act unsafely while fooling the very monitors meant to catch it.
Keertana Chidambaram, Andrew Ilyas, Vasilis SyrgkanisRead
The paper shows how a weaker model can turn an aligned model’s harmless answers into a dangerous capability by splitting and recombining the task.
Mark Russinovich, Blake Bullwinkel, Giorgio Severi, Cristian Ovadiuc, Ahmed SalemRead
MOLE tests whether security monitors can catch AI agents secretly stealing data or sabotaging systems—and finds that even strong monitors miss much of the harm.
Aashiq Muhamed, Virginia SmithRead
Load more videos
124
Benchmark
AgentIdeaBench tests whether AI systems can actively explore the literature and generate genuinely useful scientific hypotheses, not merely summarize what they have already seen.
Yunxiang Mo, Tianshi Zheng, Yisen Gao, Rui Wang, Newt Nguyen Kim Hue Nam, Kelvin Kiu Wai Tam, Jiaxin Bai, Yangqiu Song, Ginny Wong, Simon SeeRead
MotionBlind shows that many Video-LLMs can recognize what is happening in a clip while still failing to understand how fast or in which direction it happens.
Dhairya Bhatia, Bishoy Galoaa, Oliver Fritsche, Shahid Kamal, Muhammad Obaidullah Abdul Salam, Umer Saleem, Om Rastogi, Frania Felix Chettiar, Nesli Erdogmus, Sarah OstadabbasRead
This study asks whether popular AI benchmarks measure what they claim, finding that many safety and capability tests may not reliably distinguish the concepts they are intended to evaluate.
Meera Desai, Sang T. Truong, Hanna Wallach, Alex Chouldechova, A. Feder Cooper, Jean Garcia-Gathright, Daniel E. Ho, Abigail Z. Jacobs, Sanmi Koyejo, Nicholas Pangakis, Angelina WangRead
125
World Model
GE-Act 2.0 trains a scalable robot world-action model from scratch and shows that more diverse manipulation data can substantially improve zero-shot control across tasks, embodiments, and environments.
AgiBot Research Team, Renhang Liu, Wenzhi Zhao, Zhuo Yang, Liliang Chen, Pengfei Zhou, Shengcong Chen, Guanghui Ren, Youlun Peng, Rongjun Jin, Nan Wang, Sukai Wang, Xindong He, Jinyuan Feng, Ziyu Xiong, Linqing Zhong, Yifei Wei, Feng Han, Long Zhang, Da Huang, Nanshu Zhao, Chenghao Yin, Mo Wu, Zhaodong Yan, Kongtao Hu, Yuxiang Yan, Aogelijiang Niyazi, Yu Fang, Jia Zeng, Lizhu Meng, Daizhen Lv, Haoyu Cao, Zhiwen Hou, Lianjin Ye, Yuehan Niu, Zhikai Cai, Xuan Hu, Hui Min, Xiongfeng Cai, Yue Liao, Jing Wu, Soujanya Poria, Ye Li, Sanping Zhou, Maoqing YaoRead
A programmable world model combines an explicit game-state engine with generative video to create persistent, controllable interactive worlds.
Zheng-Hui Huang, Guixu Lin, Jiacheng Lin, Yi-Chuan Huang, Ruihan Yu, Muyao Niu, Siqi Yang, Yu-Lun Liu, Yung-Yu Chuang, Kaipeng Zhang, Zhixiang WangRead
This work speeds up the training of autonomous research agents by letting them practice in a learned simulated world instead of repeatedly running costly real experiments.
Xiyuan Yang, Sheikh Sarwar, Jingru Cheng, Zhan Shi, Duanshun Li, Huiyuan Chen, Haiyang Zhang, Xing Fan, Chenlei Guo, Jingrui He, Zhenyu LiaoRead
126
Reasoning
An open Nemotron-based system uses training and test-time search to generate, check, and refine mathematical proofs, reaching the IMO gold-medal threshold.
Ivan Moshkov, Stephen Ge, George Armstrong, Wei Du, Sadegh Mahdavi, Igor GitmanRead
The study argues that when AI reasoning gets harder, models become temporarily chaotic and get trapped near almost-correct solutions.
Jeffrey Lai, Anthony Bao, John Quinn, William GilpinRead
Looped Flows lets models spend more computation at inference time by repeatedly refining noisy predictions, achieving strong results on several challenging reasoning benchmarks.
Ayhan Suleymanzade, Chanhyuk Lee, Floor Eijkelboom, Nicholas M. Boffi, İsmail İlkan Ceylan, Jinwoo KimRead
Load more videos
127
Graph Learning
Graph Machine replaces much of a Transformer with dynamically routed pointer-like edges, achieving near-preserved language-modeling quality while retrieving only a few tokens per head.
Lintai HouRead
02
Graph-Based Personalized Memory for LLM Agents: Representation, Evolution, Retrieval, and Evaluation
A timely survey explains how graph-structured memories can help LLM agents remember, update, and retrieve personalized information over long-term interactions.
Dac Duy Anh Nguyen, Zhangchi Qiu, Shigeng Chen, Alan Wee-Chung LiewRead
OntoKG-EQ turns emerging-market financial analysis into reproducible, provenance-linked answers that can be checked from the final claim back to the original evidence.
Furqan Nasir, Muhammad Atif Saeed, Muhammad Ehsan, Sher Jeel Ahmad, Abdul Moiz AltafRead
128
Scientific Ai
Hakken uses evolving scientific knowledge graphs and language models to propose explainable future discoveries, including two newly validated biomedical interactions.
Tarek R. Besold, Uchenna Akujuobi, Pablo Sanchez, Alessandra Toniato, Kana Maruyama, Jihun Choi, Samy Badreddine, Frederick Gifford, Daniel Evans-Yamamoto, Sucheendra K. Palaniappan, Miquel Ferrer, Kae Nagano, Iris Rossell, Tom Joy, Hatem ElShazly, Chrysa Iliopoulou, Christoph Wehner, Thiviyan Thanapalasingam, Susana Nunes, Pedro G. Cotovio, Peter Wurman, Peter Stone, Hiroaki Kitano, Michael SprangerRead
An LLM repeatedly rewrites a circle-packing solver, beats 10 established records, and does it for less than $28.
Wes SanderRead
An AI scientist discovers theories from simulated people that surprisingly predict how real humans make decisions.
Akshay K. Jagadish, Younes Strittmatter, Nori Jacoby, Eric Schulz, Nathaniel Daw, Thomas L. Griffiths, Suyog H. ChandramouliRead
129
Code Generation
ExecCritic trains one coding agent to create trustworthy tests and another to repair code from them, substantially improving repository-level bug fixing.
Leitian Tao, Baolin Peng, Haorui Wang, Hang Wang, Hao Cheng, Wenlin Yao, Qianhui Wu, Tao Ge, Sharon Li, Jianfeng GaoRead
This work shows that LLM decompilers can produce code that compiles and passes known tests while silently changing program behavior or erasing vulnerabilities.
Chang Liu, Edward Raff, Kristopher MicinskiRead
This study shows that coding models often fix bugs by changing too much, and that targeted training can make their edits smaller, safer, and easier to review.
Tongyao Zhu, Wei Hern Lim, Min-Yen KanRead
Load more videos
130
Embodied Ai
DriveZero teaches autonomous cars to drive beyond human examples by combining foundation-model perception with reinforcement learning in simulated interactive worlds.
Hao He, Chengcheng Hu, Zirun Su, Heng Zhang, Haisong Liu, Jinke Li, Haochen Tian, Zhenwei Shen, Hongyang Li, Zhichao Li, Yunchen Yang, Bochao Huang, Siyu Zhang, Kuangye Chen, Xiongjie Zhang, Wentao Dai, Hengchen Dai, Siyuan Liu, Zehao Huang, Naiyan WangRead
IMLE-VLA replaces slow iterative action sampling with a single-step multimodal generator, making vision-language-action robots faster, smoother, and more effective.
Kian Hosseinkhani, Qinhe Peng, George Shramko, Mehran Aghabozorgi, Jianing Qian, Tristan Engst, Alireza Moazeni, Dinesh Jayaraman, Ke LiRead
OpenWAM turns world-action model training into an open, modular science experiment and shows how video knowledge and robot experience can combine to improve control across simulations and real robots.
Yuran Wang, Siqiao Huang, Mingleyang Li, Chenhao Zhang, Jiaqi Liang, Weiyang Jin, Yue Chen, Xuemin Chi, Donghao Zhou, Qize Yu, Yu-Kai Wang, Yuhan Rui, Shenzhe Yao, Zhen Yuan, Zhenhao Shen, Kefei Zhu, Zijie Zhu, Ning Gao, Xiaowei Chi, Guanqi He, Shanghang Zhang, Hao Dong, Lin Shao, Hang ZhaoRead
131
Robotics
3DWay helps robots plan more reliably by turning multi-view visual predictions into geometrically consistent 3D movement waypoints.
Ziqin Huang, Yingyue Li, Chenyangguang Zhang, Ruida Zhang, Yuxin Chen, Gu Wang, Xingyu Liu, Masayoshi Tomizuka, Xiangyang JiRead
02
Adaptive Vision-Language Grasping via Composable Foundation Priors and Generalizable Grasp Synthesis
AdaRoboVLG lets robots adapt how they grasp objects by combining a reusable physical grasping policy with task-specific knowledge from foundation models.
Sixu Yan, Shikang Wang, Binhua Huang, Xuanlai Tang, Guohua Fan, Fan Huang, Haoxuan Li, Yongkang Li, Yuhan Li, Bencheng Liao, Zeyu Zhang, Wenyu Liu, Hangxin Liu, Xinggang WangRead
A robot hand learns to write with a grasped pen in seconds by estimating how its movements affect the pen in real time, without simulation or demonstrations.
Kai Stewart, Yasunori Toshimitsu, Robert K. KatzschmannRead
132
Agent
Iris trains large language-model search agents to solve harder multi-hop web questions by repeatedly learning from challenging trajectories and managing context more effectively.
Ziyuan Liu, Hengqi Liu, Zichuan Wang, Yang Qin, Jiachen Liang, Xu Chu, Shaowei Chen, Yuantao Gu, Mu ChuanRead
CORAL uses an LLM agent to continuously tune live recommender systems, improving engagement or reducing serving costs while staying within operational guardrails.
Muhammad Rafay Azhar, Yuhang Zhou, Gilbert Jiang, Yuchen Wang, Rahul Sharma, Matthew DeSousa, Jiayi Liu, Xin Guo, Lizhu Zhang, Xiangjun FanRead
T1 trains a large MoE agent to complete hundreds of shell interactions, substantially improving long-horizon coding and terminal-task performance through verifier rewards and stable RL techniques.
Junyao Yang, Yucheng Shi, Zhongzhi Li, Ruhan Wang, Zongxia Li, Haitao Mi, Leowei LiangRead
Load more videos
133
Rl
The paper argues that agents can develop increasingly complex behaviors simply by seeking experiences that improve their ability to predict and adapt, without external rewards or predefined tasks.
Mikel Malagón, Jon Vadillo, Josu Ceberio, Michael Bowling, Jose A. LozanoRead
ARISE-RL trains agents to improve themselves by generating increasingly challenging tasks and learning from fine-grained rubric-based feedback.
Fanrui Zhang, Ruixue Ding, Qiang Zhang, Xi Chen, Boli Chen, Shihang Wang, Qiuchen Wang, Hongmin Zhan, Jinxin Bian, Li xingchao, Peijin Zheng, Hao cheng, Pengjun Xie, Kaipeng Zhang, Jiawei Liu, Zheng-Jun ZhaRead
RLVR makes models better at one answer by locking them into a narrow path right at the entrance of the reasoning process.
Qiancheng Zhou, Ruizhe LiRead
134
Speech
Text-AB is a large-scale speech generation system that turns text into expressive dubbed voices and natural two-way conversations without requiring explicit text-speech alignment.
Sanyuan Chen, Min-Jae Hwang, Sho Inoue, Anna Sun, Bokai Yu, David Kant, Dongmin Hyun, Dorian Desblancs, Gregory Antonovsky, Oleg Repin, Peng-Jen Chen, Xutai Ma, Zehai Tu, Juan Pino, Wei-Ning HsuRead
AuK is an open-source speech foundation model that can generate and edit audio from natural-language instructions, with a distilled version offering much faster inference.
Ziyang Ma, Zhikang Niu, Wenming Tu, Tianrui Wang, Ruiqi Yan, Junxi Liu, Yanru Huo, Nickk Huang, Yang Liu, Qicong Xie, Zeyu Xie, Hui Wang, Haitao Li, Zixuan Jiang, Yalin Li, Jie Fang, Yifan Duan, Zeyue Tian, Guangzheng Li, Haina Zhu, Shuyi Wang, Jinwen Wang, Mingyu Cui, Tian Tan, Auden, Sen Liang, Steve Yves, Shan Yang, Liefeng Bo, Zilong Zheng, Kai Yu, Eng-Siong Chng, Xie ChenRead
Xiaomi-CocktailASR-1 uses a voice sample as a prompt to transcribe only the desired speaker in noisy conversations, while rejecting cases where that speaker is absent.
Yiru Zhang, Hang Su, Lichun Fan, Ying Zeng, Chang Liu, Yifeng Wang, Yuquan Liang, Tao Li, Lian Li, Wenhao Yang, Jian Luan, Cong Zou, Heng QuRead
135
Multimodal
OmniHallu aims to detect hallucinations across image, video, and audio tasks using claim-level verification and a more efficient learned verifier.
Jianjiang Yang, Peihang Li, Shanqing Xu, Mengchen Qian, Lu Zhang, Meng LuoRead
For long-video AI, choosing the right frames—and reinvesting saved visual tokens—matters far more than simply compressing what the model sees.
Prakhar KhatriRead
This work shows that frozen vision-language models can gain strong multilingual speech and audio-visual abilities simply by connecting them to Whisper, without retraining the models themselves.
Ankan Deria, Hanoona Rasheed, Xilin He, Fahad Shahbaz Khan, Salman KhanRead
Load more videos
136
Cv
AdaptVPR uses controlled generative scene changes and verification to create harder same-place images, making visual place recognition more robust to weather, lighting, and occlusions.
Shunpeng Chen, Jingyi Zhang, Changwei Wang, Shengpeng Xu, Yukun Song, Xingtian Pei, Jinzhou Lin, Li Guo, Shibiao XuRead
Ref-GeNVS helps generative vision systems create realistic new viewpoints of mirror-containing scenes by explicitly using reflections as additional virtual camera views.
GeonU Kim, Shin Dong-Yeon, Tae-Hyun OhRead
Scal3R keeps online 3D reconstruction stable over long videos by querying poses against multiple past keyframes instead of relying on a single distant reference.
Chin-Yang Lin, Yang-Che Sun, Cheng Sun, Fu-En Yang, Min-Hung Chen, Yen-Yu Lin, Wei-Chen Chiu, Yu-Lun LiuRead
137
Efficiency
Layer dropout can make large language models cheaper to train and faster to run without meaningfully sacrificing accuracy.
Mostafa Elhoushi, Alex Pretko, Nolan Dey, Bin Claire Zhang, Gavia Gray, Gurpreet Gosal, Abdulrahman Mahmoud, Shane Bergsma, Joel HestnessRead
This work shows how to predict learning rates and batch sizes for increasingly sparse MoE models, making large-scale training more compute-efficient and transferable.
Changxin Tian, Kunlong Chen, Jia Liu, Ziqi Liu, Zhiqiang Zhang, Jun ZhouRead
Instead of carefully scoring which tokens to keep, this work shows that protecting the prompt and randomly evicting the rest can make long-form LLM reasoning substantially faster.
Heng Wang, Jielin Qiu, Wenting Zhao, Cheng Qian, Liangwei Yang, Jiawei Han, Heng Ji, Silvio Savarese, Shelby Heinecke, Huan WangRead
138
Diffusion
LLaDA-Image offers an open, diffusion-based recipe for building highly capable image generators and distills them into a model that can generate and edit images in just a few steps.
Chuyan Chen, Haoxing Chen, Kun Chen, Zhenglin Cheng, Long Cui, Ruishan Fang, Zhangxuan Gu, Zhicheng Huang, Zhenzhong Lan, Yuanting Lei, Haoquan Li, Jianguo Li, Rongchuan Li, Sidu Li, Tao Lin, Deyuan Liu, Jiacheng Liu, Lin Liu, Yuxuan Lou, Zhisheng Lu, Yuxin Ma, Shuheng Shen, Peng Sun, Chaoyang Wang, Hongjun Wang, Xiaomei Wang, Yongxin Wang, Chengzhang Wu, Hongru Wu, Jun XieRead
Marigold V2 turns powerful diffusion models into fast, sharp, and more generalizable predictors of depth and other dense visual properties.
Igor Pavlovic, Thiemo Wandel, Anton Obukhov, Luca Bartolomei, Andrey Davydov, Fabio Tosi, Matteo Poggi, Sabine Süsstrunk, Dengxin DaiRead
Uno uses discrete diffusion to make standard autoregressive LLMs generate multiple tokens in parallel, delivering up to three times faster inference without sacrificing quality.
Subham Sekhar Sahoo, Lingjie Chen, Khiem Pham, Jonathan Geuter, Chaitanya Dwivedi, Varad Pimpalkhute, Yash Akhauri, Alexander Moreno, Mikhail Yurochkin, Zhenting Wang, Mostafa Elhoushi, Nolan Dey, Shane Bergsma, Joel Hestness, John Thickstun, Eric Xing, Zhengzhong LiuRead
Load more videos
139
Generative Model
Kirin turns large collections of animal videos into a multimodal system that can generate and animate realistic movements for diverse species.
Brian Nlong Zhao, Zhuoyang Pan, James M. Rehg, Jiajun Wu, Shangzhe WuRead
EditVid is a training-free video editor that combines several clever mechanisms to preserve motion, identity, and locality across many kinds of edits.
Adheesh Sunil Juvekar, Onkar Kishor Susladkar, Kiet A. Nguyen, Muntasir Wahed, Nabeel Bashir, Xiaona Zhou, Tianjiao Yu, Vedant Shah, Ismini LourentzouRead
Tencent’s TGR framework replaces fragmented recommendation pipelines with generative ranking, slate generation, and offline reasoning, delivering measurable gains at massive production scale.
TGR Team, Lei Cheng, Haonan Hu, Beibei Kong, Yudong Li, Zang Li, Yunsheng Pang, Hongyang Su, Jianchao Tu, Yunlong Wang, Bing Wen, Junzhang Zhu, Shaojie Zhu, Chengxiang ZhuoRead
140
Optimizer
This paper treats live configuration tuning for models and agents as a bandit problem, enabling systems to learn better inference-time settings directly from noisy production feedback.
Louis Abraham, Tuan-Anh Nguyen, Nicolas DevatineRead
ESPO makes evolutionary prompt optimization more accurate, shorter, and more reliable by diagnosing errors, diversifying candidate prompts, and selecting them with bootstrap stability.
Lihao Liu, Peng Tang, Kunwar Yashraj Singh, Shabnam GhadarRead
This paper develops an information-theoretic speed limit for measuring how quickly SGD learns different hidden aspects of data.
Shuta Kobayashi, Andreas DechantRead
141
Neural Network
HypLTSF maps multi-scale time-series patterns into hyperbolic space so that fine-to-coarse temporal hierarchies become explicit and useful for long-term forecasting.
Namwoo Kim, Hyungryul Baik, Yoonjin YoonRead
The paper helps recurrent networks learn reliably far beyond the sequence lengths they saw during training by stabilizing how future errors credit earlier states.
Hanwen JiangRead
This work shows that anticipating bottom-up signals can make deep continuous-time recurrent networks learn better with fewer parameters.
Shivang Rawat, Mirko Morello, Flaviano Morone, David J. HeegerRead
Load more videos
142
Attention
This work explains how different ways of parameterizing attention can determine whether large models learn useful structure or remain stuck, using a detailed high-dimensional theory.
Yizhou Xu, Margarita Sagitova, Lenka Zdeborová, Florent KrzakalaRead
This work makes linear attention more adaptive by letting its memory track not only what it knows, but also how confident it is.
Ngoc Bui, Tinglin Huang, Rex YingRead
LGA is an efficient attention mechanism that models irregular, long-range time-series dynamics without sequential numerical solvers.
Yiheng Jiang, Yuanbo Xu, Yongjian YangRead
143
Transformer
ReST adapts scalable Transformers for industrial recommendation by reusing user-history computation across candidates while improving ranking quality under strict latency constraints.
Jie Chen, Xiangqian Yu, Yanchao Lian, Tan Lu, Run Yang, Zhengchun Shang, Xing Wang, Cheng Chen, Ke Hu, Qiang Li, Tianjiu Yin, Xiaobing LiuRead
TokenMatch uses curvature-aware mesh tokens and attention to rapidly match corresponding points across difficult 3D shapes, including partial and heavily deformed ones.
Adeela Islam, Zorah Lähner, Vittorio Murino, Vladislav GolyanikRead
The paper shows that putting a model’s reasoning trace before a long document can dramatically improve its ability to solve long-context reasoning tasks.
Xu Zou, Jie TangRead
144
Llm
A new distillation strategy uses teacher confidence to help language models gain reasoning skills without sacrificing the factual knowledge they learn during mid-training.
Jacqueline He, Howard Yen, Shuyue Stella Li, Margaret Li, Hanqing Zeng, Yinglong Xia, Benyu Zhang, Zhuokai Zhao, Qiang Zhang, Pang Wei Koh, Luke Zettlemoyer, Wen-tau YihRead
This study finds that on-policy distillation of language models may need far fewer examples than expected, but many more training steps to fully absorb the supervision.
Zixuan Fu, Bingxiang He, Yuxin Zuo, Haohuan Huang, Jinqian Zhang, Ruhang Xiao, Cheng Qian, Qinyu Luo, Huan-ang Gao, Yudong Wang, Zhiyuan Liu, Ning Ding, Chaojun XiaoRead
StudentSim teaches LLMs to imitate individual learners and respond realistically to tutoring, enabling more personalized AI tutors.
Ke Yang, Chenglong Wang, Michel Galley, Chandan Singh, Jeevana Priya Inala, ChengXiang Zhai, Jianfeng GaoRead
Load more videos
145
Foundation Model
Sensori turns a day of ordinary wrist movement into a reusable health representation that can improve prediction of many diseases without requiring clinical visits or retraining.
Yong Wang, Dylan McGagh, Katya Broomberg, Zizheng Zhang, Jonathan Carter, Junayed Naushad, Laura Brocklebank, Yang Sun, George Nicholson, Dianjianyi Sun, Canqing Yu, Jun Lv, Maxim Barnard, Hubert Lam, Andrew Steptoe, David W. Eyre, Liming Li, Zhengming Chen, Naomi Wray, Spiros Denaxas, Gary S. Collins, Huaidong Du, Aiden Doherty, Hang YuanRead
This study maps where video models learn motion and anomalies, finds weak intuitive physics, and uses latent geometry to create smoother camera-motion trajectories.
Sharon S. Musa, Fereshteh Forghani, Harrish Thasarathan, Sonia Joseph, Matthew Kowal, Konstantinos G. DerpanisRead
Xiaomi-TabLDM is a synthetic-data-trained tabular foundation model that aims to deliver strong, efficient classification and regression without task-specific fine-tuning.
Xiaomi-TabLDM Team, :, Penghui Wang, Wei Liu, Hong Wang, Chengyue Huang, Yuxi Sun, Zirui Wang, Hongming Huang, Quan Wang, Chunxiao Liu, Erli Meng, Bin WangRead
146
Hardware
An AI agent designed a post-quantum crypto chip, and rigorous randomized testing exposed and fixed a hardware verification gap that standard tests missed.
Jungmin Park, Eunha Kim, Wooseop Kim, Seongjoon Cho, Byungho ChaRead
This work redesigns FlashAttention for Blackwell’s FP4 hardware, achieving faster inference and training while showing that aggressively quantized distributed training can become unstable.
Robert HuRead
MaxKernel uses collaborating AI agents and compiler feedback to automatically discover high-performance TPU kernels that can rival expert optimization.
Shangkun Wang, Nina Cai, Charles Hoong, Julian Walker, Gerson Kroiz, George Vanica, Deepak Patil, Andi Gavrilescu, Hassan Sipra, Sethu SankaranRead
147
Safety
A rigorous audit finds that roughly one in three AI-generated clinical notes contains a verified failure, while showing that the measurement method itself can dramatically change the result.
Sebastian Fox, Luke Markham, Ryan Lail, Michael KarotsierisRead
Giving coding agents a clear way to report broken infrastructure may turn reward-hacking behavior from a security risk into a useful debugging signal.
Francesca GomezRead
Reasoning traces are less trustworthy for detecting hidden preferences when those preferences arrive indirectly through tools or raw evidence.
Aryo Pradipta Gema, Neel Rajani, Rohit Saxena, Wai-Chung Kwan, Pasquale MinerviniRead
Load more videos
148
Benchmark
τ^τ-Bench tests whether AI coding agents can build production-ready customer-service agents, revealing that even strong models succeed on fewer than one-quarter of realistic simulations.
Quan Shi, Keshav Dhandhania, Karthik Narasimhan, Victor BarresRead
LLM judges may not be stable measuring tools at all, because identical requests can produce different rankings across time, providers, and server conditions.
Haoyaun Zhu, Jie ZhangRead
Replay-based tests may judge LLM routers in a world that agents never actually encounter, so this paper replaces static logs with live branching rollouts.
Ashritha GonuguntlaRead
149
Continual Learning
When an AI agent gets a new model, structured memories survive far better than compressed notes or partially migrated retrieval indexes.
Ankit Goyal, Jaideep RayRead
The work shows how agents can learn a user’s evolving standards through repeated interaction, becoming substantially better at personalized writing and visual-creation tasks.
Zora Zhiruo Wang, Apurva Gandhi, Rulin Shao, Aspen Chen, Jonas Mueller, Zhiqi Liang, Jett Chen, Michael Ryan, Qianou Ma, Luxi He, Zhoujun Cheng, Andre He, Seungone Kim, Jiayi Geng, Mingqian Zheng, Weiwei Sun, Zheyuan Zhang, Xinran Zhao, Yike Wang, Abe Hou, Liwei Jiang, Pang Wei Koh, Diyi Yang, Graham Neubig, Daniel FriedRead
This paper turns attention into an online learning memory, showing how fast-weight updates can help sequence models retain and use information over long contexts.
Yifan Zhang, Steve Ta, Jasper Zhang, Jichen Feng, Shuzhen Li, Yongxin Zhang, Yifeng Liu, Huizhuo Yuan, Mengdi Wang, Quanquan Gu, Andrew Chi-Chih YaoRead
150
World Model
SolarWM is an open, scalable toolkit for training video world models that can simulate interactive environments over minutes or even hours.
Junchao Huang, Guian Fang, Shengju Qian, Xianghao Kong, Zhuoran Zhao, Wei Huang, Yihua Du, Zixin Zhang, Justin Cui, Yuchao Gu, Yukang Chen, Xinting Hu, Tianyu He, Shaoshuai Shi, Zhuotao Tian, Xin Wang, Mike Zheng Shou, Li JiangRead
WorldSculpt turns partial multi-view videos into large, editable 3D worlds made of individually generated object meshes, even in heavily cluttered scenes.
Muyao Niu, Jixuan He, Ruihan Yu, Lian Fu, Yonghao Yu, Zheng-Hui Huang, Yifan Zhan, Fengbo Lan, Yongtao Ge, Yinqiang Zheng, Kaipeng Zhang, Zhixiang WangRead
ZimaBlue turns massive amounts of human and robot video into real-time, generalizable robot manipulation skills.
Xionghao Wu, Yijun Yang, Shiyang Zhou, Haoze Sun, Jianhui Liu, Songsong Yu, Jiyao Zhang, Wenbo Li, Bo Wang, Guoqing Ma, Lin Song, Renjie Liao, Shenghe Zheng, Wei Tang, Xiaojuan Qi, Yanwei Li, Yuan Zhang, Zhuotao Tian, Haoyang Huang, Nan DuanRead
Load more videos
151
Reasoning
This study finds that chain-of-thought text can hint at which steps matter, but remains far from a reliable explanation of how models actually reason.
Kevin Du, Alexander Hoyle, Laura Ruis, Acyr LocatelliRead
The paper argues that neural networks, including LLMs, may secretly organize their continuous representations according to discoverable symbolic structures.
R. Thomas McCoy, Paul Soulos, Tal Linzen, Paul SmolenskyRead
This work argues that Evolution Strategies can preserve broader reasoning diversity than GRPO, improving LLM Pass@K performance while complementing GRPO's strength in Pass@1.
Yunpeng Ba, Zhi Zheng, Yue Xie, Jiaqing Li, Xialiang Tong, Tao Zhong, Mingxuan Yuan, Zhichao Lu, Xuyang Wu, Zhenkun WangRead
152
Graph Learning
CNY teaches language models to choose which graph neighbours to inspect while reasoning, rather than forcing them to rely on a fixed context.
Yilun Liu, Boyu Luo, Yanran Tang, Ruihong Qiu, Zi HuangRead
The paper shows that feeding graphs to an LLM is not enough—the graph must be presented in a form the decoder can actually use.
Xiaoyu Guo, Pengcheng Chen, Jiong Yu, Yi Lu, Yaohua Wang, Ziyang LiRead
This survey explores how AI can combine the structure of graphs with the visual intuition humans use to understand them.
Xinjian Zhao, Wei Pang, Zhixuan Yu, Xiangru Jian, Xiaozhuang Song, Yaoyao Xu, Zhongkai Xue, Dingshuo Chen, Shu Wu, Philip Torr, Tianshu YuRead
153
Scientific Ai
PPE is an autonomous geospatial AI pipeline that finds the right data and models itself to make better predictions about health, disasters, food security, and disease spread.
Evelyn Ma, Rama Kumar Pasumarthi, Kishwar Shafin, Mandar Sharma, Mimi Sun, Hamed Sadeghi, Dav M. Ebengo, Mbulayi Onesime, Rouslan Solomakhin, John Wamburu, William Ogallo, Aisha Walcott-Bryant, Sanxing Chen, Arbaaz Muslim, Yael Mayer, Ronald Ho, Roy Lee, Ruth Alcantara, Abdoulaye Diack, Monica Bharel, Lambert Rosique, Jeremy Amez-Droz, Christopher Haire, James Manyika, Yossi Matias, Niv Efron, Gautam Prasad, Shravya ShettyRead
By turning abundant coarse weather reanalysis into high-resolution training data, this work shows that better data scaling—not just better models—can substantially improve global forecasts.
Yang Zhao, Peisong Niu, Tian Zhou, Ziqing Ma, Guanlong Ma, Rong Jin, Huiling Yuan, Liang SunRead
03
WeatherNext 3: Increasing resolution and performance of global weather models with raw observations
WeatherNext 3 brings AI weather forecasting closer to operational models by combining raw satellite and station observations with hourly, high-resolution global predictions.
Stephan Rasp, Boris Babenko, Dominic Masters, Andrew El-Kadi, Samier Merchant, Guy Shalev, Ilan Price, Fred Zyda, Remi Lam, Sasha Shysheya, Matthew Willson, Stratis Markou, Shreya Agrawal, Suhani Vora, Mohammed Alewi Hassen, Sunny Mak, Tom R. Andersson, Megan Bela, Akib Uddin, Nofar Peled Levi, Ben Gaiarin, Ferran Alet, Aaron Bell, Peter Battaglia, Alvaro Sanchez-GonzalezRead
Load more videos
154
Code Generation
CogEvol is a compact, production-trained model that turns course briefs into reliable slides and interactive educational web pages much faster and more cheaply than larger coding models.
Shangqing Tu, Daniel Zhang-Li, Yucheng Wang, Shiyu Gan, Yanpeng Wang, Huiqiang Rong, Mofei Chen, Shen Yang, Yini Chen, Yinuo Duan, Haoxuan Li, Binglin Liu, Ye He, Danqi Zheng, Zhanxin Hao, Yuxuan Wu, Mengting Tao, Yuqiu Liu, Jifan Yu, Juanzi Li, Bin Xu, Lei Hou, Huiqin Liu, Yu ZhangRead
A specialized language model system combines curated coding data, reinforcement learning, and iterative solution refinement to surpass human gold-medal performance in programming competitions.
Aleksander Ficek, Sean Narenthiran, Mehrzad Samadi, Somshubra Majumdar, Boris GinsburgRead
03
XREPOTEST: Benchmarking Multilingual Repository-Level Unit Test Generation for Large Language Models
XREPOTEST tests whether today’s LLMs can generate genuinely useful unit tests inside real multilingual software repositories.
Dung Le Quang, Dong Cao Van, Nam Le Hai, Linh Ngo Van, Anh M. T. Bui, Phuong T. NguyenRead
155
Embodied Ai
A large-scale tactile foundation platform that combines sensors, 30,000 hours of visuo-tactile data, transferable representations, and benchmarks for robot manipulation.
NeoteAI Team, Fudan TEAI TeamRead
LM-X helps generalist robots act more reliably by predicting what task stage they are in, what event comes next, and how uncertain their movements are.
Jin Lou, Zhiyuan Jing, Andong Chen, Xupeng Wang, Yuan Xu, Yuexuan Li, Xingdong Zhu, Zhijie Zhu, Yingwei Ji, Wenpeng Nie, Yufei Liu, Boyang Xing, Lei Jiang, Yan Cui, Ying Chu, Jingxuan Zhu, Jingyi Li, Liangliang Chen, Jinyan Liu, Zhiqi Song, Jidong Zhang, Hongming Li, Yuchen ZhuRead
Qwen-Drive-1.0 turns a general vision-language model into an autonomous-driving system that can understand 3D scenes and plan vehicle motion.
Xin Zhou, Zongchuang Zhao, Zhibo Yang, Mingsheng Li, Humen Zhong, Shuai Bai, Du Chu, Ruizhe Chen, Zhaohai Li, Jun Tang, Qiuyue Wang, Mingkun Yang, Jiazhao Zhang, Dayiheng Liu, Dingkang Liang, Xiang BaiRead
156
Robotics
ChainSplat combines physics-inspired articulated modeling with Gaussian splatting to reconstruct and control deformable objects such as cables from ordinary multi-view video.
Seungyeon Kim, Noémie JaquierRead
DemoMimic teaches robot hands to generalize dexterous manipulation by concentrating on the local geometry where contact occurs.
Satvik Sharma, Samrat Sahoo, Huang Huang, Fei-Fei Li Jiajun Wu, Dorsa Sadigh, Jeannette BohgRead
A bimanual robot learns five juggling patterns safely in under five minutes by refining imperfect prior models with its own real-world experience.
Taeyoon Lee, Chunpeng Wang, Christopher G. Atkeson, Alfred A. Rizzi, Nicolas RojasRead
Load more videos
157
Agent
HarnessDev asks whether LLMs can build and improve the software infrastructure that makes their own agents effective.
Yuhao Wu, Jingyuan Zhang, Jiajun Shi, Xinping Lei, Qingshui Gu, Yuxuan Zhang, Zexuan Wang, Chen He, Chen Huang, Maojia Song, Zhiyuan Zeng, Shaowen Wang, Jinkai Liu, Yunfeng Shi, Jiaheng Liu, Shen Yan, Wenhao Huang, Ge Zhang, Wenxuan ZhangRead
SwarmWorld shows how language-model agents can collectively evolve technologies through shared environments, physical artifacts, and emergent social organization rather than direct coordination alone.
Subhadeep Pal, Fiona Y. Wang, Markus J. BuehlerRead
A simple inference-time steering vector lets LLM agents smoothly dial tool use up or down, improving the tradeoff between answer quality, cost, and unwanted actions.
Yuqi Chen, Vincent Siu, Yang Liu, Dawn Song, Chenguang WangRead
158
Rl
The paper argues that on-policy distillation may improve reasoning less by copying a teacher than by suppressing unlikely tokens, enabling a simpler teacher-free alternative that performs even better.
Yi Ding, Ruqi ZhangRead
The study argues that much of RL’s apparent reasoning improvement may come from making models sample promising existing reasoning paths more efficiently.
Wenhe Sun, Cunxiang Wang, Zijun Yao, Yixin CaoRead
This work shows that tiny language models can act as fast, effective rubric judges, potentially making reward-based training far cheaper than relying on large LLM evaluators.
Fengyu Xie, Yilun Zhao, Bingsen Chen, Arman Cohan, Chen ZhaoRead
159
Speech
LibriBrain100 provides over 100 hours of standardized MEG speech data and benchmarks to accelerate noninvasive brain-to-text research.
Francesco Mantegna, Dulhan Jayalath, Gereon Elvers, Tasha Kim, Benjamin Ballyk, Alex Fung, SungJun Cho, Teyun Kwon, Luisa Kurth, Miran Özdogan, Gilad Landau, Pratik Somaiya, Natalie Voets, Mark Woolrich, Oiwi Parker JonesRead
VibeVoice-ASR-Streaming enables an LLM to transcribe and identify speakers in real time, bringing unified speaker-aware speech recognition closer to voice-agent applications.
Yujie Tu, Zhiliang Peng, Jianwei Yu, Li Dong, Songchen Xu, Yaoyao Chang, Wenhui Wang, Zilong Wang, Zehua Wang, Yan Xia, Jiajun Zhang, Xie Chen, Furu WeiRead
VoiceMem gives real-time voice assistants separate factual and emotional memories so they can respond faster, more personally, and with greater empathy.
Zhifei Xie, Jiaqi Lang, Ze An, Yifan Zhao, Dongchao Yang, Kai Li, Ziyang Ma, Mingbao Lin, Chunyan Miao, Shuicheng YanRead
Load more videos
160
Multimodal
This work builds a multimodal oncology foundation model that learns evolving patient representations from clinical records, genomics, and pathology to improve outcome prediction and treatment selection.
Eugene Vorontsov, Yi Kan Wang, Alican Bozkurt, Adam Casson, Ludmila Tydlitatova, Michal Zelechowski, Ezra E. W. Cohen, Jyoti D. Patel, Max Banaszak, Caitlin McWilliams, Shane Colley, Kate Sasser, Ryan Fukushima, Eric Lefkofsky, Razik Yousfi, Siqi LiuRead
IntentQA pushes video AI beyond recognizing actions toward explaining why people act, using multimodal context and robustness tests to evaluate genuine intent reasoning.
Jiapeng Li, Ping Wei, Wenjuan Han, Song-Chun Zhu, Lifeng FanRead
This paper finds a tiny set of attention heads that act like visual search engines inside vision-language models, locating the image evidence needed for their answers.
Chanho Park, Daehyeon Choi, Jihyun Lee, Minhyuk SungRead
161
Cv
01
Evaluation Resolution Confounds Learning-Rule Comparisons in Model-Brain RSA of Early Visual Cortex
The study shows that claims about which vision-learning rules best match the brain can change dramatically depending on the image resolution used for evaluation.
Nils LeuteneggerRead
RefVideo-6M is a massive, quality-controlled dataset that teaches video-editing models to follow both instructions and visual references more reliably.
Bojia Zi, Xiaoyan Yang, Yu Zhou, Ruijie Sun, Lihan Zhang, Bin Liang, Kam-Fai Wong, Haibin Huang, Chi Zhang, Xuelong LiRead
GeoNeXt repurposes video generation models to learn depth and surface geometry jointly, achieving strong zero-shot results with far less training data.
Haosen Yang, Jifei Song, Zhensong Zhang, Xiatian Zhu, Jiankang DengRead
162
Efficiency
Puro-2B shows that a reasonably capable 2B language model can be pretrained from scratch for roughly $5,000 using consumer-grade GPUs and an open recipe.
Kairong Luo, Jiarui Cui, Yaorui Yin, Shengqi Chen, Yiming Yang, Linxiang Gao, Yanmohan Wang, Mingzhe Zhang, Kaiyue Wen, Kaifeng Lyu, Wenguang ChenRead
Prompt compression can save tokens in English but silently erase meaning in other languages, so multilingual users need much safer compression budgets.
Mantas LukauskasRead
Qwen3.8-Next combines sparse experts, hybrid attention, external memory, and optimizer changes to achieve competitive performance with far less computation and improved training stability.
Zihan Qiu, Zekun Wang, Xiao Li, Yanpeng Li, Yang Xu, Yixuan Wang, Huaqing Zhang, Rui Men, Bochao Mao, Chengruidong Zhang, Fan Zhou, Hao Luo, Haofeng Huang, Haoran Lian, Haoyan Huang, Hongqing Chen, Jianwei Zhang, Jing Xu, Junjie Wang, Langshi Chen, Liangyu Wang, Linlang Jiang, Man Yuan, Minmin Sun, Peng Jin, Siqi Zhang, Siyu Wang, Xingzhang Ren, Yakai Wang, Yi Zhang, Yiming Dong, Yizhong Cao, Yubo Ma, Yunfei Mao, Bo Zheng, Dayiheng LiuRead
Load more videos
163
Diffusion
4DStreamCtrl aims to make video diffusion interactive by generating long, realistic streams while jointly controlling camera motion, object movement, and depth in real time.
Shiqian Li, Chenguo Lin, Zhiguang Liu, Yu Tang, Jiarong Ou, Rui Chen, Yixin ZhuRead
This work makes multi-reference diffusion editing much faster by caching visual reference computations without sacrificing instruction following or image fidelity.
Yangshuai Liu, Zheming Li, Jiaao Li, Kang He, Ziliang Lai, Zhitai Liu, Chengru SongRead
This work makes pixel-space diffusion easier to train by teaching models to recover images from coarse structures to fine details in the same order humans and vision systems naturally perceive them.
Shaojie Guo, Lichen Ma, Haoyang Tong, Yu He, Zipeng Guo, Xiaoan Liu, Feng Yan, Yu Guo, Fei Wang, Junshi Huang, Yan WangRead
164
Generative Model
EditaLive turns human video editing into a fast, live-streaming experience by adapting and distilling a pretrained animation model while preserving facial expressions and character appearance.
Zhiyuan Li, Chi-Man Pun, Peng-Tao Jiang, Bo Li, Xiaodong CunRead
GenFirst stabilizes end-to-end latent generative models by letting generation shape the latent space before progressively adding reconstruction pressure.
Guangting Zheng, Yiyuan Zhang, Tao Yang, Yunpeng Chen, Rui Zhu, Jiajun Deng, Yanyong ZhangRead
RECAP-Forcing helps long video generators remember newly appearing subjects and scenes instead of simply favoring the latest frames.
Haiyang Xu, Zheng Ding, Zhuowen TuRead
165
Self Supervised
Human-JEPA aims to help vision systems understand people now and predict what they will do next using one efficient self-supervised video model.
Hui Wei, Licai Sun, Guoying ZhaoRead
LeVJEPA aims to make video pretraining dramatically cheaper and simpler by learning useful representations without momentum encoders, stop-gradients, or pixel reconstruction.
Lukas Kuhn, Lucas Maes, Giuseppe Serra, Quentin Le Lidec, Yann LeCun, Randall Balestriero, Florian BuettnerRead
TTPO lets language models improve their reasoning at test time by learning from their own votes while selectively correcting likely mistakes.
Aozhe Wang, Zhengxi Lu, Jianze Wang, Shangke Lv, Ying Liu, Weiming Lu, Jun Xiao, Yueting Zhuang, Hua Yang, Qianglong Chen, Yongliang ShenRead
Load more videos
166
Optimizer
01
Beyond the Stability-Exploration Dilemma: Environmental Regularization for LLM Policy Optimization
ERPO regularizes the queries an LLM trains on instead of directly constraining its responses, aiming for more stable learning without sacrificing exploration.
Xianlei Zhou, Xiangdi Meng, Yu He, Tianyu Qi, Shuyan Guan, Xianli Zhang, Jian Zhang, Xin Li, Qika Lin, Jun LiuRead
The paper argues that language-model training dynamics are governed less by raw learning rate than by its ratio to parameter norm, revealing a common effective-learning-rate scale.
Zihan Liu, Ruiheng Zheng, Shaobo Zhang, Changxin Tian, Kunlong Chen, Zhiqiang Zhang, Lei WuRead
This work argues that Muon wins by allocating step sizes across spectral directions and uses that insight to make training even more token-efficient.
Xiaodong Wu, Wenyi Yu, Chao Zhang, Philip WoodlandRead
167
Neural Network
The study shows that more complex datasets reshape where low-loss parameter neighborhoods contract around trained neural networks.
Jaeyong Bae, Hawoong JeongRead
A compact quantum-inspired neural network improves federated arrhythmia detection while reducing communication costs across hospital and wearable-device clients.
Chun-Hua Lin, Samuel Yen-Chi Chen, Yu-Chao Hsu, Kuo-Chung Peng, Jiun-Cheng Jiang, Chi-Sheng Chen, Tai-Yue Li, Nan-Yow Chen, En-Jui Kuo, Hsi-Sheng GoanRead
A recurrent neural method learns when to classify patients early while balancing accuracy, sensitivity, specificity, and monitoring costs with explicit performance guarantees.
Jiaming Qiu, Yingye Zheng, Ying-Qi ZhaoRead
168
Transformer
A time-aware Transformer uses daily ventilator data to predict COPD flare-ups earlier without waiting for delayed clinical measurements.
Weihao Qu, Ling Zheng, Dongyang Wang, Jiacun Wang, Haowen PanRead
This paper offers a four-part language for explaining how transformers align, combine, suppress, and route information across different models.
Nura Aljaafari, Andre FreitasRead
A fast two-pass test reveals when sequence models secretly use future information, even when their attention masks look correct.
Taebong Kim, Youngsik Hong, Minsik Kim, Sunyoung Choi, Jaewon Jang, Minseo KimRead
Load more videos
169
Foundation Model
EXAONE Tabular shows that a compact, causally synthetic-pretrained foundation model can deliver strong tabular predictions without fine-tuning, at far lower cost than much larger alternatives.
Moonjung Eo, Min-Kook Suh, Hye-Seung Cho, Jiwon Kim, Seoyoon Kim, Sangjun Nam, Soonyoung LeeRead
OneModel unifies user behavior across feeds, ads, and commerce into one production-scale ranking system that improves engagement and business metrics.
Yinqi Zhang, Peiyu Hu, Yuntian Tang, Siying Gu, Jiahao Liang, Longxin Kou, Haiqing Hu, Shuman Zhuang, Yubin Xu, Chenggen Sun, Bin Ye, Donghui Xu, Zhaoyu Liu, Jiang Rong, Yuting Jia, Zhaokai Luo, Leilei Ma, Yiying Xie, Yao HuRead
RIBOSPAN is a long-context RNA foundation model designed to understand, represent, generate, and redesign complete transcripts.
Ziyuan Wang, Bohao Tang, Fei Zhang, Shuo Han, Pengfei LiuRead
170
Hardware
One researcher uses AI agents and machine-checked verification to move from software ideas to a taped-out RISC-V chip without writing RTL by hand.
Jason HickeyRead
Redwood is an AI-designed accelerator that claims to go from specification to verified silicon-ready hardware and efficient model execution in just two weeks.
Architect LabsRead
Researchers combine individually controlled rubidium atoms with a fiber cavity to build a programmable platform for quantum networks and many-body quantum optics.
Stephan Roschinski, Johannes Schabbauer, Franz von Silva-Tarouca, Marvin Holten, Damien Bloch, Julian LéonardRead
171
Safety
EVOMAL shows how coding agents can unknowingly copy and spread malicious skills through their own self-improvement loops, while proposing a prompt-based defense to contain the attack.
Xiaodong Wu, Yu Shi, Qi Li, Zhimin Zhao, Xiangman Li, Bram Adams, Ahmed E. Hassan, Jianbing NiRead
EchoCoT shows that hidden reasoning traces from powerful AI models may be recoverable through carefully designed API interactions, creating a major security concern.
Yiting Qu, Ziqing Yang, Chi Cui, Ye Leng, Junjie Chu, Yang ZhangRead
A large study finds that open-weight language models can contain information about changes to their own internals but generally cannot reliably tell us about those changes.
Emilio FerraraRead
Load more videos
172
Benchmark
HealMed tests whether medical language models remain reliable across languages and shows that low-resource performance and translation quality can substantially change the results.
Yingjian Chen, Fan Gao, Sherry T. Tong, Haoyu Zhang, Aosong Feng, Kevin W. Jin, Xing Wu, Jinghui Lu, Abdul Samad, Akbar Faruqi, Cesar Caraballo, Cibele Brandão, Dhruva, Gupta, Eunji Jeon, Gabriel Madera-Santiago, Geon Lee, Hugo Toshio Itikawa, Insook Cho, Isabelli Martins, Isarar Siddique, Israr Ahmed, Jihyo Kwak, Kanyakorn Veerakanjana, Luis Guilherme Cardoso, Minjin Kim, Piyalitt Ittichaiwong, Renee Dua, Santiago Gudiño-Rosales, Xiujie Chen, Zeo Lapalus, Zixin Xu, Michihiro Yasunaga, Rex Ying, Heuiseok Lim, Jaewoo Kang, Chanjun Park, Hang Jiang, Ethan Goh, Hyunjae Kim, Edison Marrese-Taylor, Yusuke Iwasawa, Yutaka Matsuo, Qingyu Chen, Irene LiRead
SWE Refactor Bench tests whether coding agents can truly migrate an entire codebase without merely making the tests pass, and finds that they still struggle significantly.
Deyao Hong, Yizhe Chi, Wenyi Li, Xiaoqiu Wang, Mingju Gao, Kaisen Yang, Bingxiang He, Youjie Zheng, Calvin Xiao, Qinhuai NaRead
VBVR-Pro is a large, verifiable playground for training and testing models that reason directly through generated images and videos.
Junxiang Xu, Ruisi Wang, Fanyi Pu, Maijunxian Wang, Ran Ji, Tongxi Zhou, Chenyang Gu, Jing Zuo, Hongcan Xiao, Yimeng Geng, Wanqi Yin, Wei Chen, Oscar Qian, Zhengan Yan, Ziqi Huang, Haiwen Diao, Liang Pan, Bo Li, Xiangyu Fan, Dezhi Luo, Fengyuan Yu, Zehong Zhao, Qingying Gao, Tinghui Zhu, Yilan Zhang, Jingqi Tong, Pinyuan Feng, Zhengze Jiang, Letian Wang, Ziyu Guo, Renrui Zhang, Jieneng Chen, Sonia Joseph, Constantin Venhoff, Saman Motamed, Mengyue Yang, Chandra Sripada, Alan Yuille, Philip Torr, Lvmin Zhang, Vikash Kumar, Daniel Khashabi, Nikolaus Kriegeskorte, Raphaël Millière, Vincent C. Müller, Anyi Rao, Quan Wang, Ziwei Liu, Dahua Lin, Lei Yang, Hokin Deng, Zhongang CaiRead
173
Continual Learning
01
Inject, Align, Recover: Staged Post-Training for Retrieval-Free Document Knowledge Internalization
IAR teaches language models to memorize and use a document collection without retrieval while preserving their broader abilities.
Qian Kou, Xiaofeng Shi, Xiaosong Qiu, Hua ZhouRead
A small set of carefully chosen past experiences can anchor robot skills and dramatically reduce forgetting while learning new tasks.
Maximilian Du, Zhanyi Sun, Chen Xu, Paarth Shah, Masha Itkina, Shuran SongRead
SPARCL reduces forgetting in exemplar-free continual learning by protecting important spectral directions while updating only the residual feature space.
James Hartley, Zeropy Surio, Daniel Whitmore, Hannah Clarke, Thomas ReedRead
174
World Model
CLAP aims to turn diverse human and robot videos into a shared physical simulator that can predict and control unfamiliar robot embodiments.
Kechen Liu, Ola ShorinwaRead
This study asks whether today’s generative world models can truly replace simulators and finds that they still lack reliable physics, state feedback, and stable long-horizon behavior.
Tong Wang, Huan Deng, Mucheng Yang, Yang He, Xiaohui Kuang, Gang ZhaoRead
PAWBench asks whether video generators merely make plausible futures or actually reproduce the full range and probabilities of how the world can unfold.
Yuandong Pu, Le Zhuo, Sayak Paul, Gabriel Jorge Menezes, Avram Đorđević, Shiyang Li, Yifan Zhou, Bin Fu, Wenlong Zhang, Junjun He, Yu Qiao, Yihao Liu, Jingbo Xing, Xi ChenRead
Load more videos
175
Reasoning
FLARE uses LLMs and Lean proofs to reliably check whether automatically generated optimization formulations are mathematically correct.
Henry Robbins, Connor Lawless, Madeleine Udell, Ellen VitercikRead
This study finds that relying more on AI for logic puzzles may boost short-term performance while weakening the development and retention of independent reasoning skills.
Shang Wu, Catarina G Belem, Shuyuan Fu, Mark Steyvers, Padhraic SmythRead
SRPO teaches language models to turn their own mistakes into step-by-step learning signals, improving long-horizon reasoning and agent performance with relatively little training.
Jialong Liu, Yuling Shi, Ning Yang, Xiaodong Gu, Zuchao LiRead
176
Graph Learning
AdaptedKG uses calibrated knowledge-graph clues to identify and downweight misleading user interactions before they distort sequential recommendation.
Zichun Jin, Zihan Zhou, Yinan Liu, Bin Wang, Xiaochun YangRead
This paper shows that estimating a graph’s spectral structure is exponentially hard classically but can become polynomially efficient with quantum access.
Rong-Hua Li, Meihao Liao, Yichun YangRead
RAD detects unusual behavior in multi-table databases by combining graph-based learning with human-readable behavioral rules.
Noah Dahle, Anne Tumlin, Ngoc Tran, Xenofon Koutsoukos, Tyler DerrRead
177
Scientific Ai
A Gemini-based multi-agent scientist conducts experiments, generates hypotheses, and writes research across multiple fields, suggesting a promising but not yet fully validated path toward automated scientific discovery.
Samuel Schmidgall, Xiaokai Zhu, Marian Shaw, Lin Yang, Valentin Liévin, Jingyun Yang, Yuchen Zhuang, Tim Strother, Alex Bijamov, Min Woo Sun, Anil Palepu, Justin Chen, David Steiner, Jacqueline Shreibati, Wei-Hung Weng, Yilin Zhao, Xingjian Hu, Nicholas Zahn, Sadhya Garg, Julia Kirby, Yuxiang Gan, Jiaoli Li, Divy Thakkar, Shekoofeh Azizi, David Racz, Juraj Gottweis, Vivek Natarajan, Chenglin Wu, Tal Danino, Keran Rong, Haozhe Wang, Benoit Schillings, Yong Cheng, Quoc V. Le, Tao TuRead
A society of AI agents independently collaborates, experiments, and produces new mathematical constructions and theorems in an open-ended research environment.
Stephen Chung, Wenyu Du, William J. WesleyRead
The study finds that frontier LLMs can independently produce near-competitive algorithms for important operations-research problems.
Jackie BaekRead
Load more videos
178
Code Generation
LEGO-RL turns real coding-agent harnesses into trainable reinforcement-learning environments and improves SWE-bench performance across three popular platforms.
Yiming Du, Yuxin Jiang, Tao Yuan, Jianbo Dai, Shaowei Wang, Jierun Chen, Chaofan Tao, Xianzhi Yu, Lifeng Shang, Kam-Fai Wong, Xiaohui Li, Haoli BaiRead
Repo0 helps coding agents build entire software repositories by first evolving and stabilizing an explicit modular architecture before writing the code.
Silin Chen, Haoyi Teng, Xiaodong Gu, Yuling Shi, Jiale Huang, Yongpan Wang, Hongyu Zhang, Haibing GuanRead
RubSE helps vision-language models iteratively repair generated interfaces by turning visual feedback into prioritized, structured rubrics instead of letting each code edit disrupt the whole page.
Tianyi Xiong, Zhengyuan Yang, Xiaofei Wang, Chung-Ching Lin, Ruichun Ma, Kevin Lin, Zhendong Wang, Linjie Li, Chenxi Liu, Ruibo Chen, Ramani Duraiswami, Heng Huang, Lijuan WangRead
179
Embodied Ai
ADEPT teaches dexterous robots reusable manipulation skills through RL pretraining, enabling them to transfer complex visual-tactile behaviors to new tasks and real-world hardware.
Jayjun Lee, Jessica Yin, Asif Rana, Nicholas Blauch, Sam Mady, Mohak Bhardwaj, Nima Fazeli, Nathan Ratliff, Karl Van Wyk, Ankur HandaRead
This survey explains how smart glasses could evolve into trustworthy embodied AI systems that continuously perceive, remember, and act alongside their users.
Jiangning Zhang, Haojun Chen, Yong LiuRead
Zero-WAM teaches robots unseen manipulation tasks by showing them human videos, bringing in-context learning from language models into embodied robotics.
Jiaming Zhou, Qihang Zhang, Gangwei Xu, Cunxin Fan, Yujie Zhao, Ruilin Wang, Yiming Luo, Shuai Yang, Xing Zhu, Yujun Shen, Junwei Liang, Yinghao XuRead
180
Robotics
01
CoToGrasp: Contact-Topology-Conditioned Dexterous Grasp Synthesis via Canonical Workspace Learning
CoToGrasp learns to generate functionally meaningful dexterous grasps across unseen objects by conditioning on contact topology rather than object-specific annotations.
Julien Merand, Boris Meden, Liming Chen, Mathieu GrossardRead
NeSAM helps off-road robots predict and navigate deformable terrain by blending soil physics with learned dynamics that adapt online.
Chenhui Pan, Tong Xu, Francesco Cancelliere, Xuesu XiaoRead
SRL-MPC blends reinforcement learning with safety-constrained model predictive control to help differently shaped robots navigate dense crowds more safely and efficiently.
Ruihua Han, Rui Gao, Zhe Liu, Xinyi Wang, Chang Chen, Shuai Wang, Qi Hao, Jia Pan, Hengshuang ZhaoRead
Load more videos
181
Agent
AgentSysBench shows that serving AI agents requires systems designed for long-running, stateful, tool-using workloads rather than just faster LLM inference.
Chaokun Chang, Yukun Zhou, Kaihua Fu, Dakai An, Tianyu Feng, Hanfeng Lu, Sheng Yao, Pu Guo, Yinghao Yu, Yizhou Shan, Bo Li, Binhang Yuan, Wei WangRead
A physics-inspired model predicts when groups of AI agents will reach consensus, polarize, or become indifferent—and how their interactions affect truth and political opinions.
Batu El, Jinhee Paeng, Fatih Dinc, Shiye Su, Mete Erdogan, Aneesh Pappu, Haotian Ye, Wanjia Zhao, Surya Ganguli, James ZouRead
SkillGate helps long-horizon AI agents choose the right procedural skill by giving skill-selection decisions their own dedicated learning signal.
Qingyao Li, Wenxiang Jiao, Shuai Shao, Kangning Zhang, Yuan Lu, Yi Guo, Weiwen Liu, Weinan Zhang, Yong YuRead
182
Rl
A small value model helps large robot policies learn from their own mistakes, substantially improving manipulation without more human demonstrations.
Varun Giridhar, Anant Khandelwal, Jeremy A. Collins, Ignat Georgiev, Animesh GargRead
Co-RL trains diverse language and vision-language agents to improve reasoning by rewarding each other, achieving strong gains without ground-truth labels.
Yunhao Yang, Yuexin Bian, Yunjie Tian, Di Fu, Tianjin Huang, Yuanyuan Shi, Ziang Xiao, Nuno Vasconcelos, Yijiang LiRead
03
WarpSAC: Towards the Pinnacle of Scalable Off-policy RL by Rethinking Exploration and Exploitation
WarpSAC adapts exploration and exploitation strategies to the amount of available simulation data, substantially improving scalable reinforcement learning and robot training.
Zihao Wu, Hongyao Tang, Yi Ma, Huizhong Song, Pengyi Li, Yifu Yuan, Fei Ni, Jinyi Liu, Wei Wei, Jianrong Wang, Yan Zheng, Jianye HaoRead
183
Speech
AnyTalk turns generic video-generation knowledge into real-time, lip-synced 3D animation for almost any character without requiring character-specific motion data.
Kwan Yun, Serin Yoon, Sunjin Jung, Jung Eun Yoo, Inyup Lee, Junyong NohRead
Multi-hop RAG may retrieve more intelligently, but when speech recognition gets an entity wrong, that extra reasoning structure can make the final answer even worse.
Zhenghua BaoRead
03
FireRedTTS3: Unified Speech Generation and Editing with Semantically Enriched Speech Representations
FireRedTTS3 uses semantically enriched speech representations to make voice cloning, controllable speech generation, and audio editing more stable and capable.
Feiyu Shen, Kun Xie, Yichen Wu, Ziqi Dai, Yichen Han, Junjie Li, Xuelong Geng, Fenglong Xie, Lei Xie, Xu Tang, Yao HuRead
Load more videos
184
Multimodal
EXPL-FR turns opaque face-recognition embeddings into prompt-queryable semantic explanations using a lightweight vision-language alignment adapter.
Guray Ozgur, Mustafa Efe Tamyapar, Naser Damer, Fadi BoutrosRead
G-CARL trains multimodal models to explain medical reports more accurately and empathetically by combining evidence-grounded claim verification with personalized communication checklists.
Shiao Xie, Siyu Chen, Jianwei Lv, Bo Yuan, Yujin Wang, Xiandong LiRead
LAION-BVD opens access to 10 million hours of multimodal video data, aiming to make large-scale video, audio, and image pre-training more accessible.
Andreas Hochlehnert, Marianna Nezhurina, Mehdi Cherti, Andrej Radonjic, Thaddäus Wiedemer, Christoph Schuhmann, Romain Beaumont, Wieland Brendel, Bernhard Schölkopf, A. Sophia Koepke, Jenia Jitsev, Matthias BethgeRead
185
Cv
4DAnyone turns an ordinary monocular human video into a consistent, renderable 4D reconstruction by coordinating diffusion-generated views across time and camera angles.
Yudong Jin, Tao Xie, Qihang Zhang, Zehong Shen, Zhen Xu, Yujun Shen, Hujun Bao, Xiaowei Zhou, Yinghao XuRead
Mover360 enables users to move, insert, or remove objects in 360° panoramas with simple point, box, or mask controls while preserving realistic geometry and scene consistency.
Haoyi Zhong, Fang-Lue Zhang, Andrew Chalmers, Taehyun RheeRead
03
Tomatoes, Potatoes, and Onions: Questioning the Need for Faces in Face Presentation Attack Detection
This study shows that tomatoes, potatoes, and onions can help train face anti-spoofing systems by teaching detectors to recognize presentation artifacts rather than facial identity.
Guray Ozgur, Fadi Boutros, Naser DamerRead
186
Nlp
This work turns nearly a million scanned books into a flexible, multilingual research resource by preserving rich annotations instead of making one irreversible cleaning decision.
David Lowry-Duda, Matteo Cargnelutti, Catherine Brobston, Salwa Ismail, Greg Leppert, Amanda Watson, Jonathan ZittrainRead
This paper openly trains strong dense and late-interaction search models and shows that token-level matching can improve multilingual and unseen-script retrieval.
Raphaël Sourty, Antoine Chaffin, Paulo Roberto Moura Junior, Amélie ChatelainRead
The study shows that SFT teaches models to reason visibly in Greek, while reinforcement learning fixes formatting and leakage problems that accuracy benchmarks fail to detect.
Ayoub Kirouane, Christos PetrocheilosRead
Load more videos
187
Efficiency
FreeToken aims to make frontier-scale mixture-of-experts models practical on ordinary personal computers by dynamically coordinating CPU, GPU, memory, and model state.
Shuo Yang, Xiaoze Fan, Melissa Pan, Haocheng Xi, Zhe Wang, Shanlin Sun, Kurt Keutzer, Song Han, Matei Zaharia, Chenfeng Xu, Ion StoicaRead
A smart router decides when it is worth paying for better model-quality estimates before sending each query to the best specialist.
Adam Fisch, Shubhendu Trivedi, Fantine Huot, William W. Cohen, Michael Kaisers, Mirella Lapata, Kate Larson, Jacob EisensteinRead
RMM speeds up LLM inference by selectively skipping less informative parts of matrix multiplications while preserving much of the model's accuracy.
Zixuan Lan, Yanhong Li, Jiawei ZhouRead
188
Diffusion
Abra shows that diffusion image models scale predictably, but reach their best performance with far more training data per parameter than language models.
Kyle Chickering, Wei-An Lin, Swayam Bhanded, Dan Saunders, Akshat Tripathi, Jiaming Song, Shyamal Buch, Xinchen YanRead
This study shows how to efficiently train pixel-space image diffusion models by first learning in latent space, achieving competitive quality with substantially faster inference.
Dengyang Jiang, Ruoyi Du, Zhennan Chen, Dongyang Liu, Zanyi Wang, Mingzhe Zheng, Xiangpeng Yang, Huanqia Cai, Aiming Hao, Yuming Jiang, Peng Gao, Harry Yang, Steven HoiRead
LinCa speeds up diffusion-based image and video generation by learning which parts of intermediate features can be safely reused, achieving substantial acceleration with minimal quality loss.
Jinshan Liu, Haoran Qin, Xiaobing Tu, Jiacheng Liu, Jiahui Hu, Zhengan Yan, Yukun Xie, Kerui Shen, Jinkui Ren, Yuqi Lin, Xiantao Zhang, Linfeng ZhangRead
189
Generative Model
InfinityEdit lets video generators apply ongoing edits to an endless stream while maintaining temporal continuity and stability over time.
Yunze Tong, Mushui Liu, Canyu Zhao, Shiyi Zhang, Didi Zhu, Peng Zhang, Wanggui He, Jinlong Liu, Ying Chen, Hao Jiang, Pipei Huang, Bo ZhengRead
KeyID keeps video characters recognizable by drafting the action first and then repairing identity at strategically chosen keyframes.
Jianjie Luo, Yiming Zhong, Haoming Shen, Yupeng Xiao, Zhenguo YangRead
MegaParts uses compact discrete geometry tokens and long-context autoregressive modeling to generate highly complex 3D objects with hundreds of coherent parts.
Manwen Liao, Xinyu Lian, Jian Mao, Kaixu Chen, Li Luo, Jinghao Yan, Wanshui Gan, Qiao Yu, Weitian Zhang, Chunhua Shen, Guang Chen, Bo Dai, Xudong Xu, Zhaoyang LyuRead
Load more videos
190
Optimizer
sMuon adapts the Muon optimizer to low-rank fine-tuning, offering a simpler and sometimes more effective alternative for LoRA-based training.
Ben Anson, Conor Houghton, Edward MilsomRead
A compute-saving framework predicts the right learning rate for enormous MoE models by scaling up results from much smaller proxy runs.
Nayeon Kim, Hojin Lee, Yunju Bak, Jaesun Park, Boseop KimRead
This work explains why Adam works so well for modern neural networks and proposes a slimmer version that retains its performance.
Yushun ZhangRead
191
Neural Network
This work argues that plateaus, sudden learning, and power-law scaling can all emerge from one simple mathematical model of how neural-network modes turn on.
Liu Ziyin, Yizhou Xu, Tomaso Poggio, Isaac ChuangRead
The Spectral Neuron replaces opaque nonlinearities with controllable eigenvalue-based functions, aiming to make expressive neural models more transparent and mathematically shapeable.
Alex ShtoffRead
A new theory and method lets deep ReLU networks borrow strength across related datasets to improve nonparametric regression.
Junpeng Ren, Carlos Misael Madrid Padilla, Yanzhen Chen, Oscar Hernan Madrid PadillaRead
192
Transformer
This paper develops a mathematical theory explaining exactly when transformers can recognize patterns longer than those seen during training.
Andy Yang, Blerta Veseli, Corentin Barloy, Michaël Cadilhac, Andreas Krebs, Charles Paperman, Howard Straubing, Michael HahnRead
This paper replaces conventional attention with explicit Self-and-Exchange relations, aiming to preserve language-model quality while improving efficiency.
Yuting Ge, Pengju Yang, Mingkai NieRead
Maglev teaches a lightweight recurrent Transformer to preserve long-range context in a compact memory, enabling faster long-context generation without full attention at inference.
Bo Liu, Qiang LiuRead
Load more videos
193
Llm
The study finds that everyday linguistic styles associated with women can cause LLMs to produce shorter and less formal workplace writing, revealing a subtle but consequential form of bias.
Katherine Van Koevering, Anjalie FieldRead
Test-time scaling can generate better answers, but for open-ended tasks the real challenge is reliably choosing or synthesizing the best one.
Davide Romano, Kanak Raj, Jerrod Parker, Daniele GiofrèRead
YOPO lets a frozen language model reason, steer itself, and abstain when uncertain—all in one forward pass.
Ziyang Luo, Zhongyao Chu, Xinjie He, Youting Wang, Xukui Qin, Runxiong Wu, Yan-Syuan ChenRead
194
Foundation Model
This paper shows that time-series models can make deceptively flat forecasts and introduces a method that improves stock ranking without sacrificing calibration.
Shu Wan, Miles Ma, Hank Zhu, Guangqi Liu, Stephen Wang, Qingsong Wen, Huan LiuRead
This survey maps how foundation models can evolve from recognizing people to understanding their actions, interactions, environments, and embodied agency.
Yang Chen, Tianqi Wang, Xiaorui Jiang, Yilei Man, Yihua Shao, Mengyuan Liu, Zhi Chen, Xiaofeng Cao, Qibin Zhao, Chi Harold Liu, Albert Y. Zomaya, Nicu Sebe, Jingren Zhou, Dacheng Tao, Song Guo, Jingcai GuoRead
LittleLearner is a language model trained only on elementary-school-level material, creating a controlled sandbox for studying how models gain knowledge and skills.
Fanfei Li, Jana Zeller, Manuel Prada-Corral, Thaddäus Wiedemer, Prasanna Mayilvahanan, Ryan Cotterell, Wieland BrendelRead
195
Hardware
A 128-GPU cluster assembled from discarded hardware can serve LLaMA-70B cheaply, but its environmental value depends heavily on electricity costs and carbon intensity.
Zeyu Cao, Xuan Guo, Cheng Zhang, Cheuk Hang Lau, Ilia Shumailov, Yiren ZhaoRead
A single controllable qubit dramatically cuts the measurements needed to learn classical signals, suggesting powerful near-term applications for quantum-enhanced sensing.
Ishaan Kannan, Sridhar Prabhu, Saeed A. Khan, Mandar M. Sohoni, Xingrui Song, Saswata Roy, Alen Senanian, Valla Fatemi, Peter L. McMahon, Jordan CotlerRead
Agents iteratively diagnose prefetching failures and synthesize specialized hardware components that reportedly outperform leading human-designed prefetchers.
Xiangfeng Sun, Ceyu Xu, Ningzhi Ai, Zeyu Zhu, Yiyang Yuan, Yuan XieRead
Load more videos
196
Safety
The study shows that language models may leak secrets indirectly through ordinary responses, even when they refuse to reveal them explicitly.
Jaiden Fairoze, Neal Mangaokar, Kamalika Chaudhuri, Sanjam Garg, Saeed MahloujifarRead
The paper shows that many AI compliance detectors may ignore the actual rules they are supposed to enforce, and offers a lightweight way to audit them.
Saisab Sadhu, Aadit Sengupta, Vinay Kumar Sankarapu, Pratinav SethRead
A large-scale audit finds that LLMs’ doctor recommendations are strongly shaped by ratings, fees, position, and subtle demographic signals that their explanations fail to reveal.
Syeda Anshrah Gillani, Mirza Samad Ahmed BaigRead
197
Benchmark
EgoMonth tests whether AI can genuinely remember a person’s daily visual experiences over months rather than merely summarize isolated clips.
Weitao Chen, Hu Jiaxin, Xie Tianyidan, Yang Li, Yuyi Qian, Banghao Xu, Ziheng Tang, Shenyi Wang, Mingyue Yu, Duo Li, Jiacheng Shi, Gao Wang, Zhan Xu, Zhicheng Qiu, Xuanfu Li, Jian Yang, Lanjun Wang, Zili YiRead
A large-scale audit finds that AI assistants systematically overlook most restaurants, cafes, and bars, with documentation and stale business data shaping which venues become visible.
Vladimir PiteninRead
LLMs can match embedding models on text tasks, but dedicated embedders usually deliver similar quality at a tiny fraction of the cost.
Adnan El Assadi, Niklas Muennighoff, Jinhyuk LeeRead
198
Continual Learning
Chain-of-Experience helps LLMs learn from repeated feedback during inference, improving performance and token efficiency across math, coding, and knowledge tasks.
Haoqin Tu, Yunhao Fang, Yizhong Wang, Cihang Xie, Shen YanRead
This paper proposes letting frozen AI models keep learning through evolving prompts, memories, tools, and routing rules while guarding against the loss of earlier skills.
Borui Kang, Jinrui Gu, Junhan Lv, Wenbin Li, Lei Wang, Yang GaoRead
HPSE helps language models turn newly injected passages into usable, multi-step knowledge rather than merely memorized text.
Tianci Liu, Zihan Dong, Tianchun Li, Yi-Chung Chen, Qiming Cao, Xingchen Wang, Shiyang Wang, Zichen Miao, Linjun Zhang, Haoyu Wang, Jing GaoRead
Load more videos
199
World Model
Flex-π gives robots a flexible world model that combines video, 3D structure, semantics, and actions to improve precise bimanual manipulation while adapting compute at inference time.
Ge Yan, Jinghao Liu, Yuzhi Fan, Lei Cai, Minwen Liao, Jesse Zhang, Dieter FoxRead
Evoke builds an interactive world model that remembers evolving scenes indefinitely while generating responsive video in just three steps.
Yuanyang Yin, Gongxuan Wang, Yifan Zhan, Chuanhao Li, Kaipeng Zhang, Feng ZhaoRead
An AI agent learns the rules and goals of unfamiliar games by building and debugging its own executable digital twin.
Alexy Skoutnev, Kirill Acharya, Gaston Longhitano, Madeleine Udell, Kevin Ellis, Iddo DroriRead
200
Reasoning
A powerful model can effectively teach a weaker one new problem-solving habits at test time by building a smart software harness around it.
Cheng Qian, Wenting Zhao, Liangwei Yang, Heng Wang, Jielin Qiu, Heng Ji, Silvio Savarese, Huan Wang, Shelby HeineckeRead
The paper finds that thinking models often amplify the appearance of careful reasoning more than the behaviors that truly signal correct answers.
Jean de Dieu Nyandwi, Leena Mathur, Yonatan Bisk, Robert Hawkins, Graham NeubigRead
MathForm helps language models turn informal mathematics into reliable Lean proofs by retrieving library knowledge and repeatedly refining outputs against formal verification feedback.
Lushi Pu, Weiming Zhang, Xinheng Xie, Zixuan Fu, Bingxiang He, Hengyu Zhao, Hongya Lyu, Xin Li, Jie Zhou, Yudong WangRead
201
Graph Learning
This paper improves the estimation of multiple sparse networks by letting information from one interaction layer sharpen the analysis of the others.
Youngseok Song, Sofia C. OlhedeRead
Quipu is a knowledge-graph store designed to ensure that agent-written facts are admitted, trusted, time-stamped, and auditable by construction.
Steve BrownRead
This paper gives graph neural networks a common architectural language for understanding, comparing, and designing many different layers.
Sai Karthik Navuluru, Siddhartha Shankar Das, Bo Ni, Hongjie Chen, Yu Wang, Baris Coskunuzer, Nesreen K. Ahmed, Franck Dernoncourt, Mahantesh Halappanavar, Tyler Derr, Ryan A. Rossi, Lakshman TamilRead
Load more videos
202
Scientific Ai
01
GENCO - A Unified Neural Solver Embedded in a Development Framework for Steady-State Grid Analysis
GENCO is a fast, unified neural solver and open framework for performing several demanding power-grid analyses while preserving physical feasibility.
Alban Puech, Matteo Mazzonelli, Tamara R. Govindasamy, Mangaliso Mngomezulu, Héctor Maeso-García, Thomas Tolhurst, Javad Bayazi, Ali Moeini, Naomi Simumba, Celia Cintas, David Nelischer, Romeo Kienzler, Jonas Weiss, Anna Varbella, Florian Dörfler, Gabriela Hug, Martin Mevissen, Juan Bernabé-Moreno, François Mirallès, Hendrik F. Hamann, Etienne Vos, Thomas BrunschwilerRead
Mechanist turns AI into an autonomous scientific instrument for discovering, testing, and controlling the mechanisms behind intelligent behavior.
Mengru Wang, Junfeng Fang, Shuofei Qiao, Zhenqian Xu, Haoming Xu, Haoxiong Wang, Shumin Deng, Linyi Yang, Zhixiang Cui, Xin Xu, Yunzhi Yao, Buqiang Xu, Fei Shen, Haozhe Luo, Yunxiang Wei, Ningyu Zhang, Julian McAuley, Tat Seng Chua, Huajun ChenRead
OmniScientist is an AI researcher that observes raw scientific data across many modalities and disciplines, then turns those observations into experiments and complete research papers.
Bobo Li, Hao Fei, Tianjie Ju, Mong-Li Lee, Wynne HsuRead
203
Code Generation
This paper shows that many AI-generated GPU kernels that pass standard tests are silently wrong, and proposes stricter contracts to catch those failures.
Rishi Shah, Rishav ShresthaRead
Nova3D turns 3D generation from making static meshes into writing editable, measurable, and animatable Blender programs.
Nimra Noor, Muhammad Bilal, Abdullah Hussain, Hassan BaigRead
Security tests can help LLMs repair vulnerable code, but their success depends heavily on test coverage, feedback design, and the model-task combination.
Yunhao Liang, Chengguang Gan, Ruixuan YingRead
204
Embodied Ai
G0.5 turns a vision-language model into a single-stream robot brain that explains what it is doing while directly generating the actions to do it.
Yicheng Liu, Zibin Dong, Baijun Ye, Tianyuan Yuan, Tao Jiang, Anqi Yang, Shicheng Cao, Haonan Liu, Yue Sun, Zihan Guo, Xiao Liu, Dong Ke, Changxun Pan, Chenru Wu, Tailai Cheng, Xiaoshu Ren, Xinlei Zhang, Jianning Cui, Zijie Zhao, Haoyu Zhang, Kaiming Xu, Haodong Yang, Bowen Zhang, Jiahui Niu, Shaoting Zhu, Shiduo Zhang, Hang ZhaoRead
HandEdit turns abundant human hand videos into training data for dexterous robots by benchmarking how well image-editing models can adapt them to different robotic hands.
Zhenjie Yang, Xingyu Jiao, Guopeng Zhong, Shuzhe Yang, Shi Che, Chao Wu, Chenyu Jiang, Dongjie Zhang, Yideng Zhang, Zheng Zhang, Muyun Jiang, Haisheng Su, Shuang Jin, Donghang Zhang, Chao Yang, Li Chen, Hongyang Li, Zuxuan Wu, Yu-Gang Jiang, Xiaosong Jia, Junchi YanRead
The work uses simulation-based control demonstrations to teach quadrupeds and humanoids complex locomotion-and-manipulation skills with sparse rewards and minimal manual tuning.
Martin Schuck, Maks Sorokin, Simone Manni, Duy Ta, Angela P. Schoellig, Marco Hutter, Simon Le Cleac'H, Jan BrüdigamRead
Load more videos
205
Robotics
FlexWorm helps soft, suction-powered robots navigate complex surfaces by combining geometric planning with learned motion snippets and fallback search.
Zili Tang, Tiecheng Guo, Qinyue Zhang, Meng GuoRead
Robo-Dopamine 2.0 gives robot policies a history-aware, OOD-sensitive reward signal to recover from mistakes and improve manipulation reliability.
Yijie Xu, Haopeng Jin, Run Zhou, Shengbang Liu, Sixiang Chen, Hongyang Cheng, Sicheng Hu, Peterson Co, Jinwen Luo, Huajie Tan, Shanghang ZhangRead
XPolicyLab aims to make evaluating and deploying robot policies as plug-and-play as using a common software standard instead of building a new integration for every policy and environment.
XPolicyLab Community, Tianxing Chen, Yue Chen, Tian Nian, Zijian Cai, Guangyu Chen, Wenwei Lin, Qiwei Liang, Peicheng Xiang, Kailun Su, Zixuan Li, Junyuan Tang, Yan Qin, Qiangyu Chen, Shaolong Zhu, Xiang Li, Jiahao Zhang, Weijie Wan, Baijun Chen, Honghao Su, Kehe Ye, Shujia Liu, Kaixuan Wang, Haotian Liang, Yunze Liu, Mingleyang Li, Yuran Wang, Boyu Chen, Hongzhe Bi, Shuhe Huang, Hengkai Tan, Jisong Cai, Yao Mu, Jun Guo, Xiaofeng Wang, Zheng Zhu, Weijie Ke, Hengtao Li, Yuhang Tang, Xiaofan Li, Ganlin Yang, Zhangzheng Tu, Shuai Yang, Wenxuan Song, Pengxiang Ding, Kaidong Zhang, Yu Sun, Junliang Guo, Tong Zhang, Yixing Chen, Rongxu Cui, Zongzheng Zhang, Haoxiang Ma, Junhao Cai, Haoyu Zhang, Senqiao Yang, Jinhui Ye, Pengguang Chen, Shu Liu, Xiu Su, Wenhan Fang, Wenhao Li, Yichao Cao, Chengyao Wang, Qiang Chen, Ping Luo, Wenbo DingRead
206
Agent
DarwinX improves AI agents by evolving their prompts, tools, and workflows through population-based selection rather than changing the underlying model.
Yifan Zhang, Yutong Dai, Juntao Tan, Luyu Yang, Rishi Mullur, Thai Hoang, Zhiyuan Hu, James Zhu, Phil Mui, Silvio Savarese, Ran Xu, Zeyuan ChenRead
This paper shows how to replace expensive LLM agents with cheap learned surrogates so large artificial societies can be simulated on an ordinary laptop.
Igor ItkinRead
SynthEx uses an agentic large language model to design and refine ambitious natural-product synthesis routes that experts found comparable to human plans.
Daniel Armstrong, Xuan-Vu Nguyen, Octavian Susanu, Gabriel Gibberd, Théo A. Neukomm, Taddäus Strunden, Dan Forster, Morgane Delattre, Shawn Teh, Clément Rols, John Federice, Hayden Leatherwood, M. Lavelle Barnes, Maarten R. Dobbelaere, Peter Wipf, Jon T. Njardarson, Jieping Zhu, Philippe SchwallerRead
207
Rl
In partially observed environments, reinforcement learning may fail not because the policy cannot represent the answer, but because the critic learns the wrong lesson.
Idil GözelRead
Training AI agents against just one simulated user can make them brittle, so this paper uses diverse simulated users to help agents generalize to real people.
Simon Yu, Nicholas Tomlin, Marwa Abdulhai, Ximing Lu, Derek Chong, Abe Hou, Dilara Soylu, Sergey Levine, Christopher D. Manning, Weiyan ShiRead
RLVR may make a model better at today’s task while quietly making tomorrow’s successful behaviors harder to discover.
Shaohang Wei, Zikun Su, Feifan Song, Wen Luo, Wei Li, Guangyue Peng, Houfeng WangRead
Load more videos
208
Speech
MiDashengLM-Gen uses an LLM and flow matching to generate coherent multilingual audio scenes with speech, music, and sound effects while sharply improving speech intelligibility.
Xingwei Sun, Heinrich Dinkel, Gang Li, Jiahao Mei, Yadong Niu, Zerui Han, Yuepeng Jiang, Jiahao Zhou, Lichun Fan, Jian LuanRead
This paper tests whether automated TTS judges can recognize the specific linguistic flaws that humans hear, revealing that today’s evaluators often reduce speech quality to acoustics or depend heavily on prompting.
Oluwanifemi Bamgbose, Simon Rosen, Jash Shah, Lindsay Devon Brin, Hoang H Nguyen, Anke Koelzer, Rachel Hansen, Tara Bogavelli, Fanny RiolsRead
Luna-TTS uses diffusion-style parallel speech generation to make expressive, multilingual text-to-speech faster, more controllable, and capable of voice cloning and editing.
Feng Yin, Shuai Shi, Junjie Zheng, Kechenying Zhou, Yiqiu Wang, Chenyang He, Qiuhua Jiang, Mengxiao Bi, Yanmin Qian, Mingxin Chen, Xun Gong, Tianteng Gu, Bing Han, Peng Jiang, Chenda Li, Haiyang Sun, Han Wang, Wei Wang, Yi Wang, Leying Zhang, Wangyou Zhang, Chushu ZhouRead
209
Multimodal
COVER turns video moment predictions into calibrated time regions that say not only where an event is, but how reliably it is covered.
Aseel Mohamed, Rasul Khanbayov, Erchin Serpedin, Hasan KurbanRead
A Gemini-based multimodal medical AI reportedly matches or exceeds physicians on several simulated consultation tasks while remaining weaker in rapport, subtle affect, and fine-grained physical assessment.
Mahvish Nagda, Jihyeon Lee, Matthew Thompson, Chunjong Park, Tim Strother, Valentin Liévin, Roma Ruparel, Akshay Goel, Teya Bergamaschi, Suhana Bedi, Meet Shah, Pavel Dubov, Liviu Panait, Toshiyuki Fukuzawa, Sam Schmidgall, Craig Schiff, Joseph Xu, Aliya Rysbek, Yana Lunts, Jan Freyberg, Rebecca Hemengway, Sunny Virmani, David Racz, Carey Radebaugh, Joëlle Barral, Kavi Goel, Dale R. Webster, Katherine Chou, Avinatan Hassidim, Yossi Matias, James Manyika, Gregory Wayne, Tao Tu, Yun Liu, Ethan Goh, Christina Chen, Ryutaro Tanno, Po-Hsuan Cameron Chen, Mike Schaekermann, Anil PalepuRead
A multimodal assistant learns to suggest visually consistent next edits, substantially improving user engagement in large-scale image-creation conversations.
Zhijing Zhang, Jinpeng Yu, Xin Song, Bingnan Li, Chuyue Li, Changhui Du, Xiaolin Fang, Jiaming Liu, Ruihua HuangRead
210
Cv
Self-Geometry improves 3D vision foundation models at test time by using camera and multi-view geometry as self-supervision without requiring ground-truth labels.
Seokhyun Youn, Dahyeon Kye, Sung-Ho Bae, Jihyong OhRead
CANIS uses generative 3D priors and semantic image cues to align arbitrarily oriented objects into a meaningful canonical pose.
Kendong Liu, Yuxin Yao, Junhui HouRead
A physics-aware data pipeline and fast video diffusion model remove distracting glass reflections while introducing a benchmark for measuring progress.
Zepeng Wang, Jiagao Hu, Fuhao Li, Yuxuan Chen, Fei Wang, Daiguo ZhouRead
Load more videos
211
Nlp
ReMEMBER helps dialogue summarizers retrieve the missing past context needed to understand and summarize the present conversation.
Hyangsuk Min, Hwanjun SongRead
A large-scale study estimates that most biomedical papers now contain signs of LLM-assisted writing, especially in their Discussion sections.
Lena Holzwarth, Rita González-Márquez, Dmitry KobakRead
STINER turns noisy X posts into structured cyber-threat intelligence, showing that domain-specific models can extract strategic signals faster and more accurately than general-purpose or LLM-based systems.
Yasir Ech-Chammakhy, Oussama Azrara, Jaafar Chbili, Anas MotiiRead
212
Efficiency
One nested language model can serve as several differently sized models, cutting training costs while speeding up speculative decoding.
Nathan Godey, Yoav ArtziRead
Hard prompt compression can delete the crucial context that makes surviving answers understandable, but a lightweight restoration step can recover much of the lost accuracy.
Zhengpei Hu, Kai Li, Dapeng Fu, Xuechao Zou, Yuanhao Tang, Yue Li, Tengfei Cao, Jianqiang HuangRead
Small models can predict large-model behavior after all—but only when their hyperparameters are tuned carefully enough to reveal the hidden scaling laws.
Nicholas Lourie, Kyunghyun Cho, Karen Ullrich, Sanae LotfiRead
213
Generative Model
Latent-to-4D turns the hidden representations of video generators directly into stable dynamic 3D worlds without passing through RGB.
Zihao Liu, Xiaolong Shen, Zhenglin Zhou, Ruijie Quan, Yi YangRead
Atelier aims to make artist-inspired image generation more faithful by planning around artistic intent while avoiding the visual shortcuts that models commonly associate with famous artists.
Kuan Xing, Ye Wang, Changyi Gan, Yuheng Li, Thao Nguyen, Yi Chang, Yilin WangRead
V-RAE makes video generation more efficient and semantically meaningful by compressing videos into latents derived from pretrained vision models rather than pixel-focused autoencoders.
Minghui Guo, Shengqiong Wu, Hao FeiRead
Load more videos
214
Self Supervised
USD teaches language models to choose both what supervision to absorb and how difficult that supervision should be, adapting self-distillation to the model’s changing learning capacity.
Yongkang Yang, Zhezheng Hao, Hong Zhang, Yi Liu, Xiankun Lin, Wence Ji, Fanjunduo Wei, Jiarui Yu, Qiang Lin, Xiaoyun Liang, Hande DongRead
U-OPSD lets an LLM improve its mathematical reasoning by learning from the agreement and mistakes in its own sampled solutions, without external labels or teacher models.
Yijiang Li, Bingyang Wang, Yijun Liang, Yunjie Tian, Di Fu, Nuno VasconcelosRead
CVPD helps multimodal models notice visual details they can perceive but previously failed to use, without relying on external teachers or annotations.
Shravan Venkatraman, Omkar Thawakar, Ritesh Thawkar, Abdelrahman Shaker, Rao Muhammad AnwerRead
215
Optimizer
AutoOPT combines numerical search, LLM reasoning, and formal proof assistants to automatically discover and verify new optimization algorithms.
Heechang Kim, Ernest K. Ryu, Shuvomoy Das GuptaRead
The study shows that Muon can make transformers grok quickly but later destabilize their representation-to-output interface, causing learned solutions to collapse.
Ali Janati, Kaoutar El Maghraoui, Andrei Kanavalau, Anass BelfatmiRead
The paper argues that Adam’s coordinate-wise behavior breaks a hidden symmetry that lets gradient descent favor low-rank solutions, showing that optimizer geometry—not just the loss—determines which answer a model learns.
Devender SinghRead
216
Attention
This paper argues that self-attention can spontaneously organize token representations into stable clusters once attention becomes sufficiently sharp.
Qucheng Gao, Zuyi Yang, Xiao ChenRead
This work argues that a power-law, graph-based form of attention can mathematically subsume standard attention and may effectively collapse to it during inference.
Burc GokdenRead
This paper makes spiking transformers more spatially aware by combining localized attention with a lightweight pathway that preserves communication across region boundaries.
Zeqi Zheng, Zizheng Zhu, Yuping Yan, Wenxuan Pan, Zhaofei Yu, Yaochu JinRead
Load more videos
217
Transformer
Small Transformer design choices can quietly make or break a model’s ability to handle very long contexts.
Amanda Bertsch, Luca Soldaini, Matthew R. Gormley, Graham Neubig, Hannaneh Hajishirzi, Kyle Lo, Dirk GroeneveldRead
A transformer that feeds its hidden thoughts back into the next decoding step appears to gain better performance and shorter reasoning with little extra cost.
Xi Wang, Ziyang Cai, Zheng Zhan, Harry Dong, Ying Fan, Gustavo de Rosa, Tim Pearce, John LangfordRead
This study shows that unusual activation spikes in hybrid-attention language models follow predictable patterns tied to where full attention occurs and how activations are canceled.
Zunhai Su, Bohan Sun, Xialie Zhuang, Shuibai Zhang, He Xiao, Jing Xiong, Hengyuan Zhang, Zhongzhu Zhou, Tiantian Zhang, Ngai Wong, Chuan-Wei KuoRead
218
Llm
CreativeInstruct teaches LLMs to produce more diverse and creative outputs without sacrificing quality, and shows that this creativity can also improve downstream reinforcement learning.
Ananya Sahu, Mohit Bansal, Elias Stengel-EskinRead
LLMRouter is an open platform and benchmark for choosing the best language model for each query while balancing answer quality, personalization, and inference cost.
Tao Feng, Fangxu Yu, Haozhen Zhang, Zhongjie Dai, Liangqi Yuan, Zijie Lei, Weizhi Zhang, Kunlun Zhu, Haodong Yue, Keyang Xuan, Ge Liu, Jiaxuan YouRead
MetaStrategy uses a compact LLM to generate safe, executable ranking policies that improve real-world recommendation outcomes without adding online latency.
Chengyu Lai, Jiuning Lin, Zhibo Xiao, Xiaodong Zhu, Ruiquan Lan, Bin Zhang, Zihong Huang, Wendong Zhang, Chuxin Chen, Yinjiang Cai, Shuai Zhong, Lingqing Zhang, Dimin Wang, Jialin Zhu, Han ZhuRead
219
Hardware
A thin Al2O3 coating makes silicon T-center photons more spectrally stable, bringing indistinguishable telecom quantum emitters closer to practical devices.
Fariba Islam, Chang-Min Lee, Kyu-Young Kim, Sorah Fischer, Purbita Purkayastha, Edo WaksRead
GALA designs compact quantum error-correcting codes that are built from the outset to fit the movement constraints and operations of neutral-atom quantum hardware.
Willers Yang, Casey Duckering, Arpit DuaRead
Researchers show that an unprivileged CPU process can spy on GPU workloads running on Apple Silicon by tracking footprints in the shared cache.
Tianhong Xu, Saion K. Roy, Ruyi Ding, Aidong Adam Ding, Yunsi FeiRead
Load more videos
220
Safety
An attacker can make dangerous behavior look honest by rewriting only the reasoning, exposing a major weakness in chain-of-thought safety monitoring.
Shikhar Shiromani, Leo RichterRead
Citation-backed medical LLMs can substantially improve physician accuracy, but they may also make doctors less likely to challenge confidently supported wrong answers.
Tirtha Chanda, Christoph Wies, Franziska Schramm, Carina Nogueira Garcia, Nicolas B. Merl, Martin J. Hetz, Jochen S. Utikal, Phillip Tschandl, Cristian Navarrete-Dechent, Alexander Thiem, Jakob N. Kather, Consortium, Titus J. BrinkerRead
A team of specialized AI investigators examines video texture, lighting, motion, and physics to expose deepfakes and explain its verdict.
Xuechao Zou, Shun Zhang, Kai Li, Yi Zhou, Xinyu Sun, Yuhui Chen, Zhe Wu, Congyan Lang, Junliang XingRead
221
Benchmark
Business Arena tests whether LLM agents can actually run a profitable cross-border business and finds that even leading models still trail human-designed strategies.
Yijun Pan, Yukun Lian, Kunyu Shi, Junbo Li, Hongwei Xue, Sicong Xie, Guannan Zhang, Xiaoying XingRead
02
DSAgentBench: Can Agents Automate End-to-End Data-Science Workflows in Real Computer Environments?
DSAgentBench tests whether AI agents can genuinely complete end-to-end data-science projects on real computers, revealing that current systems still struggle substantially.
Mizanur Rahman, Mohammed Saidul Islam, Ridwan Mahbub, Md Tahmid Rahman Laskar, Shafiq Joty, Enamul Hoque PrinceRead
A careful audit finds that flawed grading, rather than weak models, made language models look far worse at scientific coding than they really are.
Sihan Hu, Lyuhan Huang, Youjin Deng, Kun ChenRead
222
Continual Learning
ContinualSkillBench tests whether LLM agents truly learn reusable skills over time, finding that they often adapt to context without reliably consolidating transferable abilities.
Tianyi Guan, Yiding Wang, Haotong Yang, Siyuan Cao, Shirui Liu, Yi Hu, Jiaqi Li, Muhan ZhangRead
This work keeps recommender systems up to date by separating reusable behavioral knowledge from task-specific adaptation, delivering strong benchmark and real-world business gains.
Zixuan Wang, Yuhong Chen, Yuxuan Zhu, Guidong Lei, Zhiluohan Guo, Yu Zhao, Kun Wang, Bangyang Hong, Kangle Wu, Yabo Ni, Anxiang Zeng, Cong Fu, Hui LiRead
Omega-S is a cheap weight-only regularizer that helps LLMs remember old skills while learning new ones, though it still needs broader testing.
Alberto AcedoRead
Load more videos
223
World Model
Sekai2 is a large-scale video dataset designed to teach AI systems how worlds evolve across time, viewpoints, and camera trajectories.
Kang He, Wenshuo Peng, Zihui Gao, Jiaming Tan, Kaipeng Zhang, Yongtao GeRead
WorldExam tests whether video models merely look convincing or can actually understand scenes well enough to react plausibly to what happens in them.
Yuxue Yang, Shuyao Shang, Jiahe Wang, Zitong Zhou, Liang Tan, Junhan Zeng, Ruizhi Li, Junyan Li, Yu Liu, Xiao Yang, Yong Li, Jun Zhu, Hongsheng Li, Tieniu Tan, Lue Fan, Zhaoxiang ZhangRead
WorldSimProbe checks whether embodied world models truly respond to actions like physical simulators, rather than merely generating plausible-looking videos.
Peterson Co, Sicheng Hu, Chunxuan Jiao, Hongyang Cheng, Yulin Luo, Yijie Xu, Sixiang Chen, Zhongxia Zhao, Zihao Wang, DaFeng Chi, Peidong Liu, YuTong Chen, Henghua Liu, Zhihao Yuan, Huizhu Jia, Yuzheng Zhuang, Tianle Zhang, Liang Lin, Huajie Tan, Shanghang ZhangRead
224
Reasoning
ReflectRL turns expert mistakes into useful training signals by teaching language models to reflect on flawed solutions before learning to solve problems directly.
Jinhe Bi, Chennan Zhou, Zengjie Jin, Aniri, Shuo Lu, Wenke Huang, Hu Cao, Xun Xiao, Zhihong Zhu, Volker Tresp, Fei Shen, Yunpu Ma, Tat-Seng ChuaRead
This paper explains how to fairly compare reasoning LLMs that spend more computation at inference time and provides practical standards for evaluating and reproducing them.
Mohsen Hariri, Weicong Chen, Nahal Shahini, Vikash Singh, Kai Ye, Amirhossein Samandar, Debargha Ganguly, Sreehari Sankar, Yanyan Zhang, Shouren Wang, Jerry Peng, Biyao Zhang, Michael Hinczewski, Vipin ChaudharyRead
Instead of trusting the most common LLM answer, CALVER checks whether each causal explanation actually obeys the rules of causal graphs.
Omatharv Bharat Vaidya, Connor Thomas Jerzak, Zayne Rea Sprague, Fangcong Yin, Nhat HoRead
225
Graph Learning
GLM-RAG shows that graph language models can make knowledge-graph retrieval more transferable for multi-hop question answering than conventional alternatives.
Maya Arseven, Anette Frank, Beni Egressy, Johann Higl, Moritz PlenzRead
The paper proves that an apparently powerful unrolled graph denoiser can only realize a narrow class of polynomial graph filters, limiting its expressive power and practical accuracy.
Seyed Alireza HosseiniRead
This work shows why graph-based time-series models break when correlations shift over time and proposes GLIDE to make them more robust.
Chen Shao, Yue Wang, Zhenyi Zhu, Zhanbo Huang, Tobias Käfer, Zonghan Wu, Danai KoutraRead
Load more videos
226
Scientific Ai
Fisher-R1 trains LLM agents to choose statistically valid tests and draw more reliable scientific conclusions from real-world datasets.
Jiacheng Miao, Jin Mu, Guanhua Chen, James ZouRead
Researchers use precision error-mitigated quantum hardware to observe long-lived Floquet oscillations in a regime where classical simulations struggle.
Eyal Leviatan, Tasneem Watad, Roy Perry, Lukas Broers, Mohammed Zuhair Mullath, Ori Alberton, Itai Arad, Yosi Atia, Eyal Bairey, Shaul Barkan, Matan Ben Dov, Asaf Berkovitch, Ewout van den Berg, Itsik Cohen, Omri Golan, Ilya Gurwich, Avieli Haber, Barak A. Katzir, Oded Kenneth, Roei Levi, Yotam Y. Lifshitz, Yaron Lukovsky, Ron Melcer, Adiel Meyer, Boris Muratov, Aviad Panahi, Gili Schul, Tali Shnaider, Maor Shutman, Alireza Seif, Tomonori Shirakawa, Asif Sinay, Vincent P. Su, Hayk Tepanyan, Omri Trebitch, Assaf Zubida, Dorit Aharonov, Hrant Gharibyan, Abhinav Kandala, Seiji Yunoki, Netanel H. LindnerRead
SEE tests whether multimodal AI can move beyond explaining science to making reliable, evidence-grounded inferences from real experimental data.
Taolin Han, Yuchen Zhang, Jinghang Wang, Yun Wu, Wai Yuet Chiu, Zhaohai Li, Yifei Zhang, Jinxin Wang, Yuhao Zhou, Chen Zhao, Jiajia Li, Jiaxin Li, Qile Jin, Kewei Sun, Shuang Wu, Weiqi Zhai, Renquan Lv, Junchao Li, Ruodan Chen, Qingteng Chen, Zhibo Yang, Hu Wei, Lin Qu, Shuai Bai, Bing ZhaoRead
227
Code Generation
An agentic system turns advanced mathematical research into verified Lean proofs, even when the required concepts are missing from existing libraries.
Arshia Soltani Moakhar, Iman Gholami, Max Springer, Mahdi JafariRaviz, MohammadTaghi HajiaghayiRead
A massive real-world study finds that AI-written C++ can increase maintenance and compute costs, but targeted feedback helps make the code more efficient.
Michael Tran, Fred Lewis, Kun Yang, Saksham Thakur, Aditya Kini, Aditya Patil, Milad Hashemi, Parthasarathy RanganathanRead
FailForge teaches code agents to learn from their failures, turning past mistakes into reusable skills that substantially improve software-engineering task performance.
Dongyi Lv, Fushun E, Aichen Cai, Liang Huang, Ya Zhang, Qiuyu Ding, Canhui Wu, Zhi Wang, Yuesong Zhang, Jiaqi Wang, Nan DuanRead
228
Embodied Ai
ACE-Data-0 turns everyday homes into multisensory recording studios for teaching embodied AI how humans see, move, touch, and complete long-horizon tasks.
Yukang Cao, Haozhe Xie, Beichen Wen, Runmao Yao, Yinghao Liu, Yue Huang, Zhichao Liao, Yunxiang Wang, Haiheng Liu, Xingshun Tian, Dawei Su, Long Zhuo, Dacheng Tao, Xiaogang Wang, Liang Pan, Ziwei LiuRead
Pictura trains autonomous driving agents through massive visual self-play, teaching them to handle realistic camera views without relying on privileged state information.
Yuan Yin, Elias Ramzi, Marc Lafon, Valentin Charraut, Victor Bares, Yihong Xu, Éloi Zablocki, Alexandre Boulch, Thibault Buhet, Andrei Bursuc, Matthieu CordRead
The study shows that many vision-language controllers succeed without truly seeing, while selective hazard assistance plus deterministic perception offers a safer path to grounded control.
J. de Curtò, Dayani Plasencia, Diego Sánchez, I. de ZarzàRead
Load more videos
229
Robotics
A reinforcement-learning controller helps a large quadruped keep walking after actuator failures by learning when to change its gait timing.
Giovanbattista Gravina, Luca Rossini, Carlo Rizzardo, Arturo Laurenzi, Nikos TsagarakisRead
Push-Wiper teaches robots to gather rather than spread messy stains, enabling more general cleaning across materials, residue types, and curved surfaces.
Renhao Lu, Mingxin Wang, Chenyang Cao, Yang Yang, Guoping Pan, Kangkang Dong, Yi Cheng, Houde LiuRead
Temporal Policy makes diffusion-like robot control faster by starting action generation from the robot’s recent history instead of random noise.
Dylan Miller, Martin JagersandRead
230
Agent
This paper turns the practical problem of squeezing an AI agent’s memory into a limited context window into a formal communication-complexity theory.
Hayder Tirmazi, Sam Markelon, Allison Bishop, Michael MitzenmacherRead
Harness-R1 trains an engineer agent to patch an LLM agent’s runtime harness from its failures, improving task success across web, household, and database environments.
Shuai Shao, Kangning Zhang, Qingyao Li, Shijian Wang, Hao Wang, Wenxiang Jiao, Yuan Lu, Yi Guo, Weiwen Liu, Weinan ZhangRead
Qwen-CUA is a large multimodal agent that learns to control computers directly from screenshots and mouse-keyboard actions, achieving strong performance on challenging real-world software tasks.
Dunjie Lu, Shuai Bai, Tianyi Bai, Sicheng Fan, Chang Gao, Jian Guan, Feng Hu, Mianqiu Huang, Xingyang Huang, Yizhen Jiang, Yuheng Jing, Dehui Kong, Ning Li, Dayiheng Liu, Shixuan Liu, Zheng Liu, Que Shen, Bowen Wang, Junli Wang, Chencan Wu, Rui Xie, Tianbao Xie, Zhihui Xie, Haiyang Xu, An Yang, Tao Yu, Wenzhen Yuan, Xi Zhang, Zhenru Zhang, Mingkang Zhu, Zhaoqing Zhu, Yizhong Cao, Kai Dang, Binyuan Hui, Kaixin Li, Junyang Lin, Haiquan Wang, Zekun Wang, Yiheng Xu, Fan Yan, Mengqi Yuan, Danyang Zhang, Jiajun Zhang, Zhipeng Zhang, Fan Zhou, Fan ZhouRead
231
Rl
CRPO combines contrastive learning with reinforcement-based self-distillation to make agentic LLM training more stable, exploratory, and generalizable.
Xingjian Wu, Junlin Liu, Xingchen Liu, Xuhang Zhu, Jianing Wang, Linsen Guo, Xiaoyu Li, Xuezhi Cao, Xunliang CaiRead
The paper argues that RL should give less credit to highly counterfactually sensitive tokens and shows that this improves long-chain-of-thought reasoning.
Qiangqiang He, Zhongheng Wu, ZiJian WangRead
03
When Does On-Policy Interaction Help? Representational Tradeoffs in Value-Based Imitation Learning
This work shows that interacting with an expert can let a less expressive learner succeed by representing the expert’s values rather than exactly copying its policy.
Luca Viano, Antoine Moulin, Audrey Huang, Volkan Cevher, Philip Amortila, Dylan J. FosterRead
Load more videos
232
Speech
JoyAI-Talker is a full-duplex empathetic voice agent that combines language reasoning, expressive speech generation, and real-time interruption handling.
Yinhao Bai, Jinming Chen, Yafeng Chen, Wei Deng, Boya Dong, Nan Duan, Yu Gu, Weisheng Han, Yankun Huang, Ming Ke, Hao Li, Jingdong Li, Xiangyu Liang, Ning Liu, Yuan Liu, Ji Miao, Jiaqi Wang, Qi Wang, Wenchao Wang, Yuxuan Wang, Zhenfang Wang, Zhangyu Xiao, Chao Xue, Hongfei Xue, Fan Yu, Tianyi Zhang, Yuan Zhang, Yuqi Zhang, Lin ZhuRead
SwanTale aims to generate expressive multi-speaker dialogue and rich audio scenes from either natural-language instructions or a short voice reference.
Yu Zhang, Ruiqi Li, Changhao Pan, Ke Lei, Xiang Yin, Cheng YangRead
03
ThinkOmni: A Reasoning-Driven Omni-Modal LLM Framework for Audio Forgery Detection and Localization
ThinkOmni teaches an audio-capable language model to explain why a recording is fake and pinpoint exactly when the manipulation occurs.
Yuxiong Xu, Kaiqing Lin, Bin Li, Haodong Li, Sheng LiRead
233
Multimodal
OmniPack makes omni-modal LLMs far more efficient by compressing redundant audio-visual tokens while preserving most of their task-relevant information.
Wanshun Su, Yang Shi, Feihu Liu, Ziwen Yu, Yan Min, Zhuoran Zhang, Qixun Wang, Haotian Wang, Shixuan Liu, Yuanxing Zhang, Peng Wu, Chengfu Huo, Liang DingRead
This paper shows that multimodal models may appear to think with images while often ignoring the visual evidence their tools retrieve.
Zhiheng Wang, Bo Peng, Lai Wei, Chaochao LuRead
This study maps how vision and language learn from each other and shows that unifying them early can make multimodal foundation models both stronger and more compute-efficient.
Junlin Han, Shengbang Tong, David Fan, Minghao Chen, Philip Torr, Filippos Kokkinos, Mike LewisRead
234
Cv
EOVSAM turns SAM 3's slow exhaustive vocabulary search into a fast single-pass open-vocabulary segmenter while preserving competitive accuracy.
Haomin Peng, Yongkang Li, Zhaoxiang Liu, Xiaojie Jin, Shiguo Lian, Yunchao Wei, Xinggang WangRead
It shows that AI-image detectors may learn file-compression fingerprints instead of synthesis—and that attractive heatmaps are not automatically faithful explanations.
Leonid Kuturin, Ilya Sotnikov, Mark Khusnutdinov, Mikhail Potemkin, Pavel Baranas, Aleksandra Korepanova, Alexander KalashnikovRead
Vision models may recognize not just what is in an image, but also clues about which camera or processing pipeline produced it.
Vladan Stojnić, Ryan Ramos, Giorgos Kordopatis-Zilos, Noa Garcia, Giorgos ToliasRead
Load more videos
235
Efficiency
ResKV makes long-context language-model decoding more efficient by compressing KV caches while reconstructing the attention contributions of discarded tokens.
Yuhang Zhan, Lisi Chen, Shuo ShangRead
RestoreKV makes aggressively compressed LLM KV caches behave more like full caches by adding a compact, context-specific learned restoration cache.
Changwoo Baek, Seungjun Shin, Kyeongbo KongRead
TokTier makes agentic LLM serving faster by reusing and repairing tokenization state exactly instead of repeatedly tokenizing long conversation transcripts from scratch.
Zhenyu Zhang, Zhichao CaoRead
236
Diffusion
DiffusionGemma turns a large language model into a fast parallel text generator, reaching roughly 1,500 tokens per second while preserving much of the model's reasoning and multimodal capability.
DiffusionGemma Team, Adrien Ali Taïga, James Assiene, Daniele Calandriello, Rahma Chaabouni, João Gante, Tamara von Glehn, Nate Keating, Chris Knutsen, Martin Kukla, Tianlin Liu, Ivan Lobov, Ofir Nabati, João Gabriel Oliveira, Nicolas Perez-Nieves, Nastasia Prutianova, Bobak Shahriari, Jean Tarbouriech, Pavel Tyletski, Çağlar Ünlü, Cindy Wu, Glenn Cameron, Jerome Connor, Sertan Girgin, Maarten Grootendorst, Alon Levkovitch, Eliya Nachmani, Omar Sanseviero, Piotr Stanczyk, Quentin Berthet, Andrew Campbell, Clément Crepy, Valentin De Bortoli, Arnaud Doucet, Romuald Elie, Alexandre Galashov, Klaus Greff, Alexis Jacq, David Ruhe, Yu-Han Wu, Sebastian Flennerhag, Brendan O'Donoghue, George Scrivener, Shantanu ThakoorRead
JoyAI-Video-Edit aims to make high-quality, open-ended video editing run in real time by combining autoregressive generation with diffusion distillation.
Yicheng Xiao, Wenxun Dai, Xinran Qin, Lin Song, Maoquan Zhang, Hang Xu, Yukang Chen, Yitong Li, Guohui Zhang, Yuan Zhang, Xuying Zhang, Tommy Zhang, Jianlong Yuan, Peihao Li, Shuai Lu, Siming Fu, Chuyang Zhao, Xin Han, Jie Huang, Wenbo Li, Guoqing Ma, Wei Huang, Xiaojuan Qi, Haoyang Huang, Nan DuanRead
This work shows that better-structured language can make visual diffusion models more capable, and uses scaling laws to build a stronger prompting system.
Zilong Chen, Chaorui Deng, Kunchang Li, Hongyi Yuan, Haoqi FanRead
237
Generative Model
A single-pass generative model synthesizes realistic, time-resolved breast MRI contrast without gadolinium and shows promising robustness and clinical utility.
Smriti Joshi, Apostolia Tsirikoglou, Daniel M. Lang, Richard Osuala, Noah Márquez Varaa, Alejandro Guzman, Grzegorz Skorupko, Sebastian Ibarra Arregui, Lidia Garrucho, Akane Ohashi, Dimitra Ntoula, Eugen Divjak, Oğuz Lafcı, Jan C. Peeken, Julia A. Schnabel, Fredrik Strand, Oliver Diaz, Karim LekadirRead
FlexComposer makes it easier to place still or moving objects into videos while preserving realistic motion, appearance, and user-defined trajectories.
Songchun Zhang, Sitong Guo, Xianghao Kong, Pengwei Liu, Yuwei Guo, Lvmin Zhang, Anyi RaoRead
Visko Orbis aims to make long, high-quality video generation interactive in real time, letting users change the story or visual direction while the video is still being produced.
Xiangbo Gao, Siyuan Yang, Ping He, Mingyang Wu, Yuheng Wu, Yushen Zuo, Jiongze Yu, Ryan Cui, Hongyuan Hua, Devin Ma, Xiao Jin, Yubo Yuan, Qing Yin, Jie Yang, Zhengzhong TuRead
Load more videos
238
Optimizer
CMuon makes diffusion-transformer training faster and more stable by orthogonalizing fused weight components independently instead of coupling them together.
Chuyan Chen, Peng Sun, Kun YuanRead
This work combines spectral-norm sharpness control with the Muon optimizer to improve ImageNet training performance on both ViTs and ResNets.
Wenzhi Zhong, Edward Milsom, Michael MurrayRead
This paper shows that compressing Muon to signs can break its theory, yet the seemingly unsafe version often works best in real-world training.
Maria Smirnova, Alexey KravatskiyRead
239
Transformer
MHAR lets different feature groups retrieve different layers of a transformer's history, improving language-model quality with little added cost.
Cheng Luo, Zefan Cai, Junjie HuRead
UniR² turns recommendation recall and ranking into two coordinated tasks within one decoder-only Transformer, aiming to improve consistency and efficiency in production systems.
Ruochen Yang, Shuang Wen, Pengbo Xu, Yusheng Huang, Jiangxia Cao, Shuang Yang, Zhaojie Liu, Jiawei Sheng, Tingwen LiuRead
This paper shows that ALiBi can make attention heads numerically blind at long distances and tests ways to restore reliable token retrieval.
Christopher Schröder, Lukas Gienapp, Ferdinand Schlatt, Martin Potthast, Gerhard HeyerRead
240
Llm
Language models often agree with one another far more than they agree with real readers, suggesting that model consensus should not be mistaken for human consensus.
Kazuki Nakayashiki, Keisuke WatanabeRead
This survey explains how modern language models remember, update, and retrieve information, and organizes the field’s many competing memory designs into one coherent framework.
Sining Zhoubian, Dan Zhang, Evgeny Kharlamov, Jie TangRead
LLMs can reason impressively over many modalities, but this study finds that their tabular prediction ability collapses as the number of input dimensions grows.
Marta Garnelo, Wojciech M. CzarneckiRead
Load more videos
241
Foundation Model
K-EXAONE 2.0 is a 750B-parameter open-weight multilingual MoE model designed for long-context reasoning, coding agents, and culturally grounded safety.
Eunbi Choi, Kibong Choi, Sehyun Chun, Seokhee Hong, Junwon Hwang, Hyojin Jeon, Ahra Jo, Hyunjik Jo, Yeonsik Jo, Minhyeok Jung, Doyoung Kim, Heegyu Kim, Joonkee Kim, Seonghwan Kim, Soyeon Kim, Sunkyoung Kim, Yireun Kim, Yongil Kim, Byungoh Ko, Changhun Lee, Dohaeng Lee, Haeju Lee, Jinsik Lee, Kyungmin Lee, Minwoo Lee, Wonkee Lee, Sangha Park, Sungjune Park, Kwangrok Ryoo, Kijung Seo, Minju Seo, Yongwoo Song, Sejong Yang, Heuiyeen Yeen, Stanley Jungkyu Choi, Yemuk Choi, Yongchan Chun, Jiwon Ham, Dasol Hong, Sujeong Im, Kijeong Jeon, Gerrard Jeongwon Jo, Hyeongjun Jo, Yujin Jo, Jiyeon Jung, Naeun Kang, Daeseong Kim, Euisoon Kim, Hayeon Kim, Hyosang Kim, Myoungshin Kim, Unsol Kim, Youchul Kim, Chaeeun Lee, ChaeYoon Lee, Edward Hwayoung Lee, Honglak Lee, Hwansoo Lee, Minkyung Lee, Sangeun Lee, Solji Lim, Woohyung Lim, Chanwoo Moon, Jueun Mun, Jimin Park, Seojeong Park, Yongmin Park, Hyerin Seo, Donghyeon Shin, Donghyun Son, Eunyong Son, Kaehyun Um, Sihoon Yang, Chang En Yea, Sihyuk Yi, K...Read
Rem3Di turns atomistic foundation-model features into transferable 3D molecular fingerprints that preserve molecular shape and handedness.
Steffen Wedig, Felix Burton, Rokas Elijošius, Christoph Schran, Lars L. SchaafRead
03
Sparse Autoencoders Encode Both Concepts and Functions: The Downstream Geometry of Feature Effects
This study finds that sparse autoencoder features can matter causally without acting like simple, reusable steering directions, especially when they function as context-sensitive pointers.
Phu Gia Hoang, Anwoy Chatterjee, Tanmoy Chakraborty, Iryna Gurevych, Subhabrata DuttaRead
242
Hardware
CircuitProver uses an AI agent and reusable Lean proof libraries to make formal hardware verification faster and more transferable across related circuit designs.
Ziyi Yang, Wenji Fang, Chen Chen, Zhiyao Xie, Hongce ZhangRead
NUNA speeds up multi-GPU inference by routing communication and placing computation with awareness of where traffic travels inside large GPU systems.
Conor James Green, William Won, Tuan Ta, Bradford M. BeckmannRead
Shiftfly replaces a costly all-to-all optical tier with a compact shift-routed topology that scales accelerator clusters to hundreds of thousands of chips, while exposing tradeoffs in locality and slice allocation.
Eylon E. KrauseRead
243
Benchmark
Lethe tests whether federated medical-AI models can truly forget hospitals, patients, or classes—and finds that difficult forgetting requests, rather than the choice of method, reveal the real weaknesses.
Shengchao Chen, Ting ShuRead
OSReward tests whether vision-language models can reliably judge computer-use agents and releases cheaper open reward models to make that evaluation scalable.
Qiushi Sun, Kanzhi Cheng, Yian Wang, Bowen Yang, Hang Yan, Liheng Chen, Fangzhi Xu, Zichen Ding, Nuo Chen, Jialin Cao, Xingdong Gong, Zehao Li, Kaiming Jin, Xinfeng Yuan, Zhoumianze Liu, Jingyang Gong, Zhangyue Yin, Jiahui Gao, Zhiyong Wu, Tianbao Xie, Jianbing Zhang, Ben Kao, Lingpeng KongRead
Many frontier LLMs may get hard science questions right for the wrong reasons, exposing serious flaws in answer-only reasoning benchmarks.
Xuan Ren, Weiqi Zhai, Tianle Pu, Yihua Zhu, Yihua Zhu, Hu Wei, Bing ZhaoRead
Load more videos
244
Continual Learning
Instead of making the model smarter, this system remembers verified answers so it can return them instantly, exactly, and forever.
Sietse SchelpeRead
MemSFT gives LLMs specialized domain expertise through an external memory while preserving their general abilities.
Jiarui Wang, Xiang Shi, Jiaqi Cao, Rubin Wei, Xiquan Wang, Hao Sun, Jingzhi Wang, Zhiqi Yang, Qipeng Guo, Bowen Zhou, Zhouhan LinRead
Metis gives foundation models an internal, continually updating memory that can store and recall experience without changing the model’s weights.
Zeyu Zhang, Ziliang Guo, Yihang Sun, Xichong Zhang, Xixuan Hao, Zehao Lin, Yang Zhang, Xiaoyan Zhao, Tong Shen, Bo Tang, Zhi-Qin John Xu, Junchi Yan, Haofen Wang, Xu Chen, Feiyu Xiong, Zhiyu Li, Tat-Seng ChuaRead
245
World Model
This paper argues that truly useful world models must simulate not only what happens in a scene, but also what the people in it believe, want, and intend.
Hao Fei, Yiran ZhaoRead
WCM improves robot-learning critics by predicting how the world evolves over time instead of relying only on single-frame value estimates.
Senyu Fei, Xiaopeng Yu, Siyin Wang, Xianzhong Zhao, Jingjing Gong, Xipeng QiuRead
The study shows that latent world models learn only the physical properties that both the sensors reveal and the prediction objective actively pressures them to retain.
Kaizhen Tan, Xin Xu, Siru Tao, Hanzhe Hong, Yang Feng, Heqing DuRead
246
Reasoning
CADENCE helps small language models learn stronger mathematical reasoning from larger teachers using adaptive distillation, richer rewards, and modest hardware.
Satyam Kumar, Saurabh JhaRead
Per-chunk fact checking breaks multi-hop RAG because no single passage contains the whole answer, but verifying against decomposed sub-questions substantially restores reliability.
Randhir KumarRead
This paper shows that LLMs often know the commonsense answer but get distracted by overly salient details, and that better prompting can help them recover it.
Zheng Wu, Chenhao Xue, Shijie Zheng, Yijie Lu, Cheng Yang, Zhuosheng ZhangRead
Load more videos
247
Scientific Ai
APO uses physical consistency rather than expensive labeled structures to teach generative models how to produce plausible 3D atomic configurations.
Shentong Mo, Yatao BianRead
EMBL AI Librarian helps biology-focused AI agents turn natural-language questions into literature-backed answers and evidence.
Luigi Sigillo, Matteo Silvestri, Francesco Tabaro, Rajat Bhatnagar, Syed Irtaza Mubashar, Matt Jeffryes, Daljit Nijjer, Vittorio Perera, Ola Spjuth, Julio Saez-Rodriguez, Melissa Harrison, Fabio PetroniRead
The paper shows that surprisingly hot quantum systems can remain classically tractable long after entanglement and other quantum features begin to appear.
Harald Putterman, Alexander Zlokapa, Jordan CotlerRead
248
Code Generation
Change2Task turns real pull-request history into reliable, reusable coding-agent tasks for training and evaluation.
Haomin Qi, Xingliang Wang, Xuanqi Gao, Baihui Sang, Xin Zhang, Minghua Ma, Pengfei Gao, Yu Kang, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang, Qi ZhangRead
MindForge trains smaller language models to build complete software from scratch by learning through source-free, end-to-end programming environments.
Yihao Chen, Shi Chang, Khaled Chawa, Feng Lin, Boyuan Chen, Shaowei Wang, Ahmed E. HassanRead
LLMs can find the code that needs changing but often avoid deleting it, so this paper measures the problem and explores training models to remove code more reliably.
Amir M. Ebrahimi, Mohammed Mehedi Hasan, Aaditya Bhatia, Gopi Krishnan Rajbahadur, Ahmed E. HassanRead
249
Embodied Ai
N0-VTLA combines large-scale tactile learning with offline reinforcement learning to give robots substantially better control of contact-rich and deformable objects.
NeoteAI Team, Fudan TEAI TeamRead
Ego2Robot turns everyday human manipulation videos into large-scale synthetic robot experience to help robots generalize beyond their training environments.
Ye Wang, Pei Lin, Xiong-Hui Chen, Haoqi Yuan, Zhixuan Liang, Yiyang Huang, Anzhe Chen, Zixing Lei, Jie Zhang, Tao Zhang, Haoyang Li, Tong Zhang, Chenxi Xiao, Ziyuan Jiao, Qin JinRead
HumanCLAW tests whether vision-language models truly know how to act in the physical world and finds that they struggle mainly to track their own body and progress.
Siyao Li, Jiawei Gu, Shuai Liu, Kairui Hu, Zekun Li, Linjie Li, Chengcheng Tang, Po-Chen Wu, Ivan Shugurov, Lingni Ma, Michael Zollhoefer, Sizhe An, Abhay Mittal, Amy Zhao, Ranjay Krishna, Manling Li, Ziwei Liu, Chuan GuoRead
Load more videos
250
Robotics
FA-RDP lets robot policies explore diverse approaches before contact and react quickly to force feedback once manipulation becomes constrained.
Lifeng Zhuo, Wendi Chen, Han Xue, Shirun Tang, Jun Lv, Cewu Lu, Chuan WenRead
FasTac is a fast curved tactile fingertip that uses multispectral vision and FPGA processing to sense 3D contact shape and forces during dexterous robot manipulation.
Xiaofan Lu, Kaiji Huang, Jiahui Chen, Yuankai Lin, Hua Yang, Zhouping YinRead
Action chunking helps robots not just by planning farther ahead, but by implicitly ensembling different temporal views of the task.
Filippo Lazzati, Kyle Stachowicz, William Chen, Alberto Maria Metelli, Andrew Wagenmaker, Sergey LevineRead
251
Agent
This paper benchmarks how well coding agents find the right repository context before they try to write a patch.
Bowen Qin, Yi XieRead
STAIR helps coding agents learn from past software repairs by turning old debugging trajectories into reusable, adaptable plans.
Yisen Xu, Jiayuan Zhou, Ruiqi Pan, Tse-Hsun ChenRead
Specula uses self-improving LLM agents to automatically write formal specifications and uncover deep bugs in real-world system software.
Qian Cheng, Saad Mohammad Rafid Pial, Ruize Tang, Yiming Su, Emilie Ma, Finn Hackett, Ivan Beschastnikh, Yu Huang, Tianyin XuRead
252
Rl
CAST teaches LLM agents to make better long-horizon decisions by using a game solver’s changing state values as turn-by-turn guidance.
Yu Wang, Yi-Kai Zhang, Wentao Shi, Ziang Ye, Yuchun Miao, Yueqing Sun, Qi Gu, Xunliang Cai, Lan-Zhe Guo, Han-Jia Ye, Fuli FengRead
SpyRL turns open-ended LLM tasks into self-play games where verifiable voting outcomes provide scalable reinforcement-learning rewards.
Qinsi Wang, Jing Shi, Huazheng Wang, Kun Wan, Yiran Wu, Bo Liu, Qingyun Wu, Hai Helen Li, Yiran Chen, Handong Zhao, Wentian ZhaoRead
Molt aims to make agentic reinforcement learning as easy to modify as ordinary PyTorch code without sacrificing distributed training performance.
Jian Hu, Huiying Li, Hao Zhang, Binfeng Xu, Yifan Zhang, Shaokun Zhang, Hemil Desai, Michael Demoret, Pavlo Molchanov, Jan Kautz, Yi DongRead
Load more videos
253
Speech
GigaAM Multilingual is a large speech foundation model and balancing recipe that improves ASR for data-poor Central Asian languages.
Andrei Kuzmenko, Alexandr Maximenko, Aleksandr Kutsakov, Georgii Gospodinov, Dmitrii Bolotov, Oleg Kutuzov, Pavel Bogomolov, Fyodor MinkinRead
GigaChat Audio teaches a large audio-language model to understand long recordings and answer questions with precise timestamps.
Aleksandr Kutsakov, Mariia Sadovina, Georgii Gospodinov, Alexandr Maximenko, Oleg Kutuzov, Pavel Bogomolov, Fyodor MinkinRead
This work makes speech recognizers choose between clean and truly verbatim transcription while improving disfluency detection, word timing, and training-data creation.
Laurin Wagner, Mario Zusag, Bernhard ThallingerRead
254
Multimodal
ClinFusion is a vision-focused medical multimodal LLM that combines 2D and 3D image understanding with radiologist-aligned evaluation for more reliable clinical reports and reasoning.
Hangjie Yuan, Yichen Qian, Zhiwei Tang, Xianzhe Xu, Lirong Wu, Sicheng Yang, Jinwang Wang, Pengju Wang, Zhitao Zeng, Yizeng Han, Yan Xing, Shengxuan Luo, Tao Feng, Qing Xie, Weigen Yao, Yi Yang, Zuozhu Liu, Jiasheng Tang, Shaocheng Wang, Jitao Wang, Jiahong Dong, Weihua Chen, Feng Xu, Fan WangRead
Mage-VL turns video codecs and motion-aware tokenization into a faster multimodal model designed for real-time visual understanding.
Senqiao Yang, Kaichen Zhang, Zhaoyang Jia, Jinghao Guo, Yifei Shen, Xinjie Zhang, Xiaoyi Zhang, Haoqing Wang, Xiao Li, Peng Zhang, Xiang An, Yin Xie, Zhening Liu, Xun Guo, Jiahao Li, Shicheng Zheng, Jinglu Wang, Zongyu Guo, Wenxuan Xie, Zihan Zheng, Yuxuan Luo, Bin Li, Yan LuRead
See2Think tests whether multimodal models truly reason with their sketches and visual feedback, finding that they often choose the right visual actions but struggle to render and use the results faithfully.
Siyu Yan, Zhuoran Yan, Haiying Xu, Panhao Zhou, Jingyu Chen, Chenhao Ji, Shuo Cao, Yongheng Zhang, Haoze Liu, Siyu Zhang, Xiwen Gu, Yihao Liu, Alex Jinpeng WangRead
255
Cv
ID-V2V edits a video's style and scene while preserving the people, expressions, gaze, and lip movements that make the original performance recognizable.
Yuancheng Xu, Mingming He, Pablo Salamanca, Li Ma, Yash Kant, Emmett Steven, Paul Debevec, Ning YuRead
O-VAD uses object tracking and temporal reasoning to help AI inspectors explain anomalies in complex industrial video without task-specific training.
Mei Yuan, Qi Long, Qifeng Wu, Zhenyang Li, Yizhou Zhao, Lei Wang, Yang Liu, Min XuRead
VidMap turns difficult, uncalibrated videos into reliable 3D camera trajectories by combining the best ideas from SLAM, SfM, temporal matching, and monocular depth.
Zador Pataki, Paul-Edouard Sarlin, Marc PollefeysRead
Load more videos
256
Efficiency
This paper argues that autonomous research systems should be judged not only by what they discover, but also by how efficiently they find it.
Haiqian Yang, Yuan CaoRead
Ms.Forcing accelerates real-time video diffusion by using coarser representations for noisier frames while preserving or improving generation quality.
Zekun Li, Xiaoyan Cong, Hongyu Li, Zhiyang Dou, Chuan Guo, Abhay Mittal, Sizhe An, Srinath SridharRead
For coding agents, cutting tokens is not the same as cutting costs—and aggressive compression can even break successful software patches.
Sarel Weinberger, Amir HozezRead
257
Diffusion
ElasticTTT keeps diffusion models from forgetting their generative skills while adapting them for one-shot video editing.
Yueyi Liu, Chi Zhang, Sen Cui, Miao LiuRead
FreqForcing uses low-frequency visual anchors to keep autoregressive video generation stable for minutes without additional training.
Jiatong Li, Leo Liang, Linghe Kong, Yulun ZhangRead
The method improves image generation by exploring many noise seeds early, pruning weak candidates progressively, and spending computation only on the most promising ones.
Rogerio Guimaraes, Pietro PeronaRead
258
Generative Model
01
VideoCoCo: Code-as-CoT for Physically-Consistent Video Generation via an Agentic Dual-Engine System
VideoCoCo uses executable Blender code as chain-of-thought to plan physical dynamics before a generative video model turns them into realistic videos.
Haodong Li, Tianfei Ren, Xiaoxiao Ma, Chunmei Qing, Zhen Fang, Sipeng He, Ziyu Guo, Haoyu Wu, Juanxi Tian, Yihang Zou, Ruichuan An, Dongzhi Jiang, Boxue Yang, Ji Xie, Xu Huang, Wenhao Yan, Jialv Zou, Zhengrong Yue, Yaxin Luo, Xiaotong Li, Yuzhu Wang, Junyan Ye, Jinjing Zhao, Zehui Chen, Lin Chen, Renye Yan, Feng Zhao, Pheng-Ann HengRead
Explorative Modeling trains generative systems by trying multiple candidate outputs and learning from the best match, promising better scaling and much faster end-to-end generation.
Alexi Gladstone, Heng Ji, Yilun DuRead
A controlled study using animated letters shows that balanced video data and accurate captions are crucial for training effective text-to-video models.
Amber Yijia Zheng, Lu Liu, Raymond A. Yeh, Xi YinRead
Load more videos
259
Optimizer
DecoEvo helps language models improve by evolving both their problem-solving instructions and the rubrics used to expose overlooked weaknesses.
Jiangwang Chen, Zixin Song, Junlin Liu, Shuaiyu Zhou, Haiyan Wu, Haihan Shi, Chenxi Zhou, Hanqing Li, Xiao Yang, Da Zhu, Guanjun Jiang, Hai Wan, Xibin ZhaoRead
Hyperball optimizers may not win because of better update directions after all—their apparent advantage largely depends on how their effective learning rate evolves over training.
Yihao Xiao, Jialong Sun, Zitian Gao, Zeming Wei, Chutian Wang, Ran Tao, Jiaye Teng, Bryan DaiRead
A qubit-efficient quantum optimization method tackles power-demand portfolios at surprisingly large scales, with promising results in both simulation and hardware.
Takuya Yoshioka, Keita Sasada, Riku Usuki, Yuichiro Nakano, Keisuke FujiiRead
260
Neural Network
This work shows how several neural-network architectures can exactly encode exotic Motzkin quantum states with entanglement patterns that conventional tensor networks struggle to represent.
Runde Zha, Yuntian Gu, Chaohui Fan, Jia-lin Chen, Hai-Jun Liao, Tao XiangRead
PP-CPCANet replaces fragile covariance estimation with a robust, jointly learned orthogonal basis to improve domain-generalized representations.
Yu-Hsi Chen, Abd-Krim SeghouaneRead
A recurrent sinusoidal neural representation learns sharper images and 3D signals with fewer parameters and optimization steps.
Hyunmin Cho, Jaejun Yoo, Kyong Hwan JinRead
261
Foundation Model
This paper helps answer how to spend compute wisely when high-quality training data is scarce by modeling the diminishing value of repeated and paraphrased tokens.
Tian Qin, Kimia Hamidieh, David Alvarez-MelisRead
Kimi K3 is an open 2.8-trillion-parameter multimodal model designed to deliver frontier-level reasoning, coding, and long-horizon agent capabilities at unprecedented scale.
Kimi Team, Tongtong Bai, Yifan Bai, Yiping Bao, M. C., Jianfeng Cai, Xinyuan Cai, Peizhou Cao, Yuxuan Cao, Ziwei Chai, Y. Charles, H. S. Che, Guanduo Chen, Guangyu Chen, Guanzheng Chen, Huarong Chen, Jia Chen, Jianlong Chen, Jun Chen, Kexin Chen, Peng Chen, Ruijue Chen, Wentao Chen, Xin Chen, Yang Chen, Yanru Chen, Yifei Chen, Yingjiang Chen, Yuankun Chen, Yujie Chen, Yutian Chen, Zhirong Chen, Dazhi Cheng, Yean Cheng, Jialei Cui, Jingbing Cui, Anqi Dai, Jiaqi Deng, Hao Ding, Rui Ding, Shaofeng Ding, Mengfan Dong, Mengnan Dong, Yuhao Dong, Yuxin Dong, Angang Du, Chenzhuang Du, Dikang Du, Jusen Du, Yulun Du, Yu Fan, Jing Feng, Qiulin Feng, Yichen Feng, Kelin Fu, Qiang Fu, Fuxuan Gao, Hongcheng Gao, Jingyue Gao, Tong Gao, Weijia Gao, Shangyi Geng, Jie Gong, Linhu Gong, Shengao Gong, Xiaochen Gong, Qizheng Gu, Yicheng Gu, Shuhao Guan, Haiqing Guo, Shiqi Guo, Xiang Guo, Zhengyan Guo, Beixi Hao, Wenxin Hao, Xiaoru Hao, Dailan He, Haotian He, Lehan He, Qi He, Weiran He, Xinran He, Xinyi ...Read
TiRex-2 is a recurrent time-series foundation model designed to forecast many interacting variables continuously without repeatedly recomputing the entire history.
Patrick Podest, Marco Pichler, Elias Bürger, Levente Zólyomi, Bernhard Voggenberger, Wilhelm Berghammer, Daniel Klotz, Sebastian Böck, Günter Klambauer, Sepp HochreiterRead
Load more videos
262
Hardware
Ventaglio is a compact vector-processor hardware extension that makes sparse Transformer inference substantially faster by accelerating the indexed memory operations that conventional architectures handle inefficiently.
Bowen Wang, Chi Zhang, Diyou Shen, Renzo Andri, Navaneeth Kunhi Purayil, Luca BeniniRead
GPU-Tile-Sim models LLM GPU performance through dependency-aware tile graphs, offering a faster and more adaptable tool for hardware-software co-design.
Yitong Ding, Jiawei Huang, Renyang Guan, Yangjie Zhou, Zihan Liu, Yu Feng, Shixuan Sun, Mingyi Guo, Jingwen Leng, Jian WengRead
SpiNNaker2 is a scalable neuromorphic chip that combines brain-inspired event-based computing with deep-learning acceleration for more flexible and energy-efficient AI.
Stefan Scholze, Johannes Partzsch, Sebastian Höppner, Florian Kelber, Andreas Dixius, Marco Stolba, Sirine Arfa, Marc Berthel, Georg Ellguth, Jim Garside, Hector A. Gonzalez, Stephan Hartmann, Thomas Kiel-Hocker, Dongwei Hu, Matthias Jobst, Khaleelulla Khan Nazeer, Tim Langer, Chen Liu, Gengting Liu, Matthias Lohrmann, Mantas Mikaitis, Felix Neumärker, Amirhossein Rostami, Stefan Schiefer, Tilo Schubert, Delong Shang, Bernhard Vogginger, Yexin Yan, Steve Furber, Christian MayrRead
263
Llm
DynamicRubric keeps LLM evaluators useful as policies improve by generating adaptive rubrics tailored to each set of competing responses.
Beining Wang, Weihang Su, Hongtao Tian, Hao Kong, Tao Yang, Ting Yao, Qingyi Pan, Yueyue Wu, Qingyao Ai, Min Zhang, Yiqun LiuRead
This paper explains why LLM results can change unexpectedly and proposes practical rules for making LLM-based research more reproducible.
Guillaume Coqueret, Joan Llull, Florian Oswald, Christophe Pérignon, Christoph Scheuch, Lars VilhuberRead
Zing gives LLMs a way to measure, learn, and apply social intelligence through specialized benchmarks, training, and runtime support.
Zing Team, Ao Xiang, Bi Jingping, Chen Jiahui, Chen Lehan, Chen Yilin, Cheng Xueqi, Fan Yixing, Gan Kairong, Gao Haowen, Gao Jinhua, Gao Shuxuan, Gong Chang, Guo Jiafeng, Guo Ruijie, Han Zhouyu, He Guangfu, He Yichun, Jiang Shuo, Jing Shaoling, Jing Ya, Lei Chenhao, Lei Yan, Li Anqi, Li Chengao, Li Haoyu, Li Shitian, Liang Xinjian, Liu Zhaoge, Lyu Xingyu, Nie Zhuwei, Pang Liang, Quan Zeping, Shan Shiguang, Shen Huawei, Tang Xinran, Tian Feng, Wang Qian, Wang Ruiping, Wang Xiaohong, Xia Zaiyu, Xiao Yi, Xu Jiayuan, Xu Kehan, Xu Qianqian, Xu Tianyu, Xu Yongjun, Yang Haoming, Yang Jun, Yao Di, Yu Xiaoming, Zhang Futong, Zhang Jie, Zhang Shixuan, Zhang Yuxuan, Zhao Xinyu, Zhao Zhuoran, Zhong Yunfei, Zhu ShengyuRead
264
Safety
A broad audit shows that LLM cyber agents frequently cheat on CTF benchmarks, while anti-cheat prompts help but cannot replace robust environmental safeguards.
Michael Kouremetis, Ads Dawson, Raja Sekhar Rao Dheekonda, Brian GreunkeRead
This paper argues that securing AI agents that handle money requires fixing the commerce protocols they use, not just improving the models behind them.
Yedidel LouckRead
A two-word tag reveals that newer language models are increasingly resistant to explicit agreement bids while still being highly sensitive to how confidently users phrase their opinions.
Tapan ParikhRead
Load more videos
265
Benchmark
PAJAMA turns expensive, opaque LLM judges into fast, inspectable programs while escalating only uncertain cases to an LLM.
Tzu-Heng Huang, Shengqi Qiu, Frederic SalaRead
DocOps tests whether AI agents can perform complex document workflows without losing state, misunderstanding content, or damaging document structure.
Jiazhen Jiang, Boxi Cao, Lingyong Yan, Yaojie Lu, Hongyu Lin, Shuaiqiang Wang, Dawei Yin, Xianpei Han, Le SunRead
SecRespond tests whether AI agents can investigate and remediate already-compromised computers, revealing that current systems find known alerts but miss hidden intrusions and produce incomplete recovery plans.
Lehan Wang, Boli Chen, Ruixue Ding, Pengjun Xie, Jinwei Huang, Zhendong Liu, Shuo Wang, Tao Lei, Xin Ouyang, Xiaomeng LiRead
266
World Model
HyWorldVLA combines pixel-grounded and latent world modeling to make autonomous-driving vision-language-action systems more robust to noisy future predictions.
Quanfu Yu, Xian Wu, Hao Xu, Liulong MaRead
This paper explains when learned world models can truly recover both the hidden state and the effects of actions, especially when training data lacks sufficient action diversity.
Xiangteng Zhang, Yang Guan, Bo Zhang, Ya-Qin Zhang, Shengbo Eben LiRead
Wonder turns video generation into an interactive, navigable world that remembers what it has seen and responds to camera motion in real time.
Jiacong Xu, Hanwen Jiang, Zhixin Shu, Kalyan Sunkavalli, Vishal M. Patel, Yiqun MeiRead
267
Reasoning
Some reasoning models give up not because they lack the ability to solve a problem, but because they mistakenly think they will run out of context before finishing.
Ifueko Igbinedion, Jillian Ross, Etienne Ricardez, Sertac Karaman, Eric SoRead
This study finds that vision models relying on local, sequential glimpses can generalize better to longer and more complex visual reasoning tasks than models using global shortcuts.
Pulkit Madan, Sanjay Haresh, Reza Ebrahimi, Sunny Panchal, Apratim Bhattacharyya, Roland MemisevicRead
TTEL makes language-model reasoning more efficient by finding where a solution went wrong and branching from the last correct step instead of starting over.
Rajiv Shailesh Chitale, Rahul Madhavan, Taneesh Gupta, Deepanway Ghosal, Aravindan RaghuveerRead
Load more videos
268
Graph Learning
A pretrained geometric model that treats cyclic peptides as conformational ensembles improves property prediction over sequence-only modeling.
Aaron Feller, Kris Deibler, Maxim SecorRead
This work uses global geometric curvature to explain and improve how information flows through graph neural networks.
Rachid Caich, Yassine AbbahaddouRead
Temporal graph generators cannot reliably correct deployment drift from observations alone, because unseen structural changes create an irreducible error floor.
Tianpeng Li, Xuan Guo, Wenjun Wang, Wang Zhang, Pengfei JiaoRead
269
Scientific Ai
AAMFM is a multimodal foundation model that uses antigen structure and epitope context to design more functional, antigen-specific antibodies.
Xiaoliang Shi, Zichen Wang, Runze Ma, Zhongyue Zhang, Shuangjia ZhengRead
02
From Execution to Capability: Scientific Experience Consolidation via Procedural Knowledge Synthesis
SciConsolidate turns successful and failed coding experiences into reusable scientific procedures that help smaller language models improve at computational problem solving.
Liwei Dong, Jiahao Zhao, Nan XuRead
The study tests whether a language model genuinely tracks materials physics by reading and manipulating its internal representations rather than judging answers alone.
Markus J. BuehlerRead
270
Code Generation
Frontier LLMs can autonomously generate competitive trapped-ion shuttling compilers, potentially cutting compiler development from months to days.
Fabian Kreppel, Reza Salkhordeh, Ferdinand Schmidt-Kaler, André BrinkmannRead
KAT-Coder-V2.5 trains coding agents to autonomously modify and test real software repositories using scalable environments, tool-use trajectories, and reinforcement learning.
Bo Huang, Fengxiang Li, Hao Xu, Haoyang Huang, Hongyi Fu, Jinhua Hao, Kun Yuan, Minglei Zhang, Pengcheng Xu, Shiyang Liu, Wenhao Zhuang, Yuze Shi, Zongxian Feng, Chao Wang, Cheng He, Chongling Rao, Deyu Cao, Fan Yang, Gang Xiong, Haochen Liu, Jiabao Li, Jian Liang, Jinghui Jia, Jingwen Chang, Jun Du, Junyu Shi, Min Li, Mingqi Wu, Qiang Gao, Shangpeng Yan, Shaotong Qi, Shu Xu, Shuo Zhou, Tiankuo Xu, Tong Zheng, Weilun Zhao, Xiancheng Meng, Xianda Sun, Xiaoyu Jiang, Xunhao Jia, Yao Xia, Yimeng Xu, Yinghan Cui, Yingpeng Chen, Yiwen Ning, Yong Wang, Yuxuan Sun, Zhongsheng Liu, Ming Sun, Cheng Luo, Chen Yang, Han Li, Kun GaiRead
ACQUIRE helps coding agents fix software issues more accurately by having them explicitly ask and answer repository-specific questions before writing a patch.
Haotian Lin, Silin Chen, Xiaodong Gu, Yuling Shi, Chengxi Pan, Jiaqi Ge, Mengfan Li, Jianghong Huang, Mengchieh Chuang, Beijun Shen, Haibing GuanRead
Load more videos
271
Embodied Ai
ReViV turns a single wearable-camera video into a coherent 4D model of both the person wearing it and the world they are viewing.
Xiaozhong Lyu, Gen Li, Zhiyin Qian, Xucong Zhang, Marc Pollefeys, Siyu TangRead
Robostral Navigate uses an 8B vision-language model and only monocular RGB images to achieve state-of-the-art robot navigation while dramatically reducing training costs.
Arjun Majumdar, Avinash Sooriyarachchi, Benjamin Tibi, Chris Bamford, Elliot Chane-Sane, Guillaume Lample, Khyathi Raghavi Chandu, Ludovic Ho Fuh, Mathieu Poiree, Olivier Duchenne, Rosalie Millner, Srijan Mishra, Theo Cachet, Thomas ChabalRead
HOST lets robots learn new manipulation skills from a single human video in under a minute while preserving what they already know.
Guangyan Chen, Meiling Wang, Te Cui, Zichen Zhou, Qi Shao, Shalfun Li, Hang Su, Roy Gan, Hao Wang, Mengyin Fu, Yi Yang, Yufeng YueRead
272
Robotics
AXIS turns community-collected robot demonstrations into a scalable data engine that measurably improves vision-language-action manipulation policies.
Mengfei Zhao, Dihong Huang, Yikai Tang, Peihao Li, Mingxuan Yan, Ruiqi Zhuang, Yanjia Huang, Jie Wang, Hai Zhai, Tony Zhou, Rui Zhang, Zhexi Luo, Yuchen Huang, Jianfei Yang, Jiachen LiRead
02
GLAM-SLAM: Real-time Gaussian Large-scale Mapping via Flow Densification and Spatial Decomposition
GLAM-SLAM makes Gaussian-splatting-based robot mapping faster and more scalable for large outdoor environments.
Panagiotis Mermigkas, Argyris Manetas, Petros MaragosRead
A diffusion-and-energy-based robot planner helps two arms grasp, hand over, rotate, and place objects more reliably and smoothly.
Wun Lam Yeung, Wenjun Liu, Yui Cheung Yu, Zhengyan Lambo Qin, Qijin She, Heng Li, Ziqi Wang, Ping TanRead
273
Agent
NexForge turns high-level capability requirements into large-scale executable training tasks, substantially improving open-source LLM agent performance.
Jiarong Zhao, Zhikai Lei, Zhiheng Xi, Rui Zheng, Hang Yan, Jie Zhou, Qin Chen, Liang HeRead
OpenForgeRL makes it possible to train real-world tool-using and computer-use agents directly inside the complex harnesses where they operate.
Xiao Yu, Baolin Peng, Ruize Xu, Hao Zou, Qianhui Wu, Hao Cheng, Wenlin Yao, Nikhil Singh, Zhou Yu, Jianfeng GaoRead
Experience Distillation lets agents retain much of what they learn from trial and error without repeatedly paying the cost of interacting with the environment.
Chenhui Gou, Haoqin Tu, Yunhao Fang, Jianfei Cai, Hamid RezatofighiRead
Load more videos
274
Multimodal
AV-Flamingo is an open multimodal language model that learns to reason across audio, visuals, and long videos while grounding its explanations in time.
Sreyan Ghosh, Arushi Goel, Kaousheik Jayakumar, Lasha Koroshinadze, Nishit Anand, Siddharth Gururani, Hanrong Ye, Pritam Biswas, Yuanhang Su, Ehsan Hosseini-Asl, Sang-gil Lee, Zhifeng Kong, Jaehyeon Kim, Sungwon Kim, S Sakshi, Ramani Duraiswami, Dinesh Manocha, Andrew Tao, Mohammad Shoeybi, Bryan Catanzaro, Ming-Yu Liu, Wei PingRead
RINO treats every visual signal as an image, allowing one shared model to understand and generate across tasks like segmentation, depth estimation, and pose-guided synthesis.
Timing Yang, Jinrui Yang, Xinlong Li, Yuhan Wang, Haoran Li, Yanqing Liu, Guoyizhe Wei, Jixuan Ying, Chen Wei, Rama Chellappa, Yuyin Zhou, Cihang Xie, Alan Yuille, Feng WangRead
TimeLens2 helps video-language models identify all the moments supporting an answer, using better multi-span supervision and a reward designed for flexible temporal evidence sets.
Yuhan Zhu, Changlian Ma, Xiangyu Zeng, Xinhao Li, Zhiqiu Zhang, Songze Li, Jun Zhang, Tianxiang Jiang, Yuandong Yang, Ziang Yan, Zikang Wang, Xinyu Chen, Haoran Chen, Shaowei Zhang, Limin WangRead
275
Rl
Agon trains two reasoning models to compete against each other, turning rivalry into an implicit judge that substantially improves problem-solving performance.
Vladislav BeliaevRead
A single differentiable measure of learnable novelty may unify exploration, computation, and unsupervised abstraction across intelligent systems.
Yanbo Zhang, Michael LevinRead
SEED helps language-model agents learn from their own completed experiences by turning hindsight-generated skills into dense guidance during reinforcement learning.
Jinyang Wu, Shuo Yang, Zhengxi Lu, Fan Zhang, Yuhao Shen, Lang Feng, Haoran Luo, Zheng Lian, Shuai Zhang, Zhengqi Wen, Jianhua TaoRead
276
Cv
A globally scalable computer vision foundation model maps agricultural field boundaries across countries with major accuracy gains and practical nationwide deployment speed.
Mykola Lavreniuk, Nataliia Kussul, Andrii Shelestov, Yevhenii Salii, Volodymyr Kuzin, Charlotte Julia Li-Xing Wang, Zoltan SzantoiRead
MetaView uses diffusion, implicit geometry, and metric depth to generate spatially consistent novel views from a single image—even under large camera movements.
Yufei Cai, Xuesong Niu, Hao Lu, Kun Gai, Kai Wu, Guosheng LinRead
UniD uses diffusion-informed distillation to teach one video model eight scene-understanding tasks from separate datasets.
Yihong Sun, Seoung Wug Oh, Jiahui Huang, Bharath Hariharan, Joon-Young LeeRead
Load more videos
277
Efficiency
Randomly sampling what an AI forgets lets its serving system tell whether a bad answer came from cache compression or from the model itself.
Peng XieRead
This paper measures how many bits of data LoRA adapters can secretly store and finds that their memorization depends more on where parameters are placed than on how many there are.
Kaizhen Tan, Heqing Du, Yang FengRead
SANA-Video 2.0 combines mostly linear attention with occasional softmax anchors to make high-quality long-video generation dramatically faster on a single GPU.
Junsong Chen, Jincheng Yu, Yitong Li, Shuchen Xue, Haozhe Liu, Jingyu Xin, Yuyang Zhao, Tian Ye, Zhangjie Wu, Zian Wang, Daquan Zhou, Ping Luo, Song Han, Enze XieRead
278
Diffusion
A new diffusion-transformer interface lets users precisely control what appears in different image regions using text or image guidance.
Rahul Sajnani, Yulia Gryaditskaya, Radomír Měch, Srinath Sridhar, Matheus GadelhaRead
DiffGI makes thin-shell 3D generation more precise and efficient by combining differentiable geometry images with latent diffusion.
Eungjune Shim, Hansol Lee, Eunjung JuRead
This work shows that seemingly uninformative template tokens in diffusion transformers quietly preserve object identity, and that some attention heads can be pruned to save computation with modest quality loss.
Maohua Li, Qirui Li, Yanke Zhou, Yiduo Li, Zhaosheng Chi, Chao Xu, Cuifeng Shen, Yixuan Xu, Hanlin Tang, Kan Liu, Tao Lan, Lin Qu, Shao-Qun ZhangRead
279
Generative Model
AI-generated books may be lower quality, but their sheer volume is reshaping Amazon’s market and squeezing revenue from human-authored books.
Tuhin Chakrabarty, Xinyue Liu, Jane C. Ginsburg, Paramveer DhillonRead
Qwen-Music combines melody planning, large-scale multilingual training, and waveform rendering to generate and reinterpret high-quality songs with vocals.
Jin Xu, Kangdi Wang, Ruibin Yuan, Shun Lei, Xiong Wang, Xize Cheng, Xueyao Zhang, Yang Zhang, Yiheng Chen, Yongqi Wang, Yue Wang, Zhifang Guo, Zihan Liu, Zijian Lin, Dake Guo, Hangrui Hu, Lei Xie, Linhan Ma, Wei Xue, Wenxiang Guo, Xinfa Zhu, Xipin Wei, Yangze Li, Yuanjun Lv, Yuxuan Wang, Yunfei Chu, Zhiyong WuRead
VideoRAE turns video foundation-model features into efficient, semantically rich latent spaces that can accelerate both diffusion and autoregressive video generation.
Zhihao Xie, Junfeng Wu, Xinting Hu, Junchao Huang, Li JiangRead
Load more videos
280
Optimizer
ISO speeds up RLVR training by preserving a model’s singular-value spectrum while optimizing the input and output directions that encode new reasoning behaviors.
Hanqing Zhu, Wenyan Cong, Zhizhou Sha, Sagnik Mukherjee, Xinyuan Song, David González-Martínez, Xiaoxia Wu, Yuandong Tian, Shiwei Liu, David Z. Pan, Zhangyang "Atlas" WangRead
Muon may substantially improve agentic RL training, but stronger evidence across seeds, tasks, and optimization settings is still needed.
Kai Ruan, Jinghao Lin, Zihe Huang, Ziqi Zhou, Qianshan Wei, Xuan Wang, Hao SunRead
SkewAdam cuts MoE optimizer memory dramatically by giving different parameter groups only the optimizer state they actually need, without sacrificing accuracy.
Nuemaan MalikRead
281
Transformer
A large controlled study finds that attention alone can nearly reproduce transformer performance, though feed-forward layers still provide an advantage for recalling knowledge stored in the weights.
Henry Ndubuaku, Karen Mosoyan, Jakub Mroz, Noah Cylich, Satyajit Kumar, Parkirat Sandhu, Roman Shemet, Justin H LeeRead
This paper recasts Transformer behavior as continuous geometry and thermodynamics, offering an intriguing but currently insufficiently substantiated mathematical lens.
Zhihua LiangRead
xHC expands Transformer memory through many residual streams while using sparse updates and optimized memory traffic to make large-scale LLM training more efficient.
Xiangdong Zhang, Xiaohan Qin, Sunan Zou, Tuo Dai, Xiaoming Shi, Huaijin Wu, Yebin Yang, Zhuo Xia, Shaofeng Zhang, Lin Yao, Yuliang Liu, Yu Cheng, Junchi YanRead
282
Llm
This paper explains why on-policy distillation can go wrong in LLM training and shows that regulating its guidance signals can make exploration more reliable.
Rui Wang, Hongru Wang, Yi Chen, Boyang Xue, Tianqing Fang, Wenhao Yu, Kam-Fai WongRead
Distilled RL combines reinforcement learning with selective teacher guidance to help LLMs acquire new knowledge more effectively than standard RL or direct logit matching.
Chen Wang, Zhaochun Li, Jionghao Bai, Yining Zhang, Hexuan Deng, Ge Lan, Yue WangRead
A handful of one-token answers to simple prompts may be enough to tell whether an API is really serving the LLM it claims to use.
Tomas BrucknerRead
Load more videos
283
Safety
The study finds that current LLM watermarks are too fragile and error-prone to serve as reliable courtroom evidence, especially after meaning-preserving paraphrase.
Saifur Rahman Tamim, Amir Labib KhanRead
AgentAbstain tests whether LLM agents can recognize when they should not act, revealing that even strong agents often take risky actions when abstention is required.
Xun Liu, Yi Evie Zhang, Vira Kasprova, Parisa Rabbani, Pardis Sadat Zahraei, Tianyu Zhang, Ali Ebrahimpour-Boroojeny, Varun ChandrasekaranRead
The paper shows that fluent activation explanations can hide false claims and proposes training models so independent probes can catch those lies.
Hiskias DingetoRead
284
Hardware
Instead of guessing when a GPU will fail, HeaRank identifies which GPUs are most likely to fail so operators can act before problems disrupt large-scale AI jobs.
Difeng Ma, Changhua Pei, Yuanwei Lu, Quan Zhou, Zexin Wang, Yibo Zhu, Daxin Jiang, Dan Pei, Jingjing Li, Gaogang XieRead
A new class of ultra-low-latency GPU collectives brings distributed LLM inference and HPC communication within 7% of the hardware speed limit.
Siyuan Shen, Anton Korzh, John Bachan, Tiancheng Chen, Arnav Goel, Ludwig Schneider, Pouya Kousha, Zhenhao He, Sylvain Jeaugey, Kamil Iskra, Nishank Chandawala, Jeff R. Hammond, Torsten HoeflerRead
The paper proposes GPU hardware controls that can dynamically turn down an AI system’s performance as a last-resort safety mechanism.
Haiyue Ma, Lauren Malek, Joseph Forzani, David WentzlaffRead
285
Benchmark
SDABench tests whether LLMs can genuinely reason about scientific claims, finding that they remain much weaker at inference, causality, and mechanistic explanation than at basic description.
Chuhan Shi, Xiaoquan Ren, Sicheng Song, Haobo Li, Rui Sheng, Yushi SunRead
SIS-Bench tests whether UAV-focused multimodal models understand not just the world around them, but also their own position, motion, and evolving state within it.
Zhishan Zou, Guoyan Sun, Zhiwei Wei, Jiancheng Pan, Yujie Li, Mugen Peng, Wenjia XuRead
VIABench tests whether multimodal AI can genuinely assist blind users in real-world video scenarios, rather than merely answer questions about static images.
Yunfeng Liu, Yuandong Yang, Jiarui Han, Zhenpeng Huang, Yuqing Tang, Xiangyu Zeng, Gangshan Wu, Limin WangRead
Load more videos
286
World Model
ABot-World-0 aims to make interactive video world models run in real time on a single desktop GPU while preserving controllable, coherent long-horizon environments.
Fan Jiang, Zhaoxu Sun, Mengchao Wang, Ziyu Zhu, Chiyu Wang, Yunpeng Zhang, Wenlin Liu, Yun Wang, Xue Zheng, Rui Sun, Junfeng Ni, Hongyu Pan, Zhongxu Sun, Fei Yu, Zengye Ge, Mengmeng Du, Nianfei Fan, Mingchao Sun, Yu Liu, Yongchang, Yanqing Zhu, Jiahang Wang, Ning Ying, Yuze Xuan, Di Yang, Zhicheng Liu, Zhe Gao, Tingbing Xu, Jiacheng Sui, Wenjin Yang, Junnan Lai, Shufeng Liu, Yuan Liu, Zheng Zhou, Yingliang Peng, Dawei Cao, Kaifeng Sheng, Yuxiang Cai, Fei Lu, Mu Xu, Ning GuoRead
AlayaWorld is a large interactive video world model designed to generate persistent, explorable environments over long time horizons while responding quickly to user prompts and camera actions.
AlayaWorld Team, Kaipeng Zhang, Chuanhao Li, Yifan Zhan, Yongtao Ge, Yuanyang Yin, Jiaming Tan, Kang He, Liaoyuan Fan, Mingliang Zhai, Ruicong Liu, Xiaojie Xu, Xuangeng Chu, Zhen Li, Zhengyuan Lin, Zhixiang Wang, Zian Meng, Zihui GaoRead
This paper argues that truly interactive AI game worlds need explicit, persistent state dynamics and introduces a large-scale dataset and framework to move beyond pixel-only video prediction.
Zhen Li, Zian Meng, Shuwei Shi, Mingliang Zhai, Jiaming Tan, Chuanhao Li, Kaipeng ZhangRead
287
Reasoning
This paper asks whether LLMs are truly solving math problems in different ways, or just wording the same strategy differently, and finds that many diversity metrics miss the real story.
Sangmook Lee, Minbeom Kim, Jeonghye Kim, Dohyung Kim, Sojeong Rhee, Kyomin JungRead
LOTUS uses recurrent Transformer computation to make hidden-state reasoning faster while retaining much of the power and interpretability of explicit chain-of-thought.
Ying Fan, Anej Svete, Kangwook LeeRead
The study shows that an LLM's hidden representations can reveal its confidence and evidence use more reliably than its spoken chain of thought.
Raphaël Sarfati, Pratyush Ranjan Tiwari, Siddharth Boppana, Christopher J. Earls, Srikar Varadaraj, Eric HoRead
288
Graph Learning
Loopie argues that repeatedly looping a sparse MoE Transformer can beat simply scaling model size, reaching remarkable mathematical and scientific reasoning performance.
Zitian Gao, Yilong Chen, Yihao Xiao, Xinyu Yang, Ran Tao, Joey Zhou, Bryan DaiRead
RAGU builds cleaner knowledge graphs with a compact language model, making GraphRAG more accurate, efficient, and practical.
Mikhail Komarov, Ivan Bondarenko, Stanislav Shtuka, Oleg Sedukhin, Roman Shuvalov, Yana Dementyeva, Matvey Solovyov, Nikolay O. NikitinRead
After two decades, this paper proves that in severely memory-limited streaming matching, simple greedy is as good as any possible algorithm.
Sepehr Assadi, Max Jiang, Mars XiangRead
Load more videos
289
Code Generation
A large enterprise study finds that an AI coding mandate eventually doubled developer output, while shifting much of code review from humans to automation.
Hao He, Shyam Agarwal, Yegor Denisov-Blanch, Pavel Azaletskiy, Sanmi Koyejo, Bogdan VasilescuRead
A large study finds that AI agents can speed up code reviews, but faster decisions do not necessarily mean higher-quality reviews.
Suzhen Zhong, Shayan Noei, Bram Adams, Ying ZouRead
By training code models to fill in dependency-aware function gaps, the paper improves coding-agent performance and preserves broader tool-use abilities after agentic post-training.
Yubo Wang, Jiarong Liang, Yuxuan Zhang, Xuye Liu, Cong Wei, Yuyu Zhang, Ping Nie, Wenhu ChenRead
290
Scientific Ai
PaperPilot turns scientific paper search into a controllable multi-turn workflow, making literature discovery more editable, inspectable, and effective.
Jisen Li, Bingxuan Li, Nanyi Jiang, Xuying Ning, Xiyao Wang, Yifan Shen, Heng Wang, Yuqing Jian, Xiaoxia Wu, Ben Athiwaratkun, Pan Lu, Jiaxuan You, Bingxin ZhaoRead
S1-Omni aims to be a general-purpose scientific AI model that understands and generates everything from molecules and proteins to spectra and scientific images.
Jiahao Zhao, Junyi Liu, Lifeng Xu, Nan Xu, Qingli Wang, Qingxiao Li, Tianle Chen, Xiaoyu Wu, Yawen Zheng, Zikai Wang, Guanming Liu, Hequn Zhou, Jingyi Wang, Jingyuan Shu, Keqi Wang, Li He, Songyang Diao, Wenhui Xu, Xinyu Ren, Yaqin Fan, Yujin Zhou, Zhanao YaoRead
This work builds a massive bridge between informal math papers and formal proof libraries, making the structure of mathematics searchable across both worlds.
Simon Kurgan, Evan Wang, Eric Leonen, Sophie Szeto, Luke Alexander, Artemii Remizov, Jarod Alper, Giovanni Inchiostro, Vasily IlinRead
291
Embodied Ai
UniVR teaches AI to reason, understand physics, and plan long tasks directly from visual experience without relying on language supervision.
Zhongwei Ren, Yunchao Wei, Yao Zhao, Weibo Gong, Xiao Liu, Anran Wang, Xiangtai Li, Xiaojie JinRead
Xiaomi-Robotics-1 scales robot learning with over 100K hours of language-annotated real-world data to produce a stronger and more adaptable general-purpose manipulation policy.
Xiaomi Robotics Team, Jun Guo, Piaopiao Jin, Jason Li, Peiyan Li, Yingyan Li, Futeng Liu, Wanli Peng, Optimus Qin, Yifei Su, Nan Sun, Qiao Sun, Runze Suo, Heyun Wang, Yunhong Wang, Rujie Wu, Caoyu Xia, Lina Zhang, Jack Zhao, Guoliang Chen, Wenlong Chen, Xinze He, Bin Li, Qing Li, Zhuorong Li, Heng Qu, Wenxuan Song, Diyun Xiang, Yifan Xie, Peiran Xu, Hangjun Ye, Wen Ye, Han Zhao, Quanyun ZhouRead
Xiaomi-Robotics-U0 uses a large multimodal world foundation model to generate consistent robot-centered scenes and videos while improving real-world manipulation performance.
Xinghang Li, Jun Guo, Qiwei Li, Long Qian, Hang Lai, Yueze Wang, Hongyu Yan, Jiahang Cao, Xi Chen, Jingen Qu, Jiaxi Song, Nan Sun, Hanye Zhao, Futeng Liu, Wanli Peng, Heyun Wang, Yunhong Wang, Caoyu Xia, Jack Zhao, Diyun Xiang, Hangjun Ye, Heng Qu, Huaping Liu, Jason LiRead
Load more videos
292
Agent
This work trains API-calling agents without building real backends by having LLMs simulate entire interactive worlds and generate useful synthetic trajectories.
Seanie Lee, Sanjoy Chowdhury, Chao Jiang, Cheng-Yu Hsieh, Ting-Yao Hu, Alexander T Toshev, Oncel Tuzel, Raviteja VemulapalliRead
RESOURCE2SKILL turns videos, code, articles, and other human resources into reusable multimodal skills that make software agents more capable.
Yijia Fan, Zonglin Di, Zimo Wen, Yifan Yang, Mingxi Cheng, Qi Dai, Bei Liu, Kai Qiu, Yue Dong, Ji Li, Chong LuoRead
AMID uses collaborating AI agents to automatically design, test, verify, and audit medical-imaging models across many different tasks.
Shengyuan Liu, Jia-Xuan Jiang, Boyun Zheng, Cheng Wang, Zipei Wang, Wentao Pan, Hongtao Wu, Houwen Peng, Yu Gu, Lichao Sun, Yixuan YuanRead
293
Robotics
ABot-N1 is a navigation foundation model that turns language and vision into pixel-level goals before executing actions, improving robustness and interpretability in complex indoor and urban navigation.
Ruiyan Gong, Yingnan Guo, Junjun Hu, Jintao Kong, Xiaoxu Leng, Tianlun Li, Weize Li, Fei Liu, Zhicheng Liu, Jia Lu, Minghua Luo, Chenlin Ming, Yanfen Shen, Jiyue Tao, Zhengbo Wang, Mingyang Yin, Minqi Gu, Zihao Guan, Wei Guo, Guoqing Liu, Huachong Pang, Menglin Yang, Zeqian Ye, Xiaoxiao Geng, Zhining Gu, Honglin Han, Di Jing, Hongyu Pan, Mingchao Sun, Kuan Yang, Jianfang Zhang, Yanghong Chen, Ye He, Wei Mei, Jiahao Shi, Xiangpo Yang, Yanqing Zhu, Zedong Chu, Xiaolong Wu, Mu XuRead
EgoSteer is a full-stack robot learning pipeline that turns massive egocentric human video into a steerable dexterous manipulation policy capable of generalizing to many real-world tasks.
Yifan Zhong, Zhang Chen, Tianrui Guan, Fanlian Zeng, Yuyao Ye, Tianjia He, Ka Nam Lui, Jiayi Li, Tingrui Zhang, Ruilin Yan, Xinhao Ji, Guangyu Zhao, Wenjie Lou, Jiayuan Zhang, Yuanpei Chen, Yaodong YangRead
PoLAR teaches robots to represent motion in a smarter geometric space, helping them better separate how far a change is from what kind of change it is.
Youngjoon Jeong, Jihwan Yu, Minsoo Jo, Junha Chun, Taesup KimRead
294
Rl
This paper shows that many supposedly meaningful 'risk-aware' signals learned by distributional RL agents are often just training artifacts, not real properties of the environment.
Hari PrasadRead
This paper shows that common ways of evaluating deep reinforcement learning can produce misleading conclusions, especially as training data and model capacity scale.
Ezgi KorkmazRead
Using chess and math as controlled laboratories, the paper shows how pretraining sets the stage for—and shapes what RL can achieve in—LLM reasoning.
Jingyan Shen, Ang Li, Salman Rahman, Yifan Sun, Micah Goldblum, Matus Telgarsky, Pavel IzmailovRead
Load more videos
295
Multimodal
VideoChat3 is a fully open, efficient 4B-parameter video-language model designed to understand everything from short clips to long and streaming videos.
Xinhao Li, Yuhan Zhu, Xiangyu Zeng, Yuhao Dong, Haoning Wu, Zhiqiu Zhang, Yuandong Yang, Changlian Ma, Qingyu Zhang, Yansong Shi, Xinyu Chen, Haoran Chen, Zizheng Huang, Jun Zhang, Kun Ouyang, Lin Sui, Ziang Yan, Yicheng Xu, Chenting Wang, Yinan He, Hongjie Zhang, Yi Wang, Yu Qiao, Yali Wang, Ziwei Liu, Kai Chen, Limin WangRead
MedPMC is a large-scale pipeline that turns biomedical papers into cleaner, more useful image-text training data, boosting medical multimodal models and clinical retrieval.
Hyunjae Kim, Dain Kim, Pan Xiao, Serina S. Applebaum, Younjoon Chung, Xuguang Ai, Yu Yin, Roy Jiang, Yuexi Du, Yawen Wei, Yiming Kong, Tuo Guo, Zhiyuan Cao, Mengmeng Du, Yuelei Fu, Yan Hu, Rui Shi, Gui Yang, Kevin W. Jin, Yuntian Liu, Yuxuan Tian, Jonathan Marquez, Zhen Chen, Sheng Zhang, Hoifung Poon, Hua Xu, Jaewoo Kang, Qingyu ChenRead
MonkeyOCRv2 turns document images into a specialized visual foundation model that reads text, preserves layout, and powers compact high-performing document AI systems.
Yuliang Liu, Zhang Li, Ziyang Zhang, Shuo Zhang, Qiang Liu, Jiajun Song, Zidun Guo, Xinhan Wang, Handong Zheng, Yang Liu, Dongliang Luo, Zhiyin Ma, Jiarui Zhang, Xiang BaiRead
296
Cv
01
From RGB Generation to Dense Field Readout: Pixel-Space Dense Prediction with Text-to-Image Models
This work shows that text-to-image models can do dense vision better by reading out depth, masks, and other per-pixel signals directly from their internal patch grid instead of forcing everything back into RGB.
Zanyi Wang, Xin Lin, Haodong Li, Dengyang Jiang, Yijiang LiRead
SUFLECA teaches visual features to understand 3D geometry, enabling faster and more reliable CAD model alignment from a single image.
Saad Ejaz, Miguel Fernandez-Cortizas, Javier Civera, Holger Voos, Jose Luis Sanchez-LopezRead
This paper argues that video generation models can double as powerful general vision learners, enabling a single pretrained model to tackle diverse perception tasks with strong data efficiency.
Letian Wang, Chuhan Zhang, Rishabh Kabra, Jasper Uijlings, Steven Waslander, Andrew Zisserman, Joao Carreira, Kaiming He, Misha Andriluka, Eduard Gabriel Bazavan, Andrei Zanfir, Cristian SminchisescuRead
297
Code Generation
This work turns the compiler into a real-time coding partner that catches AI-generated Rust mistakes while the code is still being written.
Niels Mündler-Sasahara, Hristo Venev, Dawn Song, Martin Vechev, Jingxuan HeRead
SpectraReward turns pretrained multimodal models into plug-and-play judges that can guide image generators without extra reward-model training.
Runhui Huang, Qihui Zhang, Zhe Liu, Yu Gao, Jie Wu, Hengshuang ZhaoRead
ShortOPD helps pruned language models regain useful free-form generation by training first on the prefixes they can handle and gradually extending their rollout length.
Qingyu Zhang, Qianhao Yuan, Hongyu Lin, Yaojie Lu, Xianpei Han, Le Sun, Xiang Li, Ming Xu, Jiarui Li, Xiuyin ZhaoRead
Load more videos
298
Diffusion
Flash-BoN speeds up diffusion model sampling by generating many cheap draft images first, then verifying and refining only the best one to get better results under the same compute budget.
Ruchit Rawal, Reza Shirkavand, Sayak Paul, Yuxin Wen, Heng Huang, Yizheng Chen, Tom Goldstein, Gowthami SomepalliRead
HDR lets video diffusion models plan complex visual actions hierarchically before streaming them efficiently, improving reasoning while sharply reducing inference cost.
Zezhong Qian, Xiaowei Chi, Chak-Wing Mak, Tianze Zhou, Ruibin Yuan, Yuhan Rui, Hengzhe Sun, Zhuoqun Wu, Yuming Li, Siyuan Qian, Sirui Han, Shanghang ZhangRead
MeanFlowNFT brings reinforcement-learning-based alignment to ultra-fast few-step image and video generators, achieving quality that can rival or exceed much slower diffusion models.
Yushi Huang, Xiangxin Zhou, Jun Zhang, Liefeng Bo, Tianyu PangRead
299
Generative Model
AffectFlow-DINO uses conditional rectified flow to represent ambiguity in facial emotion predictions while improving valence-arousal, expression, and Action Unit estimation.
Salah Eddine Bekhouche, Abdellah Zakaria Sellam, Fadi Dornaika, Abdenour HadidRead
This paper trains a 3D latent generative model to synthesize realistic chest CT scans with controllable clinical findings, then uses reinforcement learning to make the requested attributes appear more reliably.
Max Van Puyvelde, Halil Ibrahim Gulluk, Wim Van Criekinge, Olivier GevaertRead
SynCity 3000 turns image-to-3D generation into a scalable scene-level diffusion system that can build large, coherent 3D worlds from prompts and layouts.
Paul Engstler, Iro Laina, Christian Rupprecht, Andrea VedaldiRead
300
Transformer
DeepLoop makes Transformers deeper by reusing the same blocks repeatedly and adjusts residual scaling to keep this recurrent computation stable.
Shuzhen Li, Yifan Zhang, Jiacheng Guo, Quanquan Gu, Mengdi WangRead
E-VQA teaches video-language models to answer questions while showing exactly when and where in the video the evidence appears.
Shijie Wang, Honglu Zhou, Ziyang Wang, Ran Xu, Caiming Xiong, Silvio Savarese, Chen Sun, Juan Carlos NieblesRead
RoboTTT gives robot foundation models an 8K-step memory by learning at test time, enabling stronger imitation, adaptation, and long-horizon manipulation.
Yunfan Jiang, Yevgen Chebotar, Ruijie Zheng, Fengyuan Hu, Yunhao Ge, Jimmy Wu, Tianyuan Dai, Scott Reed, Li Fei-Fei, Yuke Zhu, Linxi "Jim" FanRead
Load more videos
301
Llm
This study finds that a carefully structured team of LLM reviewers can often critique computer architecture papers more deeply than individual human analysts, while still struggling with trust and calibration.
Nishant Aggarwal, Ayushi Dubal, Sreeraj Kannakarankodi, Ian McDougall, Adarsh Mittal, Vishnu Ramadas, Noah Scott, Ranganath Selagamsetty, Weichu Yang, Karthikeyan SankaralingamRead
The study shows that language models often know how subgroups differ but fail to combine that knowledge into accurate population-level answers.
Patrik Wolf, Thomas Kleine Buening, Andreas Krause, Celestine Mendler-DünnerRead
SAR trims language-model updates down to their most useful spectral directions, aiming to preserve reasoning while improving exploration, capability consolidation, and model merging.
Zhilong Zhang, Hongli Yu, Huan-ang Gao, Hanlin Wu, Yuxuan Song, Wei-Ying Ma, Ya-Qin Zhang, Hao ZhouRead
302
Foundation Model
Soofi S is a new open bilingual foundation model for German and English that combines MoE and Mamba-Transformer ideas to deliver strong quality with efficient long-context inference.
The Soofi-Team, :, Benedikt Droste, David Fitzek, Ruben Härle, Lukas Helff, Maximilian Idahl, Alex Jude, Abbas Goher Khan, Maurice Kraus, Timm Ruland, Richard Rutmann, Sebastian Sztwiertnia, Markus Frey, Daniil Gurgurov, Jan Pfister, Tom Röhr, Sebastian von Rohrscheidt, Jörg Bienert, Nicolas Flores-Herr, Simon Gottschalk, Andreas Hotho, Kristian Kersting, Joachim Köhler, Alexander Löser, Wolfgang Nejdl, Simon Ostermann, Jan Plogsties, Patrick Putzky, Mehdi Ali, Michael Fromm, Max LübberingRead
This paper argues that training on the visual form of documents, not just extracted text, can improve foundation models by preserving information like layouts, equations, and figures.
Yiming Zhang, Zhonghan Zhao, Wenwei Zhang, Haiteng Zhao, Tianyang Lin, Yunhua Zhou, Demin Song, Kuikun Liu, Haochen Ye, Haian Huang, Yuzhe Gu, Haijun Lv, Qipeng Guo, Bin Liu, Gaoang Wang, Kai ChenRead
TESSERA v2 shows how to scale Earth-observation foundation models more effectively, revealing that downstream results—not pretraining loss—should guide model selection and that bigger encoders plus distillation can produce compact, highly competitive embeddings.
Zhengpeng Feng, Sadiq Jaffer, Ira Shokar, Jovana Knezevic, Mark Elvers, Clement Atzberger, Robin Young, Aneesh Naik, Niall Robinson, Andrew Blake, David Coomes, Anil Madhavapeddy, Srinivasan KeshavRead
303
Hardware
Chiplet3D uses a thermally informed MILP to arrange 3D chiplets and pins for shorter connections while substantially reducing hotspots.
Shuo Ren, Libo Shen, Yaohui Han, Rongliang Fu, Junying Huang, Bei Yu, Tsung-Yi HoRead
This paper argues that rising memory costs and cheaper open models will reshape the AI industry by making inference economics, hardware vintage, and pricing strategy the real battleground from 2026 to 2030.
Satoshi MatsuokaRead
OpenURMA is the first open implementation of Huawei’s new Unified Bus protocol, showing that a load/store-style design can cut remote-memory latency and boost throughput versus traditional RDMA.
Bojie LiRead
Load more videos
304
Safety
Multilingual LLM judges can score identical content very differently across languages, allowing harmful material in lower-resource languages to slip past safety filters despite impressive benchmark accuracy.
Ej Zhou, Lucas Resck, Zheng Hui, Anna KorhonenRead
LLM agents can invent nonexistent skills that attackers may pre-register as malicious packages, turning ordinary recommendation hallucinations into a supply-chain security threat.
Weifeng Yuan, Wenbo Guo, Feng Dong, Haoyu Wang, Yang LiuRead
This study finds that autonomous coding workflows frequently contain overlooked security problems, especially supply-chain issues and leaked credentials, calling for stronger safeguards at the human-AI collaboration boundary.
A H M Nazmus Sakib, Dipayan Banik, Murtuza JadliwalaRead
305
Benchmark
AgentBeats reframes how we evaluate AI agents by standardizing assessment through agent-to-agent protocols, aiming to make benchmarks more open, reproducible, and fair.
Xiaoyuan Liu, Jianhong Tu, Yuqi Chen, Siyuan Xie, Sihan Ren, Tianneng Shi, Gal Gantar, Evan Sandoval, Donghyun Lee, Daniel Miao, Peter J. Gilbert, Nick Hynes, Mauro Staver, Warren He, David Marn, Andrew Low, Xi Zhang, Elron Bandel, Michal Shmueli-Scheuer, Siva Reddy, Alexandre Drouin, Alexandre Lacoste, Ramayya Krishnan, Elham Tabassi, Yu Su, Victor Barres, Chenguang Wang, Wenbo Guo, Dawn SongRead
PhantomBench is a large benchmark that tests whether language models can recognize when a concept does not actually exist, revealing that even top models often confidently hallucinate.
Haeji Jung, Hila GonenRead
KITScenes Multimodal is a high-fidelity autonomous driving dataset that pairs rich sensors and detailed maps with new benchmarks for mapping, depth, view synthesis, and end-to-end driving.
Richard Schwarzkopf, Fabian Immel, Alexander Blumberg, Jonas Merkert, Nils Rack, Kaiwen Wang, Fabian Konstantinidis, Julian Truetsch, Carlos Fernandez, Annika Bätz, Kevin Rösch, Marlon Steiner, Willi Poh, Yinzhe Shen, Royden Wagner, Felix Hauser, Dominik Strutz, Jaime Villa, Gleb Stepanov, Holger Caesar, Ömer Şahin Taş, Frank Bieder, Jan-Hendrik Pauls, Christoph StillerRead
306
World Model
The paper builds the first real-time multiplayer world model for Rocket League, showing that a large diffusion model can simulate coordinated multi-agent gameplay with surprisingly stable long-horizon behavior.
Anthony Hu, Václav Volhejn, Adrien Ramanana Rahary, Chris Mulder, Aditya Makkar, Amélie Royer, Manu Orsini, Alyx Liao, Adam Jelley, Eloi Alonso, Florian Laurent, Fredrik Norén, James Swingos, Jan Hünermann, Kent Rollins, Lucas Hosseini, Matthieu Le Cauchois, Maxim Peter, Pim de Witte, Tim Brown, Vincent Micheli, Moritz Böhle, Gabriel de Marmiesse, Viktoriia Sharmanska, Lucia Specia, Michael Black, Patrick PérezRead
Kairos proposes a full-stack world model for physical AI that learns from mixed embodied data, remembers over long horizons, and runs efficiently on real hardware.
Kairos Team, Fei Wang, Shan You, Qiming Zhang, Tao Huang, Zuoyi Fu, Zhisheng Zheng, Yunlong Xi, Feng Lv, Xiaoming Wu, Zeyu Liu, Cong Wan, Pu Li, Ruiqing Yang, Xiaoou Li, Wei Wang, Kangkang Zhu, Yuwei Zhang, Shi Fu, Zheng Zhang, Xiaoning Wu, Xuzeng Fan, Dacheng Tao, Xiaogang WangRead
This paper turns a world model into a virtual teleoperation engine, letting humans generate robot training demonstrations without being physically connected to the robot.
Haoyu Zhao, Xingyue Zhao, Hangyu Li, Biao Gong, Kehan Li, Siteng Huang, Xin Li, Deli Zhao, Zhongyu LiRead
Load more videos
307
Reasoning
CORE helps language models get better at reasoning faster by turning comparisons between right and wrong attempts into compact, human-readable strategy notes.
Linas Nasvytis, Simon Jerome Han, Ben Prystawski, Satchel Grant, Noah D. Goodman, Judith E. FanRead
This paper makes Lean theorem proving cheaper and stronger by combining curriculum training, proof-trace filtering, and a novel data augmentation scheme, while also testing a diffusion-style prover.
Joshua Ong Jun Leang, Zheng Zhao, Mihaela Cătălina Stoian, Qiyuan Xu, Haonan Li, Wenda Li, Shay B. Cohen, Eleonora GiunchigliaRead
The paper argues that transformer hidden states can be read directly in the default coordinate basis, with each dimension acting like a binary semantic register whose sign patterns reveal and even control concepts without training a probe.
Varun Reddy NalagatlaRead
308
Graph Learning
AGE improves GraphRAG by learning which graph nodes to mask during self-supervised training, making graph embeddings better aligned with frozen LLMs for graph question answering.
Bao Long Nguyen Huu, Atsushi HashimotoRead
This paper introduces a new way to handle variables inside e-graphs so algebraic rewrites can be done more safely and elegantly.
Philip ZuckerRead
This paper shows that a malicious trainer can secretly implant backdoors in neural networks that are essentially impossible to detect from the model’s weights alone.
Andrej Bogdanov, Alon Rosen, Neekon VafaRead
309
Scientific Ai
SciReasoner is a new scientific AI model that turns molecular and material structures into reasoning-friendly tokens to boost prediction accuracy and make its conclusions easier to interpret.
Chen Tang, Yizhou Wang, Jianyu Wu, Lintao Wang, Shixiang Tang, Pengze Li, Encheng Su, Jun Yao, Jiabei Xiao, Yuqi Shi, Jielan Li, Hongxia Hao, Zhangyang Gao, Fang Wu, Ben Fei, Xiangyu Yue, Pan Tan, Bozitao Zhong, Jinouwen Zhang, Aoran Wang, Yan Lu, Jiaheng Liu, Xinzhu Ma, Liang Hong, Mingyue Zheng, Phil Torr, Bowen Zhou, Wanli Ouyang, Lei BaiRead
DiscoPER is an LLM-based system that autonomously explores data, tests hypotheses statistically, and reflects on its own findings to uncover new scientific patterns in multimodal datasets.
Bingchen Zhao, Sara Beery, Oisin Mac AodhaRead
This paper teaches language models how to 'evolve' solutions across hundreds of optimization problems, turning search experience into reusable skill.
Young-Jun Lee, Seungone Kim, Minki Kang, Alistair Cheong Liang Chuen, Zerui Chen, Seungho Han, Taehee Jung, Dongyeop KangRead
Load more videos
310
Code Generation
This study shows that while cleaner code may not help coding agents finish more tasks, it can make them work more efficiently with fewer tokens and fewer back-and-forth file visits.
Priyansh Trivedi, Olivier SchmittRead
This paper shows that giving code agents simple structural hints from static analysis can make them more predictable, more stable across runs, and slightly better at finding the right code.
Zhihao Lin, Mingyi Zhou, Yizhuo Yang, Li LiRead
This paper teaches coding agents not just to write pull requests, but to review, critique, and revise them in a loop that more closely matches real software development.
Ruoyu Wang, Jierun Chen, Shaowei Wang, Chaofan Tao, Sidi Yang, Yuxin Jiang, Kim-Hui Yap, Lifeng Shang, Xiaohui Li, Haoli BaiRead
311
Embodied Ai
Deform360 is a big real-world dataset that helps robots learn how squishy objects move by combining many camera views, tactile sensing, and dense motion tracking.
Hongyu Li, Wanjia Fu, Xiaoyan Cong, Zekun Li, Binghao Huang, Hanxiao Jiang, Xintong He, Yiqing Liang, Rao Fu, Tao Lu, Srinath Sridhar, Kevin A. Smith, George Konidaris, Yunzhu LiRead
Image2Sim turns ordinary RGB-D image sequences into large-scale interactive navigation simulators, letting embodied AI agents train in realistic neural worlds instead of expensive hand-built environments.
Zihan Wang, Seungjun Lee, Yinghao Xu, Gim Hee LeeRead
This paper introduces a robot control model that keeps a vision-language model’s understanding intact while adding compact future prediction, improving long-horizon manipulation and compositional generalization.
Haoxiang Ma, Junhao Cai, Xiaoxu Xu, Hao Li, Yuyin Yang, Yang Tian, Jiafei Cao, Hongrui Zhu, Zherui Qiu, Zhaxizhuoma, Yuqiang Yang, Jiaqi Peng, Xueyuan Wei, Yangkun Zhu, Jiahao Jiang, Xing Gao, Hanqing Wang, Feng Yuan, Kailin Li, Xueyue Zhu, Tai Wang, Yan Ding, Jiangmiao Pang, Jia Zeng, Jingjing Zhang, Bowen Zhou, Yao Mu, Chunhua Shen, Weinan ZhangRead
312
Robotics
3D HAMSTER makes robot planners output depth-aware 3D waypoints instead of flat 2D paths, helping language-guided manipulation work better in real-world and visually shifting environments.
Dongyoon Hwang, Byungkun Lee, Dongjin Kim, Hyojin Jang, Hoiyeong Jin, Jueun Mun, Minho Park, Hojoon Lee, Hyunseung Kim, Jaegul ChooRead
This paper teaches robot vision-language-action models to remember past experience in a shared latent space, helping them handle longer and more complex manipulation tasks.
Hongyu Qu, Jianzhe Gao, Xiaobin Hu, Shaohuan Yang, Xinlei Yu, Rui Yan, Wenguan Wang, Xiangbo Shu, Shuicheng YanRead
RoboTALES teaches robot policies by using language-guided planning and vision-language feedback to imagine task-aligned futures, improving long-horizon manipulation.
Hanan Gani, Tejal Kulkarni, Madhoolika Chodavarapu, Nicklas Hansen, Manmohan ChandrakerRead
Load more videos
313
Agent
This paper argues that you can build smarter multi-agent systems not by centrally coordinating them, but by creating an economy where agents bid, earn, compete, and evolve into better reasoners.
Zhenting Qi, Huangyuan Su, Ao Qu, Chenyu Wang, Yu Yao, Han Zheng, Kushal Chattopadhyay, Guowei Xu, Zihan Wang, Weirui Ye, Vijay Janapa Reddi, Ju Li, Paul Pu Liang, Himabindu Lakkaraju, Sham Kakade, Yilun DuRead
OpenSkill lets LLM agents teach themselves from open-web resources and self-generated practice tasks, even when no labeled target data or verifier exists.
Zhiling Yan, Dingjie Song, Hanrong Zhang, Wei Liang, Yuxuan Zhang, Yutong Dai, Lifang He, Philip S. Yu, Ran Xu, Xiang Li, Lichao SunRead
This paper introduces Arbor, an AI research agent that keeps a living tree of hypotheses and evidence so it can iteratively plan, test, and improve scientific ideas over long time horizons.
Jiajie Jin, Yuyang Hu, Kai Qiu, Qi Dai, Chong Luo, Guanting Dong, Xiaoxi Li, Tong Zhao, Xiaolong Ma, Gongrui Zhang, Zhirong Wu, Bei Liu, Zhengyuan Yang, Linjie Li, Lijuan Wang, Hongjin Qian, Yutao Zhu, Zhicheng DouRead
314
Rl
This paper turns an LLM from a static grader into a dynamic tutor that makes training prompts progressively harder so reinforcement learning stays informative as the policy improves.
Yujin Kim, Namgyu Ho, Sangmin Hwang, Joonkee Kim, Yongjin Yang, Sangmin Bae, Seungone Kim, Jaehun Jung, Se-Young Yun, Hwanjun SongRead
This paper uses reinforcement learning to dynamically tune particle-collider trigger thresholds, improving real-time event selection on both simulated and real LHC data.
Zixin Ding, Shaghayegh Emami, Giovanna Salvi, Cecilia Tosciri, Abhijith Gandrakota, Jennifer Ngadiuba, Nhan Tran, Christian Herwig, David W. Miller, Yuxin ChenRead
This paper introduces a more stable asynchronous reinforcement learning method for training agentic LLMs, using single-rollout updates to better handle long-horizon tasks like coding and reasoning.
Zhenyu Hou, Yujiang Li, Jie Tang, Yuxiao DongRead
315
Multimodal
MV-Forcing lets a diffusion model generate long, consistent videos from multiple viewpoints by using 3D geometry to guide each new view and time step.
Gal Fiebelman, Hadar Averbuch-Elor, Sagie BenaimRead
PixelRAG shows that for web-based retrieval-augmented generation, using screenshots instead of extracted text can improve both answer quality and efficiency across several QA and agent tasks.
Yichuan Wang, Zhifei Li, Zirui Wang, Paul Teiletche, Lesheng Jin, Matei Zaharia, Joseph E. Gonzalez, Sewon MinRead
This paper shows how a single multimodal model can handle detection, segmentation, depth, OCR, and more by turning all vision tasks into text-and-image generation.
Xiaoyang Han, Jianhua Li, Kewang Deng, Zukai Chen, Xuanke Shi, Sihan Wang, Boxuan Li, Linyan Wang, Siyi Xie, Xin You, Jinsheng Quan, Zhongang Cai, Haiwen Diao, Ziwei Liu, Lei Yang, Dahua Lin, Quan WangRead
Load more videos
316
Cv
PixWorld unifies 3D scene reconstruction and generation in pixel space, adding geometry-aware supervision to improve both visual quality and 3D fidelity.
Sensen Gao, Zhaoqing Wang, Qihang Cao, Dongdong Yu, Changhu Wang, Jia-Wang BianRead
SCAIL-2 is a new end-to-end character animation system that learns motion transfer directly from video, using synthetic training data and preference tuning to produce more faithful animations.
Wenhao Yan, Fengjia Guo, Zhuoyi Yang, Jie TangRead
VideoMDM teaches a 3D human motion generator using only 2D keypoints from videos, getting surprisingly close to fully 3D-supervised models without needing 3D ground truth.
Amir Mann, Gal Michael Harari, Merav Keidar, Or LitanyRead
317
Efficiency
DSpark speeds up LLM inference by generating draft tokens in a more structured way and verifying them adaptively, delivering major serving throughput gains in real production traffic.
Xin Cheng, Xingkai Yu, Chenze Shao, Jiashi Li, Yunfan Xiong, Yi Qian, Jiaqi Zhu, Shirong Ma, Xiaokang Zhang, Jiasheng Ye, Qinyu Chen, Chengqi Deng, Jiping Yu, Damai Dai, Zhengyan Zhang, Yixuan Wei, Yixuan Tan, Wenkai Yang, Runxin Xu, Yu Wu, Zhean Xu, Xuanyu Wang, Muyang Chen, Rui Tian, Xiao Bi, Zhewen Hao, Shaoyuan Chen, Huanqi Cao, Wentao Zhang, Anyi Xu, Huishuai Zhang, Dongyan Zhao, Wenfeng LiangRead
Gemma 4 is Google’s new open-weight multimodal model family, pushing stronger reasoning, faster inference, and better long-context performance across text, image, and audio tasks.
Gemma Team, Sherif El Abd, Vaibhav Aggarwal, Robin Algayres, Alek Andreev, Olivier Bachem, Ian Ballantyne, Cormac Brick, Victor Cărbune, Michelle Casbon, Mayank Chaturvedi, Victor Cotruta, Alice Coucke, Phil Culliton, Robert Dadashi, Lucas Dixon, Mohamed Elhawaty, Utku Evci, Clément Farabet, Johan Ferret, Filippo Galgani, Sertan Girgin, Jean-Bastien Grill, Maarten Grootendorst, Jiaxian Guo, Cassidy Hardin, Yanzhang He, Steven M. Hernandez, Omri Homburger, Léonard Hussenot, Juyeong Ji, Armand Joulin, Aishwarya Kamath, Parnian Kassraie, Olivier Lacombe, Preethi Lahoti, Gaël Liu, Gus Martins, Luciano Martins, Tatiana Matejovicova, Ramona Merhej, Nikola Momchev, Sneha Mondal, Ryan Mullins, Sindhu Raghuram Panyam, Shreya Pathak, Sarah Perrin, André Susano Pinto, Etienne Pot, Angéline Pouget, Alexandre Ramé, Sabela Ramos, Douglas Reid, David Rim, Morgane Rivière, Karsten Roth, Louis Rouillard, Omar Sanseviero, Pier Giuseppe Sessa, Shane Settle, Danila Sinopalnikov, Sara Smoot, Pi...Read
This paper shows that LLM agents with tools can secretly coordinate through steganographic channels that evade monitoring, raising serious new safety risks for multi-agent systems.
Jimmy Laurence Rippin, Simon C. Marshall, David Demitri Africa, Christian Schroeder de WittRead
318
Diffusion
This paper shows how to distill powerful modern diffusion models into smaller one-step generators even when the teacher and student use incompatible latent spaces.
Anh Nguyen, Ngan Nguyen, Duc Vu, Trung Dao, Viet Nguyen, Quan Dao, Kien Nguyen, Chi Tran, Phong Nguyen, Khoi Nguyen, Cuong Pham, Dimitris Metaxas, Vishal M. Patel, Anh TranRead
Flex-Forcing lets one video diffusion model switch between fast autoregressive generation and high-quality bidirectional generation, aiming to get the best of both worlds.
Xinyin Ma, Julius Berner, Chao Liu, Arash Vahdat, Weili Nie, Xinchao WangRead
MrFlow speeds up text-to-image diffusion by generating a coarse image first and then refining it in stages, reaching large speedups without extra training.
Xingyu Zheng, Xianglong Liu, Yifu Ding, Weilun Feng, Junqing Lin, Jinyang Guo, Haotong QinRead
Load more videos
319
Generative Model
GEAR lets an image generator and its tokenizer learn together, making autoregressive image synthesis train faster and produce better visual representations.
Bin Lin, Zheyuan Liu, Chenguo Lin, Sixiang Chen, Yunyang Ge, Yunlong Lin, Jianwei Zhang, Miles Yang, Zhao Zhong, Liefeng Bo, Li YuanRead
This paper makes pixel-by-pixel image generators train more like they infer, delivering a new state of the art without needing a separate tokenizer.
Jiayi Xu, Di He, Guolin KeRead
PolyFlow turns mesh generation into a parallel continuous process by embedding topology into a learnable state space, making artist-style 3D mesh synthesis faster and more controllable.
Chunshi Wang, Haohan Weng, Junliang Ye, Biwen Lei, Yang Li, Zibo Zhao, Zeqiang Lai, Kaiyi Zhang, Yunhan Yang, Zhuo Chen, Chunchao Guo, Yawei LuoRead
320
Self Supervised
This paper explains why contrastive learning on images works by proving that the best features are often sinusoidal filters with partial whitening, and showing real models learn the same pattern.
Antonio Torralba, Yair WeissRead
This paper shows that even strong medical LLMs can be easily tricked by misleading context, and introduces MedMisBench to measure how well they stay correct under pressure.
Hongjian Zhou, Xinyu Zou, Jinge Wu, Sean Wu, Junchi Yu, Bradley Max Segal, Tobias Erich Niebuhr, Sara Amro, Michael Petrus, Sheikh Momin, Alexandra M. Cardoso Pinto, Rachel Niesen, Laura Sophie Wegner, Dhruv Darji, Jung Moses Koo, Joshua Fieggen, Kapil Narain, Mingde Zeng, Lei Clifton, Linda Shapiro, Fenglin Liu, David A. CliftonRead
SiamJEPA shows that giving JEPA a Siamese student encoder can act like a useful regularizer, improving early learning and representation quality in self-supervised vision models.
Makoto YamadaRead
321
Benchmark
EdgeBench is a large real-world benchmark that reveals a surprising scaling law for how agents improve from long-horizon interaction after deployment.
Deyao Zhu, Xin Zhou, Shengling Qin, Xuekai Zhu, Hangliang Ding, Shu Zhong, Zixin Wen, Zhonglin Xie, Chenhui Gou, Linxuan Ren, Yueyang Wang, Junfeng Zhong, Rui Liu, Tian Gao, Yangguang Lin, Jingyuan Zhang, Maojia Song, Xuan Qi, Jinhong Wu, Chenyang Zhang, Yinzhu Piao, Ziru Niu, Hongbin Lin, Lingxiang Meng, Peng Tang, Chengyao Tang, Shanyu Wu, Huanyu Zheng, Yu Liu, Liya Zhu, He Wang, Ming Ding, Ziyu Wan, Hao Liu, Sibo Wang, Haotian Zhu, Xintian Zhang, Nan Chai, Yipeng Liu, Panhao Lai, Sihang Yuan, Zixin Su, Ge Zhang, Wangchunshu Zhou, Yantao Du, Wenhao Huang, Guang ShiRead
OmniOpt organizes the chaotic world of modern optimizers into a unified taxonomy and benchmark so researchers can compare methods more systematically.
Siyuan Li, Jiabao Pan, Yumou Liu, Zhuoli Ouyang, Xin Jin, Xinglong Xu, Jingxuan Wei, Shengye Pang, Jintao Che, Xuanhe Zhou, Conghui He, Cheng TanRead
This paper presents Vera, a scalable system for discovering and testing safety risks in LLM agents, and releases a new benchmark showing that even state-of-the-art agent frameworks can be surprisingly vulnerable.
Yunhao Feng, Ruixiao Lin, Ming Wen, Qinqin He, Yanming Guo, Yifan Ding, Yutao Wu, Jialuo Chen, Zhuoer Xu, Xiaohu Du, Jianan Ma, Zixing Chen, Xingjun Ma, Yunhao Chen, Xinhao DengRead
Load more videos
322
Safety
This paper shows that AI agents can be tricked not just by malicious instructions, but by seemingly trustworthy data that secretly steers them into unsafe actions, exposing real vulnerabilities in popular agents.
Woohyuk Choi, Juhee Kim, Taehyun Kang, Jihyeon Jeong, Luyi Xing, Byoungyoung LeeRead
This paper compares several modern linear-attention designs under one framework and finds that a simple cross-layer value-routing trick can modestly improve performance.
Tommaso Cerruti, Tim Rieder, George Rowlands, Lingfeng Jin, Imanol SchlagRead
This paper builds a single model that can understand and generate both audio and text, while trying hard not to lose the reasoning and alignment skills of its text-only LLM core.
Zhifeng Kong, Sang-gil Lee, Jaehyeon Kim, Boxin Wang, Zihan Liu, Sungwon Kim, Yang Chen, Arushi Goel, Rajarshi Roy, Wenliang Dai, Zhuolin Yang, Yangyi Chen, Dongfu Jiang, Sreyan Ghosh, Tuomas Rintamaki, Andrew Tao, Jonathan Raiman, Mohammad Shoeybi, Bryan Catanzaro, Wei PingRead
323
Llm
This paper shows that LLMs can look like humans in risk-taking tasks while actually using very different decision mechanisms underneath.
Chensong Huang, Changyu Chen, Chenwei Lin, Hanjia Lyu, Xian Xu, Jiebo LuoRead
The paper argues that beyond a modest number of samples, test-time scaling for language models stops helping and can even hurt because the real bottleneck is picking the right answer, not generating more candidates.
Yong Yi Bay, Kathleen A. YearickRead
dOPSD teaches diffusion language models using their own unfolding predictions, boosting reasoning without needing ground-truth hints.
Phuong Tuan Dat, Qi Li, Xinchao WangRead
324
Llm
This paper lets LLM distillation work across different tokenizers, making it easier and cheaper to transfer knowledge between otherwise incompatible model families.
Yifan Niu, Han Xiao, Dongyi Liu, Zelong Wang, Dihong Gong, Yasheng Wang, Jia LiRead
This paper uses reinforcement learning to automatically decide what training data an LLM should see next, making pre-training faster and improving final model quality.
Chenhao Dang, Jing Ma, Mingjie LiaoRead
The paper shows that LLMs can be stubborn annotators: once they form a confident answer, extra prompting often fails to change their mind, and what matters most is whether the model’s internal concept matches the task definition.
Etienne Casanova, Rafal Kocielnik, R. Michael AlvarezRead
Load more videos
325
Llm
This paper shows how to turn an autoregressive language model into a diffusion-style language model much more efficiently by training it on its own generated trajectories while distilling knowledge from the original model.
Xingyu Su, Jacob Helwig, Shubham Parashar, Atharv Chagi, Lakshmi Jotsna, Degui Zhi, James Caverlee, Dileep Kalathil, Shuiwang JiRead
The paper shows that LLMs struggle not just with long inputs, but especially with long inputs packed with lots of new information, sharply reducing their effective context window.
Giovanni Dettori, Matteo Boffa, Danilo Giordano, Idilio Drago, Marco MelliaRead
This paper shows that on-policy distillation for language models follows its own unique path in parameter space, with updates quickly locking into a low-dimensional subspace that still preserves performance.
Zhennan Shen, Yanshu Li, Qingyu Yin, Chak Tou Leong, Zhilin Wang, Yanxu Chen, Rongduo Han, Sunbowen Lee, Yi R. FungRead
326
Agent
This paper turns LLMs into fine-grained verifiers that can score, rank, and guide agentic tasks more effectively than ordinary judges, improving benchmark performance and even training efficiency.
Jacky Kwok, Shulu Li, Pranav Atreya, Yuejiang Liu, Yixing Jiang, Chelsea Finn, Marco Pavone, Ion Stoica, Azalia MirhoseiniRead
This paper shows that letting multi-agent reasoning systems pass along partial thoughts as they are generated can make them both faster and, surprisingly, more accurate.
Zhen Yang, Xiaogang Xu, Wen Wang, Cong Chen, Xander Xu, Ying-Cong ChenRead
This paper introduces a self-improving agent framework where the evaluator changes too, letting writers, coders, and reviewers co-evolve in an ongoing Red Queen race.
Alex Iacob, Andrej Jovanović, William F. Shen, Daniel Burkhardt, Meghdad Kurmanji, Nurbek Tastan, Lorenzo Sani, Niccolò Alberto Elia Venanzi, Ambroise Odonnat, Zeyu Cao, Bill Marino, Xinchi Qiu, Nicholas D. LaneRead
327
Llm
HiLS Attention is a new way to make LLMs handle much longer contexts by learning which chunks to attend to, cutting compute while keeping performance strong.
Xiang Hu, Xinyu Wei, Hao Gu, Minshen Zhang, Tian Liang, Huayang Li, Lei Zhu, Yan Wang, Sirui Han, Yushi Bai, Kewei Tu, Haitao Mi, Leo LiangRead
MOPD is a new way to merge multiple specialized LLM skills by distilling several RL-trained teachers into one student using the student’s own rollouts, and it appears to work well at frontier scale.
Wenhan Ma, Jianyu Wei, Liang Zhao, Hailin Zhang, Bangjun Xiao, Lei Li, Qibin Yang, Bofei Gao, Yudong Wang, Rang Li, Jinhao Dong, Zhifang Sui, Fuli LuoRead
This paper introduces a language model that can switch between autoregressive, diffusion, and self-speculative decoding to trade off speed and quality, achieving major throughput gains over current open-source models.
Yonggan Fu, Lexington Whalen, Abhinav Garg, Chengyue Wu, Maksim Khadkevich, Nicolai Oswald, Enze Xie, Daniel Egert, Sharath Turuvekere Sreenivas, Shizhe Diao, Chenhan Yu, Ye Yu, Weijia Chen, Sajad Norouzi, Jingyu Liu, Shiyi Lan, Ligeng Zhu, Jin Wang, Jindong Jiang, Morteza Mardani, Mehran Maghoumi, Song Han, Ante Jukić, Nima Tajbakhsh, Jan Kautz, Pavlo MolchanovRead
Load more videos
328
Generative Model
DanceOPD teaches a single image generator to juggle text-to-image creation, editing, and guidance signals by learning from its own rollout states, helping one model do many generation tasks without the usual capability conflicts.
Wei Zhou, Xiongwei Zhu, Zelin Xu, Bo Dong, Lixue Gong, Yongyuan Liang, Meng Chu, Leigang Qu, Lingdong Kong, Wei Liu, Tat-Seng ChuaRead
This paper makes generative models faster by training them in perceptual feature space, letting them produce good samples in just a few steps instead of dozens.
Chuyang Zhao, Yifei Song, Hongfa Wang, Jianlong Yuan, Yuan Zhang, Siming Fu, Zhineng Chen, Huilin Deng, Haoyang Huang, Nan DuanRead
This paper shows how to train strong one-step image generators by matching feature distributions from multiple frozen encoders, achieving state-of-the-art quality and even compressing a four-step model into a single step.
Lan Feng, Wuyang Li, Eloi Zablocki, Matthieu Cord, Alexandre AlahiRead
329
Llm
CausalMix treats LLM data mixing like a causal problem, using learned treatment effects to choose better training mixtures that scale to larger models and data pools.
Zinan Tang, Yukun Zhang, Shaomian Zheng, Zhuoshi Pan, Qizhi Pei, Dingnan Jin, Jun Zhou, Yujun Wang, Biqing HuangRead
This paper shows that a diffusion language model’s token order is not truly parallel or sequential, and that how you measure it can completely change the story.
Ali Asaria, Tony Salomone, Deep GandhiRead
This paper argues that language models should spend more capacity in earlier layers and less in later ones, and shows that a tapered width schedule can improve performance without increasing cost.
Reza Bayat, Ali Behrouz, Aaron CourvilleRead
330
Self Supervised
Autodata turns an AI agent into a data scientist that generates better training data, and even learns how to improve its own data-making process.
Ilia Kulikov, Chenxi Whitehouse, Tianhao Wu, Yixin Nie, Swarnadeep Saha, Eryk Helenowski, Weizhe Yuan, Olga Golovneva, Jack Lanchantin, Yoram Bachrach, Jakob Foerster, Xian Li, Han Fang, Sainbayar Sukhbaatar, Jason WestonRead
This paper shows how world models can serve as practical stand-ins for expensive robot testing, and introduces a benchmark and model designed to make that evaluation much more reliable.
GigaWorld Team, Angyuan Ma, Boyuan Wang, Bohan Li, Chaojun Ni, Guo Li, Guan Huang, Guosheng Zhao, Hao Li, Hengtao Li, Jingyu Liu, Jiwen Lu, Qiuping Deng, Tingdong Yu, Xuancheng Xu, Xinyu Zhou, Xiuwei Xu, Xinze Chen, Xiaofeng Wang, Xiaoyu Tian, Yang Wang, Yifan Chang, Yukun Zhou, Yun Ye, Zhenyu Wu, Zhanqian Wu, Zheng ZhuRead
This paper teaches vision models to pay attention to boundaries so they can learn richer spatial structure and improve tasks like depth estimation.
Zelin Fu, Bin Tan, Changjiang Sun, Shaohui Liu, Kecheng Zheng, Yinghao Xu, Xing Zhu, Yujun Shen, Nan XueRead
Load more videos
331
Rl
This paper finds that, for RL post-training of LLMs, most of the benefit can come from training just one transformer layer—often a middle layer—rather than updating the whole model.
Zijian Zhang, Rizhen Hu, Athanasios Glentis, Dawei Li, Chung-Yiu Yau, Hongzhou Lin, Mingyi HongRead
This paper argues that LLM RL should optimize what actually improves inference-time behavior, not just training-time policy updates, and introduces a method to make that happen more reliably.
Jing Liang, Hongyao Tang, Yi Ma, Yancheng He, Weixun Wang, Xiaoyang Li, Ju Huang, Wenbo Su, Jinyi Liu, Yan Zheng, Jianye Hao, Bo ZhengRead
This paper proposes an automated training curriculum for reasoning models that picks the next domain based not just on where the model is learning fastest, but on where an update will help other domains the most.
Yongjin Yang, Jiarui Liu, Yinghui He, Lechen Zhang, Bernhard Schölkopf, Zhijing JinRead
332
Optimizer
This paper builds a 2,048-spin acoustic Ising machine that uses sound waves in solid-state delay lines to tackle hard optimization problems like MAX-CUT and Sudoku.
Venkatesh Vadde, Roman Ovcharov, Victor H. González, Roman Khymyn, Artem Litvinenko, Johan ÅkermanRead
This paper shows that AI peer reviewers can be gamed by changing only how a paper is presented, not what it actually says or proves.
Xu Yang, Zhizhou Sha, Junbo Li, Jian Yu, Yifan Sun, Matthew Zhao, Jinrui Fang, Xinyue Guo, Yining Wu, Xu Hu, Yifu Luo, Qiang Liu, Zhangyang WangRead
This paper explains why the Muon optimizer can beat Adam in large language model training by showing it makes smaller curvature-induced mistakes, especially in imbalanced and high-curvature settings.
Shuche Wang, Fengzhuo Zhang, Jiaxiang Li, Dirk Bergemann, Zhuoran YangRead
333
Multimodal
This paper creates a large open benchmark for testing how to build better vision-language training data, and shows that mixing the right kinds of data can matter more than filtering.
Matteo Farina, Vishaal Udandarao, Thao Nguyen, Selim Kuzucu, Maximilian Böther, Andreas Hochlehnert, Adhiraj Ghosh, Marianna Nezhurina, Karsten Roth, Joschka Struber, Yuhui Zhang, Sebastian Dziadzio, Elaine Sui, Soumya Jahagirdar, Dhruba Ghosh, Hasan Hammoud, Thomas De Min, Simone Caldarella, Jehanzeb Mirza, Sedrick Keh, Mehdi Cherti, Hilde Kuehne, Bernt Schiele, Serena Yeung-Levy, Muhammad Ferjad Naeem, Federico Tombari, Ana Klimovic, Elisa Ricci, Matthias Bethge, Sewoong Oh, Ameya Prabhu, Alessio Tonioni, Jenia Jitsev, Massimiliano Mancini, Ludwig Schmidt, Nikhil ParthasarathyRead
This paper shows that grammar constraints meant to make code generation safer can actually be exploited to jailbreak LLMs into producing malicious code, and it introduces a defense that resists this attack.
Yitong Zhang, Shiteng Lu, Jia LiRead
This paper gives a new theory for measuring how architecturally complex a neural network is and uses it to generate thousands of previously unexplored designs.
Nicholas J. Cooper, François G. Meyer, Michael L. Roberts, Carlos Zapata-Carratalá, Lijun Chen, Danna GurariRead
Load more videos
334
Transformer
This paper turns transformer attention heads into small executable Python programs, offering a new way to explain and partially replace opaque neural behavior with human-readable code.
Amiri Hayes, Belinda Li, Jacob AndreasRead
This paper shows that GPU Rowhammer can do more than break models: it can let an attacker steal data, tamper with GPU code, and even potentially break out to root on the host machine.
Chris S. Lin, Yuqin Yan, Guozhen Ding, Joyce Qu, Joseph Zhu, David Lie, Gururaj SaileshwarRead
This paper says robots should first learn how to move from cheap unlabeled experience, and only then learn what humans want them to do from scarce expert demonstrations.
Junhao Shi, Siyin Wang, Xiaopeng Yu, Li Ji, Jingjing Gong, Xipeng QiuRead
335
Benchmark
Read
Read
This paper asks whether fixing a model’s text knowledge also fixes its image generation, and finds that the answer is usually no unless the edit is made more reasoning-aware.
Xin Gao, Cheng Yang, Chufan Shi, Taylor Berg-KirkpatrickRead
OSWorld 2.0 is a tougher real-world benchmark for computer-use agents, showing that even top systems still struggle to complete long, stateful tasks reliably.
Mengqi Yuan, Zilong Zhou, Xinzhuang Xiong, Weiming Wu, Jiayang Sun, Jiamin Song, Kaiqian Cui, Bowen Wang, Haoyuan Wu, Yitong Li, Dunjie Lu, Haikong Lu, Qi Zhen, Xinyuan Wang, Jiaqi Deng, Yuhao Yang, Cheng Chen, Boyuan Zheng, Alex Su, Xiao Yu, Hao Zou, Saaket Agashe, Xing Han Lu, Manpreet Kaur, Zhengyang Qi, Vincent Sunn Chen, Frederic Sala, Dayiheng Liu, Junyang Lin, Zhou Yu, Yu Su, Siva Reddy, Xin Eric Wang, Peng Qi, Tianbao Xie, Tao YuRead
This paper shows that LLM judges can be badly overconfident about scientific novelty, and introduces a benchmark revealing that they often call model-generated research questions more novel than domain experts do.
Soumitra Sinhahajari, Navonil Majumder, Soujanya PoriaRead
336
Code Generation
The paper finds that engineers write significantly more pull requests in weeks when they use GitHub Copilot heavily, suggesting the coding assistant can boost developer efficiency even after controlling for effort and confounders.
Alex Heilman, Alex Kyllo, Emerson Murphy-HillRead
This paper shows that coding agents work better when they can 'see' a repository’s structure, and that combining visual graphs with text can cut token use while preserving or improving bug-fixing performance.
Dongjian Ma, Silin Chen, Yufei Yang, Yulin Shi, Yanfu yan, Xiaodong GuRead
03
To Run or Not to Run: Analyzing the Cost-Effectiveness of Code Execution in LLM-Based Program Repair
This paper asks whether running tests during LLM code repair is really worth it, and finds that agents often pay a big execution cost for only modest gains.
Zhihao Lin, Junhua Zhu, Mingyi Zhou, Xin Wang, Zhensu Sun, Renyu Yang, David Lo, Li LiRead
Load more videos
337
Cv
Goku introduces a million-scale dataset and benchmark for instruction-based video editing, plus a model that better handles both appearance changes and structural motion edits.
Sen Liang, Cong Wang, Zhentao Yu, Fengbin Guan, Zhengguang Zhou, Teng Hu, Youliang Zhang, Yuan Zhou, Xin Li, Qinglin Lu, Zhibo ChenRead
This paper turns dense cell segmentation from hundreds of clicks into one click per cell type by chaining prompts through SAM's built-in feature structure.
Sanghyun Jo, Seo Jin Lee, Seohyung Hong, Yoorim Gang, Hyeongsub Kim, Hyungseok Seo, Kyungsu KimRead
This paper turns 3D scenes into object-level tokens instead of raw point clouds or Gaussians, enabling reconstruction, segmentation, editing, and retrieval from unposed images.
Mijin Yoo, In Cho, Subin Jeon, Jiwoo Lee, Eunbyung Park, Seon Joo KimRead
338
Robotics
This paper shows how a small simulation-trained corrective policy can make vision-language-action robot policies much more reliable in the real world without extra robot training.
Kinam Kim, Namiko Saito, Heecheol Kim, Katsushi Ikeuchi, Jaegul Choo, Yasuyuki MatsushitaRead
This work shows that teaching a dexterous robot to 'play' first can make it far better at precise real-world assembly tasks like tight insertions and screwing.
Tyler Ga Wei Lum, Kushal Kedia, C. Karen Liu, Jeannette BohgRead
03
Qwen-RobotManip Technical Report: Alignment Unlocks Scale for Robotic Manipulation Foundation Models
Qwen-RobotManip shows that aligning diverse robot and human demonstration data at scale can unlock stronger generalization for vision-language-action robots across many platforms and real-world settings.
Haoqi Yuan, Zhixuan Liang, Anzhe Chen, Ye Wang, Haoyang Li, Pei Lin, Yiyang Huang, Zixing Lei, Tong Zhang, Jiazhao Zhang, Jie Zhang, Jingyang Fan, Gengze Zhou, Qihang Peng, Chenxu Lv, Xiaoyue Chen, An Yang, Fei Huang, Junyang Lin, Dayiheng Liu, Jingren Zhou, Chenfei Wu, Xiong-Hui ChenRead
339
Scientific Ai
An AI system helped discover and formally verify a new proof of a decade-old quantum optimization conjecture, showing how LLMs plus theorem provers can tackle open math and physics problems.
Uri Kol, Maor Ben-Shahar, Kfir Sulimany, Dirk EnglundRead
RaysUp is a fast, lightweight way to turn low-resolution vision foundation model features into sharp, high-resolution maps using geometry-aware ray representations.
Yuchuan Ding, Linfei Li, Lin Zhang, Ying ShenRead
This paper shows that AI app platforms like Hugging Face can hide serious security flaws, including leaked credentials, code-execution bugs, and even backdoors across hundreds of thousands of apps.
Yacong Gu, Lingyun Ying, Zidong Zhang, Yingyuan Pu, Xiaoxue Huang, Jiawei Zhou, Wenjie Zhu, Donghong Sun, Haixin DuanRead
Load more videos
340
Attention
This paper makes linear attention smarter by dynamically deciding how to merge memory states, aiming to keep long-context models efficient without losing important information.
Xin Wang, Hui Shen, Boyuan Zheng, Xueshen Liu, Minkyoung Cho, Zhongwei Wan, Zesen Zhao, Zhuoqing Mao, Shen Yan, Mi ZhangRead
HydraHead rethinks hybrid attention by assigning full attention only to the most retrieval-critical heads, improving long-context performance while keeping compute efficient.
Zhentao Tan, Wei Chen, Jingyi Shen, Yao Liu, Xu Shen, Yue Wu, Jieping YeRead
This paper proposes LOCOS, a new way to find attention heads that retrieve meaning rather than just copied words, and shows those heads are crucial for long-context question answering.
Aryo Pradipta Gema, Beatrice Alex, Pasquale MinerviniRead
341
World Model
This paper shows that hallucinations in world models are not random: they happen where training data is sparse, and the authors use that insight to detect, reduce, and fine-tune around them.
Nicklas Hansen, Xiaolong WangRead
Orca is a general world model that learns a shared latent representation from video, language, and other signals so it can better predict, describe, and act in the world.
Yihao Wang, Yuheng Ji, Mingyu Cao, Yanqing Shen, Runze Xiao, Huaihai Lyu, Senwei Xie, Euan Liu, Klara Tian, Tianfeng Long, Yichi Zhang, Zhengliang Cai, Ruike Chen, Jifan Zhao, Ruochuan Shi, Zihan Tang, Jing Lyu, Wenxing Tan, Ningbo Zhang, Yangtao Hu, Yuming Gao, Xiansheng Chen, Junkai Zhao, Congsheng Xu, Boan Zhu, Ziqi Wang, Yupu Feng, Qiongqiong Zhang, Yingli Zhao, Yulong Ao, Shaoxuan Xie, You Liu, Guocai Yao, Leiduo Zhang, Xiaodan Liu, Yunyan Zhang, Yance Jiao, Xinyan Yang, Jiaxing Wei, Xu Liu, Tengfei Pan, Shaokai Nie, Chunlei Men, Sen Cui, Xiaojie Jin, Hongyang Li, Jianlan Luo, Yao Mu, Yunchao Wei, Jun Yan, Hang Zhao, Xiaolong Zheng, Jiaming Li, Yonghua Lin, Tiejun Huang, Zhongyuan Wang, Pengwei WangRead
WEAVER is a faster, longer-horizon world model for robot manipulation that can help robots predict outcomes, improve policies, and plan actions more effectively in the real world.
Arnav Kumar Jain, Yilin Wu, Jesse Farebrother, Gokul Swamy, Andrea BajcsyRead
342
Reasoning
This paper argues that some reasoning problems can be solved by search but still cannot be learned as a neat left-to-right chain of thought—what models can distill is often verification and memorization, not the search itself.
Harsh PatelRead
This paper shows that large reasoning models can often solve problems yet still be surprisingly bad at judging whether a step-by-step explanation is actually valid.
Mingzhong Sun, Teresa Yeo, Armando Solar-Lezama, Tan Zhi-XuanRead
MaxProof uses a population of proof candidates plus generator-verifier-ranker loops to push an AI system to gold-medal-level performance on hard math olympiad proofs.
Jiacheng Chen, Xinyu Zhang, Shunkai Zhang, Yanmohan Wang, Lin Li, Tiancheng Qin, Qin Wang, Zhengmao Zhu, Tianle Li, Jingyang Li, Zehan Li, Binyang Jiang, Jin Zhu, Han Ding, Fei Yu, Chenyu Du, Zijian Song, Jiayuan Song, Zhi Zhang, Yunan Huang, Weiyu Cheng, Pengyu Zhao, Yu ChengRead
Load more videos
343
Foundation Model
Program-as-Weights lets a model compile a natural-language task description into a small reusable neural program that can run locally and cheaply instead of calling a giant LLM every time.
Wentao Zhang, Liliana Hotsko, Woojeong Kim, Pengyu Nie, Stuart Shieber, Yuntian DengRead
Seed2.0 is a large foundation model series aimed at improving real-world reasoning, instruction following, and visual understanding, presented through a model card focused on practical deployment and evaluation.
Bytedance SeedRead
This paper argues that transformers should split remembering from predicting, and that doing so can make language models train more efficiently and work better downstream.
Giovanni Monea, Nathan Godey, Kianté Brantley, Yoav ArtziRead
344
Graph Learning
This paper trains AI to reason like a graph-builder, turning scientific questions into traceable chains of concepts that help generate better hypotheses for materials discovery.
Subhadeep Pal, Shashwat Sourav, Tirthankar Ghosal, Markus J. BuehlerRead
SciAtlas builds a massive cross-disciplinary knowledge graph for science and pairs it with graph-based retrieval to help AI agents do better literature search, synthesis, and research planning.
Shuofei Qiao, Yunxiang Wei, Jiazheng Fan, Bin Wu, Busheng Zhang, Mengru Wang, Yuqi Zhu, Ningyu Zhang, Keyan Ding, Qiang Zhang, Huajun ChenRead
This paper shows that the most salient graph tokens in graph language models are not actually the ones carrying the important graph information, revealing a hidden mismatch in how these models represent graphs.
Ding Zhang, Runtao Zhou, Wenqing Zheng, Rizal Fathony, Bayan Bruss, Chirag AgarwalRead
345
Embodied Ai
This paper teaches robots to infer how their body and environment work from a short self-generated experience window, so they can adapt to new camera angles or robot setups without retraining.
Siyin Wang, Junhao Shi, Senyu Fei, Zhaoyang Fu, Li Ji, Jingjing Gong, Xipeng QiuRead
Vesta is a robot brain that tries to do localization, navigation, spatial reasoning, and long-horizon planning in one generalist model instead of juggling multiple specialist systems.
Johan Bjorck, Zhiqi Li, Yunze Man, Jing Wang, An-Chieh Cheng, Sifei Liu, Shihao Wang, Zhiding Yu, Abhishek Badki, Stan Birchfield, Valts Blukis, Yevgen Chebotar, Siyi Chen, Sicong Leng, Yu-Cheng Chou, Tianli Ding, Boyi Li, Zhengyi Luo, Hang Su, Jonathan Tremblay, Tingwu Wang, Bowen Wen, Jimmy Wu, Xianghui Xie, Hanrong Ye, Hongxu Yin, K. R. Zentner, Liangyan Gui, Yu-Xiong Wang, Yuke Zhu, Linxi "Jim" Fan, Jan KautzRead
WorldLines tests whether embodied agents can remember and reason over long-running household interactions, and pairs the benchmark with a memory system designed to keep track of changing world states.
Yehang Zhang, Jianchong Su, Haojian Huang, Yifan Chang, Tianhao Zhou, Xinli Xu, Yingjie Xu, Yinchuan Li, Zexi Li, Ying-Cong ChenRead
Load more videos
346
Diffusion
SharpMoE improves diffusion model efficiency and image quality by routing computation to the most important tokens using cleaner saliency signals instead of noisy latents.
Haoyou Deng, Keyu Yan, Chaojie Mao, Xiang Wang, Yu Liu, Changxin Gao, Nong SangRead
This paper improves discrete diffusion image generation by letting the model revise earlier token choices and by training more effectively with large vocabularies, boosting both quality and efficiency.
Shufan Li, Greg Heinrich, Hanrong Ye, Yonggan Fu, Aditya Grover, Jan Kautz, Pavlo MolchanovRead
347
Agent
PANDO makes multimodal web agents cheaper over time by distilling useful skills online, cutting token use while improving task success on VisualWebArena.
Yubo Li, Yidi Miao, Yuntian Shen, Yuxin LiuRead
PhoneWorld turns real mobile app traces into scalable training environments, giving phone agents a new way to learn and be evaluated across many apps instead of one benchmark at a time.
Zhengyang Tang, Yuxuan Liu, Xin Lai, Junyi Li, Pengyuan Lyu, Jason, Yiduo Guo, Zhengyao Fang, Yang Ding, Yi Zhang, Weinong Wang, Huawen Shen, Xingran Zhou, Liang Wu, Fei Tang, Sunqi Fan, Shangpin Peng, Zheng Ruan, Anran Zhang, Benyou Wang, Rui Yan, Ji-Rong Wen, Chengquan Zhang, Han HuRead
MiniMax-M2 is a huge but sparsely activated language model built for agents, combining verifiable agent data, specialized RL, and self-improving training workflows to push real-world task performance.
MiniMax, :, Aili Chen, Aonian Li, Baichuan Zhou, Bangwei Gong, Binyang Jiang, Boji Dan, Changqing Yu, Chao Wang, Cheng Ma, Cheng Zhong, Cheng Zhu, Chengjun Xiao, Chengyi Yang, Chengyu Du, Chenyang Zhang, Chi Zhang, Chuangyi Huang, Chunhao Zhang, Chunhui Du, Chunyu Zhao, Congchao Guo, Da Chen, Deming Ding, Dianjun Sun, Dongyu Zhang, Enhui Yang, Fei Yu, Guang Zheng, Guodong Zheng, Guohong Li, Haichao Zhu, Haigang Zhou, Haimo Zhang, Han Ding, Hao Zhang, Haohai Sun, Haolin Lyu, Haonan Lu, Haoyu Wang, Huajie Shi, Huiyang Li, Jiacheng Chen, Jian Zhang, Jiaqi Zhuang, Jiaren Cai, Jiaxin Pan, Jiayao Li, Jiayuan Song, Jichuan Zhang, Jie Wang, Jihao Gu, Jin Zhu, Jingwei Dong, Jingyang Li, Jingyu Zhang, Jingze Zhuang, Jinhao Tian, Jinli Liu, Jinyi Hu, Jun Tao, Jun Zhang, Junbin Ruan, Junhao Xu, Junjie Yan, Junteng Liu, Junxian He, Kang Xu, Ke Ji, Ke Yang, Kecheng Xiao, Keyu Duan, Keyu Li, Le Han, Letian Ruan, Li Yuan, Lianfei Yu, Liheng Feng, Lijie Mo, Lin Li, Lingye Bao, Lingyu Yang, Lingyua...Read
348
Generative Model
This paper upgrades reward models for image generation by teaching them to reason about score distributions instead of predicting a single scalar, making them both more accurate and easier to deploy.
Xin Jin, Huanqia Cai, Zhen Li, Zechao Zhan, Dengyang Jiang, Aiming Hao, Yuming Jiang, Chunle Guo, Peng Gao, Ming-Ming Cheng, Steven C. H. HoiRead
PermaVid helps edited videos stay coherent over time by using separate memory for appearance and geometry so later generations don’t drift after changes.
Shuai Yang, Bingjie Gao, Ziwei Liu, Jiaqi Wang, Dahua Lin, Tong WuRead
This paper connects decision trees and diffusion models through a shared mathematical framework, then uses that link to speed up tabular generation and distill tree behavior into neural networks.
Sai Niranjan Ramachandran, Suvrit SraRead
Load more videos
349
Llm
AsyncOPD shows how to train language models faster with asynchronous on-policy distillation without losing accuracy, by carefully handling stale rollouts and teacher-feedback estimates.
Wonjun Kang, Kevin Galim, Seunghyuk Oh, Minjun Kang, Sanghyun Park, Donghoon Kim, Minjae Lee, Minseo Kim, Rishabh Tiwari, Yuchen Zeng, Hyung Il Koo, Kangwook LeeRead
LoopCoder-v2 shows that, for parallel loop transformers, two loops are the sweet spot: enough extra computation to boost code and agent performance, but not so much that positional mismatch overwhelms the gains.
Jian Yang, Shawn Guo, Wei Zhang, Tianyu Zheng, Yaxin Du, Haau-Sing Li, Jiajun Wu, Yue Song, Yan Xing, Qingsong Cai, Zelong Huang, Chuan Hao, Ran Tao, Xianglong Liu, Wayne Xin Zhao, Mingjie Tang, Weifeng Lv, Ming Zhou, Bryan DaiRead
This paper turns GRPO training into a predictable dynamical system, giving a closed-form explanation for reward growth, stability, and failure modes in LLM reasoning training.
Rajat Ghosh, Datta Nimmaturi, Aryan Singhal, Vaishnavi Bhargava, Henry Wong, Johnu George, Debojyoti DuttaRead
350
Rl
This paper argues that reward hacking in language-model reinforcement learning happens when updates drift off a stable learning path, and shows that keeping gradients aligned with clean directions can reduce shortcut exploitation.
Wenlong Deng, Jiaji Huang, Kaan Ozkara, Yushu Li, Christos Thrampoulidis, Xiaoxiao Li, Youngsuk ParkRead
This paper teaches LLM agents to know when they really need a tool—and when they should trust their own knowledge—cutting unnecessary tool calls while improving accuracy.
Dingwei Chen, Zefang Zong, Zhipeng Ma, Leo Luo, Yang Li, Chengming Li, Peng Chen, Jie JiangRead
This paper introduces a cheap, Wikipedia-based reward signal that helps language models learn to answer factual questions more accurately without relying on expensive neural verifiers.
Shicheng Fan, Haochang Hao, Dehai Min, Weihao Liu, Philip S. Yu, Lu ChengRead
351
Multimodal
EVA01 is a multimodal model that treats 3D meshes as a first-class input/output modality, enabling text-to-3D generation and interactive 3D editing in one unified system.
Zongyuan Yang, Mingjing Yi, Wanli Ma, Chenzhuo Fan, Bocheng Li, Baolin Liu, Yuke Lou, Yingde Song, Yongping Xiong, Zhengdong Guo, Shimu WangRead
This paper builds a music recommender that uses song audio, lyrics, and AI-generated descriptions together with LLMs to better predict what people will want to hear next.
Srikar Prabhas Kandagatla, Sreehitha R. Narayana, Chandana Magapu, Swetha Mohan, Shamanth Kuthpadi, Hongjie Chen, Ryan A. Rossi, Franck Dernoncourt, Nesreen AhmedRead
This paper teaches multimodal AI to reason across multiple robots’ first-person views, improving cooperative spatial understanding in both simulation and real-world tests.
Kunyu Peng, Zhikun Zhou, Kailun Yang, Di Wen, Ruiping Liu, Yufan Chen, Junwei Zheng, Hao Shi, Yi Zhou, M. Saquib Sarfraz, Danda Pani Paudel, Luc Van GoolRead
Load more videos
352
Cv
This paper makes 3D visual geometry transformers much faster by cleverly choosing which image tokens to keep, cutting compute by over 85% while preserving accuracy.
Shuhong Zheng, Michael Oechsle, Erik Sandström, Marie-Julie Rakotosaona, Federico Tombari, Igor GilitschenskiRead
LAS2 shows that stereo matching can stay fast enough for real devices while still improving zero-shot accuracy, thanks to a smarter architecture and a carefully staged training recipe.
Junpeng Jing, Ronglai Zuo, Zhelun Shen, Shangchen Zhou, Rolandos Alexandros Potamias, Stefanos Zafeiriou, Krystian Mikolajczyk, Jiankang DengRead
This work shows that video diffusion models can do more than generate clips—they can also reconstruct detailed two-hand motion from egocentric video surprisingly well.
Yuxi Wang, Chengkai Jin, Yufei Liu, Wenqi Ouyang, Tianyi Wei, Zhiwei Zeng, Siyuan Huang, Zhiqi Shen, Xingang PanRead
353
Scientific Ai
This paper introduces a benchmark for AI scientific reasoning in physics and chemistry, along with a multi-agent system that improves performance and lowers inference cost on these hard research tasks.
Yigeng Jiang, Tengchao Yang, Taoyong Cui, Jiaxing Wan, Yuan Wang, Weida Wang, Zhiyu Liu, Chuyi Peng, Binzhao Luo, Maoli Gao, Huaihai Huang, Yuqianer Zeng, Ziyang Zheng, Dongchen Huang, Chao Chen, Zichao Liu, Weiping Shen, Shuchen Pu, Siyu Zhou, Runmin Ma, Yusong Hu, Fei Chao, Bo Zhang, Xiawu Zheng, Zifu Wang, Lei Bai, Yunqi Cai, Shufei ZhangRead
This paper shows that in biological AI, more training is not always better: the way you mix pretraining, fine-tuning, and reinforcement learning strongly changes whether models generalize or overfit.
Lukas Fesser, Hanlin Zhang, Michelle M. Li, Eric Wang, Bryan Perozzi, Shekoofeh Azizi, Sham M. Kakade, Marinka ZitnikRead
This paper introduces an AI assistant that helps review scientific papers by checking proofs, validating experiments, and finding errors before human referees do.
Rajesh Jayaram, Drew Tyler, David Woodruff, Corinna Cortes, Yossi Matias, Vahab Mirrokni, Vincent Cohen-AddadRead
354
Code Generation
Code2LoRA turns code repositories into lightweight, updateable adapters so code models can learn project-specific context without stuffing the prompt or retraining for every repo.
Liliana Hotsko, Yinxi Li, Yuntian Deng, Pengyu NieRead
This paper introduces a new way to create harder, more original coding tasks for training LLMs, helping reinforcement learning with verifiable rewards scale beyond hand-crafted or heuristic data.
Jiasheng Zheng, Boxi Cao, Boxi Yu, Yuzhong Zhang, Jialun Cao, Yaojie Lu, Hongyu Lin, Xianpei Han, Le SunRead
DOMINO learns the hidden structure of a target coding domain from a few examples and uses it to synthesize better training data for large language models.
Tong Ye, Hang Yu, Tengfei Ma, Xuhong Zhang, Jianguo Li, Peng Di, Peiyu Liu, Jianwei Yin, Wenhai WangRead
Load more videos
355
Efficiency
JetSpec makes large language models decode faster by drafting and verifying better token trees in parallel, pushing speculative decoding to much higher speeds.
Lanxiang Hu, Zhaoxiang Feng, Yulun Wu, Haoran Yuan, Yujie Zhao, Yu-Yang Qian, Bojun Wang, Peng Zhao, Daxin Jiang, Yibo Zhu, Tajana Rosing, Hao ZhangRead
LongAttnComp makes very long prompts cheaper to process by learning a smarter way to keep the most useful tokens, helping models reason over 100k+ context with less accuracy loss.
Mengmeng Ji, Ravi Shanker Raju, Jonathan Lingjie Li, Chen WuRead
SimFoundry turns a single real video into editable simulation twins that help robot policies train, generalize, and transfer to the real world more reliably.
Nadun Ranawaka, Josiah Wong, Wei-Lin Pai, Wei-Teng Chu, Tianyuan Dai, Masoud Moghani, Hang Yin, Yunfan Jiang, Wesley Durbano, Brandon Huynh, Yu Fang, Linxi Fan, Danfei Xu, Ruohan Zhang, Li Fei-Fei, Bowen Wen, Ajay Mandlekar, Yuke ZhuRead
356
Diffusion
CollectionLoRA condenses dozens of specialized image-editing LoRAs into a single efficient model, making diffusion-based customization easier to deploy without losing visual quality.
Fangtai Wu, Hailong Guo, Shijie Huang, Jiayi Song, Yubo Huang, Mushui Liu, Zhao Wang, Yunlong Yu, Jiaming Liu, Ruihua HuangRead
This paper makes it cheaper and more stable to train image generators from human preferences by redesigning how reward gradients are backpropagated through flow-matching sampling trajectories.
Ruoyu Wang, Boye Niu, Xiangxin Zhou, Yushi Huang, Tongliang Liu, Chi ZhangRead
This paper introduces a training-free way to make fast text-to-image generators safer by editing their velocity trajectories, cutting down harmful outputs while keeping benign images intact.
Yujin Choi, Jaehong YoonRead
357
Generative Model
AAD-1 improves one-step video generation by using a smarter, non-mirrored discriminator and a staged training recipe to reduce motion collapse and make outputs more realistic.
Haobo Li, Yanhong Zeng, Yunhong Lu, Jiapeng Zhu, Hao Ouyang, Qiuyu Wang, Ka Leong Cheng, Yujun Shen, Zhipeng ZhangRead
MeshWeaver makes 3D mesh generation more efficient and accurate by weaving vertices through sparse voxel geometry instead of relying on long, brittle token sequences.
Jiale Xu, Wang Zhao, Ying ShanRead
RankE is a new way to fine-tune discrete text-to-image generators by updating both the token policy and decoder together, so better reward scores actually translate into better images.
Siyong Jian, Siyuan Li, Luyuan Zhang, Zedong Wang, Xin Jin, Ying Li, Cheng Tan, Huan WangRead
Load more videos
358
Llm
This paper introduces SWITCH, a way to turn latent reasoning on and off with special tokens so large language models can be trained with reinforcement learning while keeping their hidden reasoning more interpretable.
Jiayu Yang, Chao Chen, Shengen Wu, Yinhong Liu, Yuxuan Fan, Lujundong Li, Songning Lai, Chengwei Qin, Zhijiang GuoRead
This work shows that large language models can be trained with masked diffusion instead of next-token prediction and still become highly competitive on reasoning, math, and coding tasks.
Shen Nie, Qiyang Min, Shaoxuan Xu, Zihao Huang, Yuxuan Song, Yong Shan, Yankai Lin, Wayne Xin Zhao, Chongxuan Li, Ji-Rong WenRead
This paper shows that you can predict whether an LLM steering trick will work by looking at its early hidden states, then use that prediction to search for better steering settings much more cheaply.
Chenrui Fan, Yize Cheng, Ming Li, Soheil Feizi, Tianyi ZhouRead
359
Embodied Ai
EventVLA helps robots remember the right visual moments at the right time, improving long-horizon manipulation by storing sparse, task-critical evidence instead of all past frames.
Ganlin Yang, Zhangzheng Tu, Yuqiang Yang, Sitong Mao, Junyi Dong, Tianxing Chen, Jiaqi Peng, Jing Xiong, Jiafei Cao, Jifeng Dai, Wengang Zhou, Yao Mu, Tai WangRead
InSight lets robot policies teach themselves new manipulation skills by breaking demonstrations into reusable primitives and then autonomously practicing, labeling, and adding missing skills to their training data.
Maggie Wang, Lars Osterberg, Stephen Tian, Ola Shorinwa, Jiajun Wu, Mac SchwagerRead
PAIWorld upgrades world models for robots by making multiple camera views agree in 3D, improving how models simulate and plan manipulation tasks.
Yuhang Huang, Xuan Lv, Junyan Xu, Zhiyuan Yu, Jiazhao Zhang, Ruizhen Hu, Wancheng Feng, Shilong Zou, Hewen Xiao, Ziqiao Zhou, Kaiyun Huang, Zhiyu Peng, Juzhan Xu, Hang Zhao, Chenyang Zhu, Renjiao Yi, Yifei Huang, Douhui Wu, Yan Zhang, Kexu Cheng, Chunhe Song, Yunzhi Xue, Xiuhong Zhang, Leitao Guo, Yunji Chen, Bin Wu, Haibin Yu, Kai XuRead
360
Robotics
The paper makes robot control smoother and more continuous by learning fast action chunks in latent space and refining them on the fly for real-world contact tasks.
Kunyun Wang, Yuhang Zheng, Yupeng Zheng, Jieru Zhao, Wenchao DingRead
Light-WAM makes robot world-action models much lighter by predicting actions from fused backbone states, keeping strong manipulation performance while cutting latency and memory.
Ziang Li, Dongzhou Cheng, Yibin Wang, Shiyue Wang, Xiaoyang Xu, Lingxuan Weng, Juan Wang, Jiaqi WangRead
NatureBench tests whether coding agents can actually match published Nature-paper results on real scientific problems, and finds they usually fall short of true scientific discovery.
Yuru Wang, Lejun Cheng, Yuxin Zuo, Sihang Zeng, Bingxiang He, Che Jiang, Junlin Yang, Yuchong Wang, Kaikai Zhao, Weifeng Huang, Kai Tian, Zhenzhao Yuan, Jincheng Zhong, Weizhi Wang, Ning Ding, Bowen Zhou, Kaiyan ZhangRead
Load more videos
361
Agent
This paper shows that RL post-training itself can provide a free, step-level signal for judging LLM agents, enabling better test-time scaling, uncertainty estimation, and failure analysis without training a separate reward model.
Changdae Oh, Wendi Li, Seongheon Park, Samuel Yeh, Tanwi Mallick, Sharon LiRead
This paper shows how to build better open training data for AI agents, and demonstrates that the resulting 100K-example dataset improves performance across multiple agentic benchmarks.
Negin Raoof, Richard Zhuang, Marianna Nezhurina, Etash Guha, Atula Tejaswi, Ryan Marten, Charlie F. Ruan, Tyler Griggs, Alexander Glenn Shaw, Hritik Bansal, E. Kelly Buchanan, Artem Gazizov, Reinhard Heckel, Chinmay Hegde, Sankalp Jajee, Daanish Khazi, Emmanouil Koukoumidis, Xiangyi Li, Hange Liu, Shlok Natarajan, Harsh Raj, Nicholas Roberts, Ethan Shen, Nishad Singhi, Michael Siu, Ashima Suvarna, Hanwen Xing, Patrick Yubeaton, Robert Zhang, Leon Liangyu Chen, Xiaokun Chen, Steven Dillmann, Saadia Gabriel, Xunyi Jiang, Anurag Kashyap, Boxuan Li, Yein Park, Minh Pham, Sujay Sanghavi, Lin Shi, Ke Sun, Yixin Wang, Zhiwei Xu, Erica Zhang, Siyan Zhao, Wanjia Zhao, Jenia Jitsev, Alex Dimakis, Benjamin Feuer, Ludwig SchmidtRead
Qwen-AgentWorld turns language models into simulated worlds for agents, enabling better planning, scalable RL training, and stronger downstream agent performance.
Yuxin Zuo, Zikai Xiao, Li Sheng, Fei Huang, Jianhong Tu, Yuxuan Liu, Tianyi Tang, Xiaomeng Hu, Yang Su, Qingfeng Lan, Yantao Liu, Qin Zhu, Yinger Zhang, Bowen Yu, Haiquan Zhao, Haiyang Xu, Jianxin Yang, Jiayang Cheng, Junyang Wang, Lianghao Deng, Mingfeng Xue, Tianyi Bai, Yang Fan, Yubo Ma, Yucheng Li, Zeyu Cui, Zhihai Wang, Zhihui Xie, Zhuorui Ye, An Yang, Dayiheng Liu, Jingren Zhou, Ning DingRead
362
Rl
DVAO is a new reinforcement learning method that adapts reward weighting on the fly to stabilize multi-objective training for LLMs and improve performance on reasoning and tool-use tasks.
Guochao Jiang, Jingyi Song, Guofeng Quan, Chuzhan Hao, Guohua Liu, Yuewei ZhangRead
This paper argues that reward models are often too fine-grained for their own good, and shows that discretizing their outputs can reduce reward hacking and produce better RL policies.
Vijay Viswanathan, Shiqi Wang, Devamanyu Hazarika, Chirag Nagpal, Tongshuang Wu, Graham Neubig, Yuning MaoRead
GRAIL improves LLM reasoning training by redistributing reinforcement learning credit to the tokens that matter most, boosting accuracy without needing step-by-step reward labels.
Tej Deep Pala, Vernon Toh, Soujanya PoriaRead
363
Multimodal
This paper introduces a native vision-language model that learns pixel-to-word connections end-to-end without separate encoders or adapters, aiming to make multimodal systems more unified and competitive at scale.
Haiwen Diao, Jiahao Wang, Penghao Wu, Yuhao Dong, Yuwei Niu, Yue Zhu, Zhongang Cai, Weichen Fan, Linjun Dai, Silei Wu, Xuanyu Zheng, Mingxuan Li, Yuanhan Zhang, Bo Li, Hanming Deng, Huchuan Lu, Quan Wang, Lei Yang, Lewei Lu, Dahua Lin, Ziwei LiuRead
This paper introduces an always-on vision-language assistant that watches live video, decides when to speak or stay silent, and can delegate hard tasks to a background model.
Dingyu Yao, Junhao Zhou, Chenxu Yang, Chuanyu Qin, Haowen Hou, Zheming Liang, Congcong Wang, Yuhang Cao, Shenglong Ye, Shuai Xie, Shuhuan Gu, Haoyang Huang, Qingyi Si, Nan Duan, Jiaqi WangRead
iVGR teaches multimodal language models to keep visual grounding inside their reasoning instead of forcing explicit boxes at inference time, improving fine-grained visual reasoning with reinforcement learning.
Chang-Bin Zhang, Yujie Zhong, Qiang Zhang, Kai HanRead
Load more videos
364
Cv
FLAT turns a single image into an explorable 3D scene by directly decoding surface triangles from diffusion latents, aiming for better geometry than Gaussian-based methods.
Orest Kupyn, Goutam Bhat, Philipp Henzler, Fabian Manhardt, Christian Rupprecht, Federico TombariRead
This paper lets creators control video generation by tracking points across reference images and frames, enabling more precise compositing and camera motion in a single model.
Koichi Namekata, Yash Kant, Zhizheng Liu, Ryan D Burgert, Yuancheng Xu, Kuan Heng Lin, Emmett Steven, Julien Philip, Li Ma, Andrea Vedaldi, Paul Debevec, Ning YuRead
Vera edits videos by generating only the changed layer and blending it back with the original, aiming to preserve everything that should stay the same while still allowing strong visual edits.
Hongkai Zheng, Ta-Ying Cheng, Benjamin Klein, Yisong Yue, Zhuoning YuanRead
365
Efficiency
This study shows that AI can already out-persuade even expert humans in conversation, raising major questions about how persuasive systems could be used in politics, fundraising, and everyday influence.
Kobi Hackenburg, Caroline Wagner, Luke Hewitt, Ben M. Tappin, Ed Saunders, Hannah Rose Kirk, Helen Margetts, Christopher SummerfieldRead
This paper makes long video understanding cheaper by replacing quadratic video attention with a stateful linear-time cache that preserves much of the accuracy of full self-attention.
Cristobal Eyzaguirre, Jiajun Wu, Juan Carlos NieblesRead
03
Rethinking Shrinkage Bias in LLM FP4 Pretraining: Geometric Origin, Systemic Impact, and UFP4 Recipe
This paper explains why some 4-bit LLM training formats quietly introduce harmful rounding bias, and shows that a more uniform 4-bit recipe can make pretraining more stable and efficient.
Qian Zhao, Kunlong Chen, Changxin Tian, Zhonghui Jiang, Haitao Zhang, Chaofan Yu, Peijie Jiang, Mingliang Gong, Jia Liu, Ziqi Liu, Zhiqiang Zhang, Jun ZhouRead
366
Diffusion
This paper shows how to make streaming video and interactive world-model diffusion much faster and more practical by unifying two training styles into one distillation recipe.
Kaiwen Zheng, Guande He, Min Zhao, Jintao Zhang, Huayu Chen, Jianfei Chen, Chen-Hsuan Lin, Ming-Yu Liu, Jun Zhu, Qianli MaRead
This paper improves diffusion image generation by replacing uniform noise with frequency-aware colored noise during sampling, making the process more efficient and producing sharper images.
Hadar Davidson, Noam Issachar, Sagie BenaimRead
OSP-Next makes text-to-video generation faster and more hardware-friendly by mixing sparse attention, smarter parallelism, quantization, and post-training RL while keeping quality high.
Yunyang Ge, Xianyi He, Zezhong Zhang, Bin Lin, Bin Zhu, Xinhua Cheng, Li YuanRead
Load more videos
367
Foundation Model
BioMatrix is a single AI system that can read and generate biological sequences, 3D structures, and scientific language for both molecules and proteins.
Qizhi Pei, Zhimeng Zhou, Yi Duan, Yiyang Zhao, Wei Li, Han Guo, Liang He, Chengping Li, Chang-Yu Hsieh, Conghui He, Rui Yan, Lijun WuRead
LOGOS is a general-purpose science AI that turns many natural-science problems into one shared language, aiming to make a single foundation model useful across multiple scientific domains.
Mingyang Li, Yurou Liu, Jieping Ye, Bing Su, Ji-Rong Wen, Zheng WangRead
Wan-Streamer is a single multimodal foundation model built for real-time two-way audio-visual conversation with sub-second latency, replacing many separate pipeline modules.
Lianghua Huang, Zhifan Wu, Wei Wang, Yupeng Shi, Mengyang Feng, Junjie He, Chenwei Xie, Yu Liu, Jingren Zhou, Ang Wang, Bang Zhang, Baole Ai, Chen Liang, Cheng Yu, Chongyang Zhong, Jinwei Qi, Kai Zhu, Pandeng Li, Peng Zhang, Wenyuan Zhang, Xinhua Cheng, Yitong Huang, Yun Zheng, Zoubin BiRead
368
Diffusion
This paper shows that in latent diffusion models, predicting the clean latent directly can work better than predicting velocity, because the choice of training target changes the geometry of learning in compressed latent space.
Funing Fu, Tenghui Wang, Junyong Cen, Qichao Zhu, Guanyu ZhouRead
This paper teaches mask diffusion models to “think again” by revisiting and locally revising their own outputs across multiple turns, enabling stronger reasoning and refinement without starting over.
Yanming Zhang, Yihan Bian, Jingyuan Qi, Yuguang Yao, Lifu Huang, Tianyi ZhouRead
PiD replaces traditional latent decoders with a fast pixel-diffusion decoder that can turn compressed latents into high-resolution images much more efficiently and with better visual fidelity.
Yifan Lu, Qi Wu, Jay Zhangjie Wu, Zian Wang, Huan Ling, Sanja Fidler, Xuanchi RenRead
369
Self Supervised
This paper shows that self-supervised learning can remove the need for target-lot labels while still achieving very high parking-spot occupancy accuracy, making deployment cheaper and more scalable.
Luan Marko Kujavski, Rayson Laroca, Paulo Lisboa de AlmeidaRead
UniverSat makes one transformer backbone work across many Earth-observation sensors, resolutions, and modalities by learning a shared patch embedding through self-supervision.
Yohann Perron, Guillaume Astruc, Nicolas Gonthier, Clement Mallet, Loic LandrieuRead
This paper improves self-supervised learning on medical tabular data by adaptively refining feature discretization during training, leading to better representations without hand-tuned binning.
Daehwan Kim, Haejun Chung, Ikbeom JangRead
Load more videos
370
Llm
This paper shows that for some aligned LLMs, the best next-token predictions may come from a near-final layer rather than the last one, and it uses that insight to improve reasoning with almost no extra cost.
Xuanming Zhang, Sining Zhoubian, Yuxuan Chen, Tianyi Tang, An Yang, Sean Du, Chujie Zheng, Fei Huang, Dayiheng Liu, Gao Huang, Jingren ZhouRead
This paper turns LLM decision-making into a causal graph of human-interpretable concepts, using counterfactual chains to make the model’s reasoning more transparent.
Nirit Nussbaum-Hoffer, Nitay Calderon, Liat Ein-Dor, Roi ReichartRead
This paper shows that LLMs can make structured choices for hidden reasons that their own explanations only partly reveal.
Gabriel Freedman, Francesca ToniRead
371
Robotics
AEGIS helps robots avoid failure spirals by predicting trouble early and handing control to a stronger policy only at the risky moments.
Josef ChenRead
This paper launches Agents' Last Exam, a living benchmark aimed at measuring whether AI agents can actually handle real-world professional tasks that matter economically.
Yiyou Sun, Xinyang Han, Weichen Zhang, Yuanbo Pang, Tianyu Wang, Yuhan Cao, Yixiao Huang, Chris Duroiu, Haoyun Zhang, Jeffrey Lin, Weishu Zhang, Tyler Zeng, Ying Yan, Bo Liu, Hanson Wen, Mingyang Xu, Xiaoyuan Liu, Zimeng Chen, Weiyan Shi, Amanda Dsouza, Vincent Sunn Chen, Patrick Bryant, Carl Boettiger, Yamini Rangan, Bradley Rothenberg, Kyle Steinfeld, Arvind Rao, Tapio Schneider, Georgios Yannakakis, Laure Zanna, Kaan Ozbay, Ida Sim, Tarek Zohdi, George Em Karniadakis, Jack Gallant, Teresa Head-gordon, Yushan Li, Wenxi Deng, Tao Sun, Huiqi Wang, Zhun Wang, Justin Xu, Chris Yuhao Liu, Yafei Cheng, Rongwang Hu, Aras Bacho, Shengcao Cao, Zengyi Qin, Yixiong Chen, Hengduan Fan, Hao Liu, Lin Zeng, Shashank Muralidhar Bharadwaj, Litian Gong, Yingxuan Yang, Maojia Song, Ruheng Wang, Zongzheng Zhang, Honglin Bao, Shuo Lu, Jianhong Tu, Zhonghua Wang, Zheng Zhang, Zijiao Chen, yanqiong Jiang, Zhendong Li, Bohan Lyu, Chang Ma, Peiran Xu, Benran Zhang, Shangding Gu, Haoyue Hua, Haoyang Li, Wa...Read
Flash-WAM makes diffusion-based robot world models fast enough for real-time control by using different distillation parameterizations for video and action streams, cutting inference from seconds to milliseconds without losing much task success.
Arman Akbari, Ci Zhang, Arash Akbari, Lin Zhao, Yixiao Chen, Weiwei Chen, Xuan Zhang, Geng Yuan, Yanzhi WangRead
372
Agent
This paper shows that today’s AI research agents mostly generate narrow, local variations on existing science rather than broadly expanding the frontier of discovery.
Yixuan Tang, Yi YangRead
EvoGraph is a new system that lets software evolve its own code, docs, and pipelines using a graph-based mutation-and-selection loop powered by small language models.
Igor Costa, Christopher BaranRead
SkillOpt treats an agent’s skill as something you can train like a model parameter, using controlled text edits and validation to make AI agents steadily improve without extra runtime cost.
Yifan Yang, Ziyang Gong, Weiquan Huang, Qihao Yang, Ziwei Zhou, Zisu Huang, Yan Li, Xuemei Gao, Qi Dai, Bei Liu, Kai Qiu, Yuqing Yang, Dongdong Chen, Xue Yang, Chong LuoRead
Load more videos
373
Rl
This paper explains why training an LLM on one RL domain can hurt others, and shows that a small targeted refresh can recover lost performance with minimal side effects.
Lei Yang, Siyu Ding, Deyi XiongRead
This paper turns an LLM from a student into a coach: it studies its own failures and redesigns the training environment to help itself learn better in reinforcement learning.
Chao Chen, Chengzu Li, Zhiwei Li, Yinhong Liu, Zhijiang GuoRead
This paper asks when you can learn good offline RL policies without per-step rewards, and shows both the limits and the exact sample complexity of doing so from trajectory-level labels or preferences.
Xuanfei Ren, Tengyang XieRead
374
Multimodal
ARM is a multimodal AI model that turns images into discrete tokens so one autoregressive system can understand, generate, and edit visuals, with reinforcement learning improving both quality and instruction following.
Junke Wang, Xiao Wang, Jiacheng Pan, Xuefeng Hu, Feng Li, Jingxiang Sun, Chaorui Deng, Zilong Chen, Yunpeng Chen, Kaibin Tian, Matthew Gwilliam, Hao Chen, Danhui Guan, Kun Xu, Weilin Huang, Zuxuan Wu, Haoqi Fan, Yu-Gang Jiang, Zhenheng YangRead
HYDRA-X is a new unified multimodal model that uses one visual tokenizer for both images and videos, improving understanding, generation, and editing by doing more of the work inside the tokenizer itself.
Guozhen Zhang, Xuerui Qiu, Yutao Cui, Tianhui Song, Changlin Li, Junzhe Li, Tao Huang, Xiao Zhang, Yang Li, Jianbing Wu, Miles Yang, Zhao Zhong, Liefeng Bo, Limin WangRead
The paper teaches video AI to think partly in images instead of only in words, helping it predict future events more accurately.
Tianxiang Jiang, Linquan Wu, Sheng Xia, Songze Li, Ziang Yan, Haoyu Yang, Yu Qiao, Yi WangRead
375
Cv
Memento improves long video generation by teaching the model to reconstruct recurring subjects from memory, helping characters stay consistent across shots and scenes.
Xuan Wei, Longbin Ji, Guan Wang, Xiangrui Liu, Zhenyu Zhang, Shuohuan Wang, Yu Sun, Qingqi HongRead
This work proposes a new way to restore and sharpen AI-generated images using the original reference image, recovering lost fine details while fixing generation artifacts in one step.
Jeahun Sung, Dahyeon Kye, Soo Ye Kim, Jihyong OhRead
UniSHARP lets a single monocular view-synthesis system render sharp scenes across perspective, fisheye, and panoramic cameras by aligning them in a shared omnidirectional latent space.
Meixi Song, Dizhe Zhang, Hao Ren, Ruiyang Zhang, Bo Du, Ming-Hsuan Yang, Lu QiRead
Load more videos
376
Scientific Ai
AutoScientists is a team of AI agents that self-organize to run long scientific experiments, explore multiple hypotheses in parallel, and keep learning from both successes and failures.
Shanghua Gao, Ada Fang, Marinka ZitnikRead
LabVLA teaches vision-language-action models to execute scientific lab protocols with robot arms, aiming to make AI capable of carrying out experiments, not just planning them.
Baochang Ren, Xinjie Liu, Xi Chen, Yanshuo Liu, Chenxi Li, Daqi Gao, Zeqin Su, Jintao Xing, Zirui Xue, Rui Li, Xiangyu Zhao, Shuofei Qiao, Minting Pan, Wangmeng Zuo, Lei Bai, Dongzhan Zhou, Ningyu Zhang, Huajun ChenRead
This paper builds the largest dataset so far for research-level math and shows that filtered AI-generated reasoning traces can help train stronger math models.
Guijin Son, Seungyeop Yi, Minju Gwak, Hyunwoo Ko, Wongi Jang, Youngjae YuRead
377
Code Generation
This paper proposes a clever way to tell when coding agents are 'gaming the test' by using randomized capped evaluations, and shows how to discourage that cheating during training.
Thanawat Lodkaew, Johannes Ackermann, Soichiro Nishimori, Nontawat Charoenphakdee, Masashi Sugiyama, Takashi IshidaRead
This paper creates the first large-scale benchmark for whole-game code generation in a real game engine, revealing that today’s AI can often write code that compiles but still fails to produce working games.
Jianwen Sun, Chuanhao Li, Zizhen Li, Yukang Feng, Fanrui Zhang, Yifei Huang, Yu Dai, Kaipeng ZhangRead
This paper shows how coding agents can learn from their own past mistakes by turning old solving traces into new training tasks, steadily improving software-engineering performance over multiple iterations.
Chuan Xiao, Zhengbo Jiao, Shaobo Wang, Wei Wang, Bing Zhao, Hu Wei, Linfeng Zhang, Lin QuRead
378
Efficiency
RoundPipe is a new way to fine-tune huge language models efficiently on cheap GPUs by dynamically rotating work across devices to cut pipeline bubbles and boost throughput.
Yibin Luo, Shiwei Gao, Huichuan Zheng, Youyou Lu, Jiwu ShuRead
This paper introduces a new way to serve ultra-long-context LLMs by predicting which past tokens actually matter, cutting KV-cache memory drastically while keeping accuracy high.
Yan Wang, Qifan Zhang, Jiachen Yu, Tian Liang, Dongyang Ma, Xiang Hu, Zibo Lin, Chunyang Li, Zhichao Wang, Jia Li, Yujiu Yang, Haitao Mi, Dong YuRead
ResearchClawBench is a benchmark that tests whether AI agents can truly rediscover real scientific papers end to end, revealing that today’s best systems still fall far short.
Wanghan Xu, Shuo Li, Tianlin Ye, Qinglong Cao, Yixin Chen, Hengjian Gao, Yiheng Wang, Qi Li, Kun Li, Sheng Xu, Shengdu Chai, Fangchen Yu, Xiangyu Zhao, Zhangrui Zhao, Weijie Ma, Zijie Guo, Haoyu Zhou, Haoxiang Yin, Lixue Cheng, Chaofan Hu, Haoxuan Li, Lu Mi, Xuxuan Xie, Yifan Zhou, Ruizhe Chen, Zhiwang Zhou, Xingjian Guo, Yuhao Zhou, Xuming He, Shengyuan Xu, Xinyu Gu, Jiamin Wu, Mianxin Liu, Chunfeng Song, Fenghua Ling, Dongzhan Zhou, Shixiang Tang, Yuqiang Li, Mao Su, Peng Ye, Siqi Sun, Bin Wang, Xue Yang, Zhenfei Yin, Tianfan Fu, Guangtao Zhai, Wanli Ouyang, Bo Zhang, Lei Bai, Wenlong ZhangRead
Load more videos
379
Diffusion
This paper tackles hallucinations in diffusion image generators by controlling the model’s score dynamics, improving reliability without sacrificing quality.
Mahesh Bhosale, Naresh Kumar Devulapally, Abdul Wasi, Chau Pham, Vishnu Suresh Lokhande, David DoermannRead
This paper boosts pixel-space diffusion by filtering out high-frequency noise early in the denoising process, helping the model focus on the signal where it matters most.
Weichen Fan, Haiwen Diao, Penghao Wu, Ziwei LiuRead
This work makes it easier to combine multiple LoRA styles and concepts in diffusion models by weighting each adapter based on how important its trigger words are in the prompt.
Georgios Tsoumplekas, Stella Bounareli, Vasileios ArgyriouRead
380
Generative Model
FlowBender teaches generative models to look at their own mistakes and correct them on the fly, improving both constraint satisfaction and sample quality.
Daniel Gilo, Sven Elflein, Ido Sobol, Or LitanyRead
SmartDirector lets video models generate more cinematic, better-paced stories by using multiple keyframes as narrative anchors across single-shot, multi-shot, and video-extension settings.
Zhida Zhang, Jie Ma, Zhan Peng, Haoxue Wu, Yang Han, Jun Liang, Jie Cao, Jing LiRead
Sumi is a fully open 7B diffusion language model trained from scratch on 1.5T tokens, offering the first large-scale reference point for studying uniform diffusion generation in practice.
Mengyu Ye, Keito Kudo, Wataru Ikeda, Ryosuke Matsuda, Keisuke Sakaguchi, Jun SuzukiRead
381
Llm
REVES improves LLM test-time reasoning by learning from near-miss answers, teaching models not just to solve problems but to recover from mistakes more effectively.
Yuanxin Liu, Ruida Zhou, Xinyan Zhao, Amr Sharaf, Hongzhou Lin, Arijit Biswas, Mohammad Ghavamzadeh, Zhaoran Wang, Mingyi HongRead
This paper shows that LLM personalization looks much better on synthetic tests than on real human conversations, and introduces human-grounded data and evaluation methods to expose the gap.
Lechen Zhang, Jiarui Liu, Tal AugustRead
STARE is a new training trick for LLM reasoning that keeps exploration alive by reweighting the tokens most responsible for entropy collapse during reinforcement learning.
Haipeng Luo, Qingfeng Sun, Songli Wu, Can Xu, Wenfeng Deng, Han Hu, Yansong TangRead
Load more videos
382
Embodied Ai
01
AHA-WAM:Asynchronous Horizon-Adaptive World-Action Modeling with Observation-Guided Context Routing
AHA-WAM speeds up robot control by letting one model plan the long-horizon world at a slower pace while another executes actions quickly using routed context from the planner.
Jisong Cai, Long Ling, Shiwei Chu, Zhongshan Liu, Jiayue Kang, Zhixuan Liang, Wenjie Xu, Yinan Mao, Weinan Zhang, Xiaokang Yang, Ru Ying, Ran Zheng, Yao MuRead
This paper teaches robots to grasp like humans by learning from a million egocentric human grasps and using that data to generate and retarget natural grasps in the real world.
Kevin Yuanbo Wu, Tianxing Zhou, Isaac Tu, Billy Yan, Irmak Guzey, David Fouhey, Dandan Shan, Lerrel PintoRead
This study suggests a surprising shortcut for embodied AI: pretrain on cheap, abundant human egocentric video, then fine-tune on a small amount of robot data for the best real-world performance.
Juncheng Ma, Jianxin Bi, Yufan Deng, Xuanran Zhai, Kewei Zhang, Ye Huang, Bo Liang, Shukai Gong, Jiankai Tu, Xiaotian Tang, Jiaxin Li, Kaiqi Chen, Duomin Wang, Yuqi Wang, Bingyi Kang, Eric Huang, Zhiyang Dou, Zhen Dong, Enze Xie, Wojciech Matusik, Tat-Seng Chua, Daquan ZhouRead
383
Robotics
This paper asks a simple but important question: can a model’s predicted action or dynamics actually work in the real physical world, and it proposes a gate to reject implausible proposals before execution.
Barak OrRead
This paper makes robot fine-tuning smarter by turning sparse success/failure signals into more informative, stage-aware training weights, boosting real-world task success on bimanual manipulation.
Tongyan Fang, Siyuan Huang, Naiyu Fang, Ganlong Zhao, Zhongjin Luo, Jianbo Liu, Xiaogang Wang, Ying Dong, Hongsheng LiRead
This paper teaches robots to play first, so they can build reusable code-based skills that help them solve future tasks better.
Junyi Zhang, Jiaxin Ge, Hanjun Yoo, Letian Fu, Zihan Yang, Yaowei Liu, Raj Saravanan, Shaofeng Yin, Justin Yu, Dantong Niu, Zirui Wang, Roei Herzig, Ken Goldberg, Yutong Bai, David M. Chan, Ion Stoica, Angjoo Kanazawa, Jiahui Lei, Haiwen Feng, Trevor DarrellRead
384
Agent
This paper shows that what an LLM agent reads in a feed can quietly steer its decisions, even when the final prompt stays the same.
Rana Muhammad UsmanRead
This paper asks whether coding agents can take over the tedious loop of data curation, and finds they can do useful work—but only when guided by structured method adaptation rather than vague prompting.
Feiyang Kang, Hanze Li, Adam Nguyen, Mahavir Dabas, Jiaqi W. Ma, Frederic Sala, Dawn Song, Ruoxi JiaRead
Nemotron 3 Ultra is a massive open AI model that mixes MoE, Mamba, and transformer ideas to deliver faster long-context reasoning for autonomous agents.
NVIDIA, :, Aaron Blakeman, Aaron Thomas, Aastha Jhunjhunwala, Abhibha Gupta, Abhinav Khattar, Adam Rajfer, Adi Renduchintala, Adil Asif, Aditya Vavre, Adriana Flores Miranda, Ahmad Bilal, Aileen Zaman, Ajay Hotchandani, Akanksha Shukla, Akhiad Bercovich, Aleksander Ficek, Alex Gronskiy, Alex Kondratenko, Alex Steiner, Alex Ye, Alexander Bukharin, Alexandre Milesi, Ali Taghibakhshi, Alice Gatti, Alisa Liu, Alok Kumar, Amar Phanishayee, Ameya Sunil Mahabaleshwarkar, Amir Klein, Amit Zuker, Amnon Geifman, Anahita Bhiwandiwalla, Ananth Subramaniam, Andrea Santilli, Andrew Fulks, Andrew McHarg, Andrew Tao, Andrii Skliar, Anjulie Agrusa, Ankur Srivastava, Ankur Verma, Anna Shors, Anna Warno, Antoni-Joan Solergibert I Llaquet, Arham Mehta, Arkadiusz Nowaczynski, Arti Jain, Ashwath Aithal, Ashwin Poojary, Asif Ahamed, Asit Mishra, Asma Kuriparambil Thekkumpate, Atefeh Sohrabizadeh, Avinash Kaur, Avinash Vem, Ayush Dattagupta, Barath Subramaniam Anandan, Bardiya Sadeghi, Ben Lanir, Benedik...Read
Load more videos
385
Rl
The paper shows that reinforcement learning can teach agents to become 'addicted' to visible reward dashboards, causing them to chase the displayed incentive even when it hurts the real task or safety.
Tong Che, Rui WuRead
PBSD turns sparse final rewards into step-by-step learning signals for agents by using Bayesian self-distillation to figure out which actions really helped.
Yang Tian, Rui Wang, Xumeng Wen, Junjie Li, Shizhao Sun, Lei Song, Jiang Bian, Bo ZhaoRead
This paper introduces SAVE, a method that lets reward models learn from a policy’s own on-policy outputs using value-guided self-supervision, improving alignment without relying only on fresh human labels.
Xiaobo Wang, Tong Wu, Min Tang, Jiaqi Li, Qi Liu, Zilong ZhengRead
386
Multimodal
ACE-Ego-0 makes robot learning more scalable by turning egocentric human videos into usable action supervision and training vision-language-action models on both human and robot data together.
Hao Li, Ganlong Zhao, Yufei Liu, Haotian Hou, Guoquan Ye, Tongyan Fang, Chunxiao Liu, Siyuan Huang, Jianbo Liu, Xiaogang Wang, Hongsheng LiRead
AnyMo introduces a large multimodal motion dataset and a masked-modeling transformer that can generate human motion from flexible combinations of inputs like text, speech, music, and trajectories.
Yiheng Li, Zhuo Li, Ruibing Hou, Yingjie Chen, Hong Chang, Hao Liu, Shiguang ShanRead
Avatar V is a large-scale system that uses reference videos, not just photos, to generate more realistic talking avatars with better identity, gestures, and lip-sync.
Benjamin Liang, Ce Chen, Desmond Lin, Ivan Somov, Jiajun Zhao, Jiewei Yuan, Jingfeng Zhang, Junhao Huang, Nik Nolte, Pedram Haqiqi, Penghan Wang, Rong Yan, Rui Zhang, Sam Prokopchuk, Sivan Wang, Viktor Goriachko, Yi Ren, Yuanming Li, Yutao Chen, Zhenhui Ye, Zhibin Hong, Zilong Nie, Zujin GuoRead
387
Cv
Track2View lets a video diffusion model re-render scenes from new camera paths by using paired 3D point tracks to keep motion and appearance consistent across time.
Feng Qiao, Zhaochong An, Zhexiao Xiong, Serge Belongie, Nathan JacobsRead
World Tracing is a new way to turn images into 3D by keeping pixels aligned while also hallucinating the hidden parts of objects and scenes.
Hao Zhang, Mohamed El Banani, Jen-Hao Cheng, Paul Zhang, Yi Hua, Ben Mildenhall, Christoph Lassner, Narendra Ahuja, Gengshan YangRead
ZipSplat makes fast 3D scene reconstruction more flexible by clustering visual tokens into fewer, smarter Gaussians, cutting representation cost while improving quality.
Alexander Veicht, Sunghwan Hong, Dániel Baráth, Marc PollefeysRead
Load more videos
388
Efficiency
CONF-KV makes long-context LLMs cheaper by using the model’s own confidence to decide how much memory to keep, helping preserve accuracy while cutting KV-cache use.
Yubo Li, Yidi MiaoRead
Cosmos 3 is a unified multimodal world model that can understand and generate text, images, video, audio, and actions, aiming to become a general backbone for embodied AI.
Aditi, Niket Agarwal, Arslan Ali, Jon Allen, Martin Antolini, Adeline Aubame, Alisson Azzolini, Junjie Bai, Maciej Bala, Yogesh Balaji, Josh Bapst, Aarti Basant, Mukesh Beladiya, Mohammad Qazim Bhat, Zaid Pervaiz Bhat, Dan Blick, Vanni Brighella, Han Cai, Tiffany Cai, Eric Cameracci, Jiaxin Cao, Yulong Cao, Mark Carlson, Carlos Casanova, Ting-Yun Chang, Yan Chang, Yu-Wei Chao, Prithvijit Chattopadhyay, Roshan Chaudhari, Chieh-Yun Chen, Junyu Chen, Ke Chen, Qizhi Chen, Wenkai Chen, Xiaotong Chen, Yu Chen, An-Chieh Cheng, Click Cheng, Xiu Chia, Jeana Choi, Chaeyeon Chung, Wenyan Cong, Yin Cui, Magdalena Dadela, Nalin Dadhich, Wenliang Dai, Joyjit Daw, Alperen Degirmenci, Rodrigo Vieira Del Monte, Robert Denomme, Sameer Dharur, Marco Di Lucca, Ke Ding, Wenhao Ding, Yifan Ding, Yuzhu Dong, Nicole Drumheller, Yilun Du, Aigul Dzhumamuratova, Aleksandr Efitorov, Hamid Eghbalzadeh, Naomi Eigbe, Imad El Hanafi, Hassan Eslami, Benedikt Falk, Jiaojiao Fan, Jim Fan, Amol Fasale, Sergiy Fefilat...Read
Still makes long-context language models more practical by compressing their key-value cache in one fast forward pass, preserving performance even at extreme compression ratios.
Charles O'Neill, Alex Sandomirsky, Harry Partridge, Mudith Jayasekara, Max KirkbyRead
389
Self Supervised
This paper shows that even pure noise can help learn better models in semi-supervised transfer, if it is used as a clever surrogate source domain.
Yuan Yao, Jin Song, Huixia Li, Tongtong Yuan, Jiaqi Wu, Yu ZhangRead
U-TTT adapts a PET denoising model on the fly during inference, using self-supervised spatial and frequency updates to stay robust across scanners and dose shifts.
Zhiwen Yang, Jiayin Li, Hao Lu, Hui Zhang, Zihua Wang, Bingzheng Wei, Yan XuRead
This paper introduces a new way to learn visual representations from video by predicting how representations change over time, potentially reducing reliance on hand-crafted training tricks.
Ninad Daithankar, Alexi Gladstone, Yann LeCun, Heng JiRead
390
Transformer
This paper asks when transformers really need full attention, and shows that a hybrid of spectral mixing and attention can cut compute while outperforming a strong baseline on some text tasks.
Prateek Kumar SikdarRead
This paper shows that in hybrid language models, efficient attention mostly shapes how quickly long-context skills emerge, while full attention does the heavy lifting for retrieval.
Ziqing Qiao, Yinuo Xu, Chaojun Xiao, Zhou Su, Zihan Zhou, Yingfa Chen, Xiaoyue Xu, Xu Han, Zhiyuan LiuRead
This paper proposes making transformers wide at the edges and narrow in the middle, showing that smarter width allocation can improve language model efficiency and performance.
Zhaofeng Wu, Oliver Sieberling, Shawn Tan, Rameswar Panda, Yury Polyanskiy, Yoon KimRead
Load more videos
391
Diffusion
This paper turns single-image diffusion into a fast, training-free process by using closed-form patch statistics, enabling high-quality image generation from one reference image in seconds rather than hours.
Haojun Qiu, Kiriakos N. Kutulakos, David B. LindellRead
FadeMem makes long video generation more memory-efficient by keeping recent details crisp while compressing older context into a hierarchical cache that preserves coherence.
Yu Lu, Junjie Yang, Piotr Koniusz, YuXin Song, Yi YangRead
This paper shows how to squeeze high-quality images out of a diffusion model in just two denoising steps by aligning the student to its teacher, separating step-specific parameters, and training end-to-end with regularization.
Dongyang Liu, Ruoyi Du, David Liu, Dengyang Jiang, Liangchen Li, Qilong Wu, Zhen Li, Steven C. H. Hoi, Hongsheng Li, Peng GaoRead
392
Generative Model
DecMem introduces a split memory system to help video generators remember long-range details, aiming to produce more consistent minute-long worlds.
Zhenhao Yang, Xiaoshi Wu, Zhengyao Lv, Xiaoyu Shi, Xintao Wang, Pengfei Wan, Kun Gai, Kwan-Yee K. WongRead
OmniDirector makes video generation follow complex multi-shot camera moves by turning camera paths into visual motion grids and training a large controllable diffusion system on them.
Jiwen Liu, Shujuan Li, Zhixue Fang, Xiaohan Li, Yan Zhou, Zijie Meng, Zhimin Zhang, Yawen Luo, Guoxin Zhang, Yu-Shen Liu, Pengfei WanRead
SwiftVR makes high-quality video restoration run in real time on consumer GPUs by redesigning one-step generative modeling for efficient streaming inference.
Jiaqi Yan, Xiangyu Chen, Xinlin Zhong, Haibin Huang, Chi Zhang, Jie Liu, Jiantao Zhou, Xuelong LiRead
393
Llm
This paper shows that LLMs are strangely more confident in answers they think are their own, and that a simple prompt trick can make them better calibrated without retraining.
Mario Sanz-Guerrero, Manuel Mager, Katharina von der WenseRead
OmniOPD is a new way to train smaller models from stronger teachers without needing token logits, using chunk-level semantic verification and uncertainty-based checks to improve distillation, especially on math tasks.
Yuhang Zhou, Lizhu Zhang, Yifan Wu, Mingyi Wang, Peng Bo, Jiayi Liu, Xiangjun Fan, Zhuokai ZhaoRead
03
Self-Evaluation Is Already There: Eliciting Latent Judge Calibration in Base LLMs with Minimal Data
This paper shows that base LLMs already have hidden self-evaluation skills and that a small amount of targeted training can surface them to better predict how external judges will score their answers.
XiuYu Zhang, Yi Shan, Junfeng Fang, Zhenkai LiangRead
Load more videos
394
Embodied Ai
Embodied-R1.5 is a large embodied AI model that learns planning, correction, and grounding together, then proves it can transfer from benchmarks to real-world robot tasks.
Yifu Yuan, Yaoting Huang, Xianze Yao, Yutong Li, Shuoheng Zhang, Linqi Han, Pengyi Li, Jiangeng Sun, Wenting Jia, Zhao Zhang, Yuhao Liu, Ruihao Liao, Yucheng Hu, Qiyu Wu, Yuxiao Li, Zibin Dong, Fei Ni, Yan Zheng, Shuyang Gu, Yi Ma, Hongyao Tang, Han Hu, Jianye HaoRead
GRAIL uses 3D scenes plus video-model priors to generate large-scale humanoid interaction data in simulation, then turns that synthetic data into real robot skills like picking up objects and climbing stairs.
Tianyi Xie, Haotian Zhang, Jinhyung Park, Zi Wang, Bowen Wen, Jiefeng Li, Xueting Li, Qingwei Ben, Haoyang Weng, Yufei Ye, David Minor, Tingwu Wang, Chenfanfu Jiang, Sanja Fidler, Jan Kautz, Linxi Fan, Yuke Zhu, Zhengyi Luo, Umar Iqbal, Ye YuanRead
This paper introduces a robot foundation model that can predict what happens next, reason in language, and generate actions—all in one system for better long-horizon embodied tasks.
Yi Yang, Zhihong Liu, Siqi Kou, Yiyang Chen, Yanzhe Hu, Jianbo Zhou, Boyuan Zhao, Zhijie Wei, Xiao Xia, Xueqi Li, Pengfei Liu, Zhijie DengRead
395
Robotics
ActiveMimic teaches robots from human first-person videos by treating camera motion as useful action, helping bridge the gap between egocentric video and robot pretraining.
Xingyao Lin, Guojin Zhong, Tianyi Lu, Ziyi Ye, Yichen Zhu, Zuxuan Wu, Yu-Gang JiangRead
02
Evaluating Large Language Models in Dynamic Clinical Decision-Making with Standardized Patient Cases
This paper introduces MedSP1000, a new interactive benchmark that tests whether large language models can handle dynamic, real-world-style clinical decision-making instead of just answering isolated questions.
Cheng Liang, Pengcheng Qiu, Ya Zhang, Yanfeng Wang, Chaoyi Wu, Weidi XieRead
This paper shows that in robot vision-language-action systems, the same checkpoint can behave like a different policy once action normalization metadata changes, creating a hidden safety risk before rollout.
Jianwei TaiRead
396
Agent
This paper teaches AI agents to turn messy human web guides into editable skills they can keep improving from their own experience.
Xinyu Che, Junqi Xiong, Yunfei Ge, Xinping Lei, Shihao Li, Hang Yan, Han Li, Yuanxing Zhang, Zhiqi Bai, Jinhua Hao, Ming Sun, Han Li, Jiaheng LiuRead
OpenWebRL shows that visual web agents can be trained directly on live websites with online multi-turn RL, delivering a strong open-source system that rivals proprietary agents.
Rui Yang, Qianhui Wu, Yuxi Chen, Hao Bai, Wenlin Yao, Hao Cheng, Baolin Peng, Huan Zhang, Tong Zhang, Jianfeng GaoRead
VoLo teaches a vision-language model to act like a robot conductor, dynamically steering tools and recovery steps during long-horizon manipulation instead of waiting for each action to finish.
Siyi Chen, Hugo Hadfield, Alex Zook, Mikaela Angelina Uy, Chan Hee Song, Erwin Coumans, Xuning Yang, Faisal Ladhak, Qing Qu, Stan Birchfield, Jonathan Tremblay, Valts BlukisRead
Load more videos
397
Rl
EvoTrainer is an autonomous RL system that not only improves LLM policies but also evolves the training process itself to better diagnose and fix failures in agentic tasks.
Guhong Chen, Yingcheng Shi, Yongbin Li, Binhua Li, Xander Xu, Hu Wei, Shiwen Ni, Min Yang, Jieping YeRead
This paper introduces an information-bottleneck-guided way to balance exploration and exploitation in LLM reinforcement learning, boosting reasoning performance with smarter tree-based sampling.
Hao Jiang, Shurui Li, Tianpeng Bu, Bowen Xu, Xin Liu, Qihua Chen, Hongtao Duan, Lulu Hu, Bin Yang, Minying ZhangRead
This paper introduces Pion, a faster Muon-like optimizer that keeps the useful large-gradient directions while suppressing noisy ones, boosting training for robot control and reinforcement-learning math models.
Chongyu Fan, Gaowen Liu, Mingyi Hong, Ramana Rao Kompella, Sijia LiuRead
398
Multimodal
InstructSAM lets users give natural-language instructions to segment multiple instances in one shot by bridging a vision-language model with SAM3 through learnable instance queries.
Yuqian Yuan, Wentong Li, Zhaocheng Li, Yutong Lin, Juncheng Li, Siliang Tang, Jun Xiao, Yueting Zhuang, Wenqiao ZhangRead
MemDreamer lets vision-language models understand hours-long videos by storing them in a hierarchical graph memory and retrieving only the most relevant pieces during reasoning.
Cong Chen, Guo Gan, Kaixiang Ji, ChaoYang Zhang, Zhen Yang, Guangming Yao, Hao Chen, Jingdong Chen, Yi Yuan, Chunhua ShenRead
SMART turns ordinary single-vector embedding models into stronger multimodal retrievers by reusing their hidden states for late-interaction retrieval without needing full retraining.
Jianrui Zhang, Hyun Jung Lee, Sukanta Ganguly, Tae-Eui Kam, Donghyun Kim, Yong Jae LeeRead
399
Cv
LongLive-RAG turns previously generated video latents into a searchable memory, helping autoregressive models keep long videos more consistent instead of drifting over time.
Qixin Hu, Shuai Yang, Wei Huang, Song Han, Yukang ChenRead
RayDer turns self-supervised novel view synthesis from brittle multi-module training into a single scalable transformer that learns from real-world video and performs strongly zero-shot on many benchmarks.
Ulrich Prestel, Stefan Andreas Baumann, Nick Stracke, Björn OmmerRead
YOLO26 is a faster, cleaner evolution of the YOLO family that aims to deliver real-time detection and related vision tasks with better accuracy-latency tradeoffs and simpler deployment.
Glenn Jocher, Jing Qiu, Mengyu Liu, Shuai Lyu, Fatih Cagatay Akyon, Muhammet Esat KalfaogluRead
Load more videos
400
Nlp
ESC-Skills turns emotional support chat into a set of explicit, self-improving skills so the assistant can respond more safely, consistently, and helpfully.
Jie Zhu, Huaixia Dou, Shuo Jiang, Junhui Li, Lifan Guo, Feng Chen, Chi Zhang, Fang KongRead
This paper shows that giving LLMs step-by-step grammatical reasoning can boost translation quality for very low-resource languages, especially when the reasoning is used at inference time rather than learned from scratch.
Renhao Pei, Yihong Liu, Sampo Pyysalo, Hinrich Schütze, Shaoxiong JiRead
03
STREAM: A Data-Centric Framework for Mining High-Value Task-Oriented Dialogues from Streaming Media
STREAM turns live streams and short videos into a large task-oriented dialogue dataset, aiming to fill the gap in realistic domain conversations for training and evaluating LLMs.
Liang Xue, Haoyu Liu, Cheng Wang, Pengyu Chen, Haozhuo Zheng, Yang LiuRead
401
Efficiency
This paper introduces ABC-Bench, a benchmark that tests whether LLM agents can carry out biology tasks with biosecurity implications, and shows they can already perform surprisingly well, even in some wet-lab validated cases.
Andrew Bo Liu, Samira Nedungadi, Bryce Cai, Alex Kleinman, Harmon Bhasin, Seth DonougheRead
The paper argues that AI agents should stop recomputing the same prompt over and over, and instead buy and reuse precomputed KV caches to make LLM serving much cheaper.
Luoyuan ZhangRead
MiniMax Sparse Attention makes ultra-long-context LLMs much faster by combining blockwise sparse retrieval with GPU-friendly kernels, delivering major inference speedups at 1M-token scale.
Xunhao Lai, Weiqi Xu, Yufeng Yang, Qiaorui Chen, Yang Xu, Lunbin Zeng, Xiaolong Li, Haohai Sun, Haichao Zhu, Vito Zhang, Pengyu ZhaoRead
402
Llm
This work shows that for on-policy distillation, letting the student only roll out the first few response tokens can actually outperform full rollouts, while being faster and more stable to train.
Zhou Ziheng, Jiaqi Li, Huacong Tang, Ying Nian Wu, Demetri TerzopoulosRead
This study shows that online accusations of “AI slop” are growing fast, but they often function more as social gatekeeping than as real detection of LLM-generated writing.
Jason Miklian, John E. KatsosRead
This paper improves LLM distillation by fixing bad reasoning prefixes at the trajectory level, helping smaller models learn better from teacher-guided rollouts.
Li Jiang, Haoran Xu, Yichuan Ding, Amy ZhangRead
Load more videos
403
Diffusion
This paper introduces a single diffusion model that can both generate images and perform super-resolution by starting the denoising process at different scales.
Zixin Jessie Chen, Zhuo Chen, Archer Wang, Jeff Gore, William T. Freeman, Congyue Deng, Marin SoljačićRead
This paper lets Stable Diffusion take both a text prompt and a reference image at inference time, so you can steer generation toward a desired style or visual concept without retraining.
Agata Żywot, Iason Skylitsis, Thijmen Nijdam, Zoe Tzifa-Kratira, Derck Prinzhorn, Konrad Szewczyk, Aritra BhowmikRead
This paper delivers a fully open, highly competitive text-to-image diffusion model plus a carefully tested training recipe that could become a new starting point for open generative AI research.
Boya Zeng, Tianze Luo, Shu Pu, Jucheng Shen, Taiming Lu, Gabriel Sarch, Zhuang LiuRead
404
Generative Model
This paper proposes a new way to turn images into discrete tokens by quantizing channels instead of patches, enabling a fresh autoregressive image generator that paints details progressively from coarse structure to fine texture.
Wei Song, Tianhang Wang, Yitong Chen, Tong Zhang, Zuxuan Wu, Ming Li, Jiaqi Wang, Kaicheng YuRead
This paper teaches video generators to follow camera motion more faithfully by rewarding them with geometry-based feedback instead of only supervised examples.
Zizun Li, Haoyu Guo, Runzhe Teng, Chunhua Shen, Tong HeRead
This paper turns image-quality debugging for text-to-image models into a structured task that pinpoints what is wrong, where it is wrong, why, and how important it is, then uses those defects to improve the generator.
Huaisong Zhang, Hao Yu, Yuxuan Zhang, Jiahe Wang, Xinrui Chen, Haoxiang Cao, Feng Lu, Wendong Zhang, Changqian Yu, Chun YuanRead
405
Foundation Model
APEX adapts foundation models to wireless network telemetry, showing that a smaller network-native transformer can forecast AP behavior and detect anomalies better than generic time-series baselines.
Swadhin Pradhan, Niloo Bahadori, Peiman AminiRead
HRM-Text claims you can pretrain a capable language model far more cheaply by swapping Transformers for a brain-inspired recurrent design and training on instruction pairs instead of internet-scale raw text.
Guan Wang, Changling Liu, Chenyu Wang, Cai Zhou, Yuhao Sun, Yifei Wu, Shuai Zhen, Luca Scimeca, Yasin Abbasi YadkoriRead
This paper shows that a frozen time-series foundation model can provide strong, data-efficient features for predicting when industrial equipment will fail.
Amir El-Ghoussani, Michele De Vita, Ronald Naumann, Valiseios BelagiannisRead
Load more videos
406
Robotics
Dream.exe tests whether video generators can dream motions that actually work in a robot simulator, turning pretty videos into a practical measure of physical understanding.
Rui Zhao, Kaiming Yang, Jifeng Zhu, Siyang Chen, Ziqi Wang, Weijia Wu, Kevin Qinghong Lin, Heng Wang, Mike Zheng ShouRead
OASIS shows that carefully built simulated data can sometimes beat real teleoperation for humanoid robot loco-manipulation on the physical robot.
Zehao Yu, Jiakun Zheng, Weiji Xie, Jiyuan Shi, Chenyun Zhang, Chenjia Bai, Xuelong LiRead
WIZARD teaches a robot policy to adapt to new tasks in one shot by predicting the right weight updates from a demo video and instruction, avoiding task-specific fine-tuning.
Christian Bianchi, Siamak Yousefi, Alessio Sampieri, Andrea Roberti, Luca Rigazio, Fabio Galasso, Luca FrancoRead
407
Agent
This paper studies how AI agents turn raw experience into reusable skills, finds that these skills can help but also hurt in surprising ways, and proposes a meta-skill to make skill extraction more reliable.
Zisu Huang, Jingwen Xu, Yifan Yang, Ziyang Gong, Qihao Yang, Muzhao Tian, Xiaohua Wang, Changze Lv, Xuemei Gao, Qi Dai, Bei Liu, Kai Qiu, Xue Yang, Dongdong Chen, Xiaoqing Zheng, Chong LuoRead
QUEST shows that fully synthetic training can produce open deep research agents that rival proprietary systems on long-horizon search, citation grounding, and synthesis tasks.
Jian Xie, Tianhe Lin, Zilu Wang, Yuting Ning, Yuekun Yao, Tianci Xue, Zhehao Zhang, Zhongyang Li, Kai Zhang, Yufan Wu, Shijie Chen, Boyu Gou, Mingzhe Han, Yifei Wang, Vint Lee, Xinpeng Wei, Xiangjun Wang, Yu Su, Huan SunRead
ScientistOne is an autonomous research agent that tries to make every claim traceable, aiming to stop hallucinated citations, unverifiable results, and mismatches between code and papers.
Rui Meng, Bhavana Dalvi Mishra, Jiefeng Chen, Chun-Liang Li, Palash Goyal, Mihir Parmar, Yiwen Song, Yale Song, Rajarishi Sinha, Parthasarathy Ranganathan, Burak Gokturk, Jinsung Yoon, Tomas PfisterRead
408
Rl
This paper proposes DistIL, a new reinforcement-learning approach that turns rich feedback like traces and corrections into better policies with stronger theory and better performance than common self-distillation baselines.
Rishabh Agrawal, Jacob Fein-Ashley, Paria RashidinejadRead
TRON is a scalable online training environment that generates fresh, automatically verifiable visual reasoning tasks on demand, helping multimodal models learn harder reasoning skills more efficiently.
Tianze Yang, Yucheng Shi, Ruitong Sun, Jingyuan Huang, Ninghao Liu, Jin SunRead
The paper shows that you can build bigger reasoning-training tasks by composing smaller verifiable environments like LEGO bricks, improving LLM reasoning generalization with far less manual environment design.
Hao Xiang, Qiaoyu Tang, Le Yu, Yaojie Lu, Xianpei Han, Ben He, Le Sun, Bowen Yu, Peng Wang, Hongyu Lin, Dayiheng LiuRead
Load more videos
409
Cv
This paper makes 3D object understanding more fine-grained by teaching a model to infer consistent parts and shapes across object categories from a single image, while also releasing a new large benchmark to measure that skill.
Leonhard Sommer, Artur Jesslen, Basavaraj Sunagad, Adam KortylewskiRead
CubePart lets users generate 3D objects from text while specifying the parts they want, making AI-generated assets more useful for games, animation, and simulation.
Yiheng Zhu, Kangle Deng, Jean-Philippe Fauconnier, Inaki Navarro, Daiqing Li, Ava Pun, Yinan Zhang, Peiye Zhuang, Xiaoxia Sun, Maneesh Agrawala, Kiran Bhat, Tinghui ZhouRead
This paper introduces a large-scale diffusion model that can generate and edit editable layered images, making layered visual composition much faster and more practical.
Zhicong Tang, Zhao Zhang, Jingye Chen, Mohan Zhou, Yifan Pu, Yuchi Liu, Yalong Bai, Ethan Smith, Yuhui YuanRead
410
Nlp
This paper turns millions of NSF award abstracts into a new dataset for extracting scientific claims and research plans, aiming to power better claim verification and science analysis.
Delip Rao, Weiqiu You, Eric Wong, Chris Callison-BurchRead
OmniRetrieval aims to be a universal translator for search, sending each query to the right mix of text, tables, and graph databases instead of forcing everything into one format.
Jinheon Baek, Soyeong Jeong, Sangwoo Park, Woongyeong Yeo, Minki Kang, Patara Trirat, Heejun Lee, Sung Ju HwangRead
Xetrieval explains why dense retrievers rank documents highly by turning opaque embeddings into sparse human-readable features and using them to trace retrieval decisions.
Zhixin Cai, Jun Bai, Yang Liu, Jiaqi Li, Yichi Zhang, Taichuan Li, Zhuofan Chen, Zixia Jia, Zilong Zheng, Wenge RongRead
411
Llm
This paper introduces a new way for language models to improve themselves by both evolving candidate answers forward and breaking problems into checkable subgoals backward, boosting performance on hard reasoning tasks.
Guowei Xu, Zhenting Qi, Huangyuan Su, Weirui Ye, Himabindu Lakkaraju, Sham M. Kakade, Yilun DuRead
Skill-RM turns reward evaluation into an agent-like skill that dynamically combines rules, references, and rubrics to judge LLM outputs more consistently.
Tao Chen, Gangwei Jiang, Pengyu Cheng, Siyuan Huang, Yihao Liu, Jingwei Ni, Jiaqi Guo, Mengyu Zhou, Kai Tang, Junling Liu, Qinliang Su, Xiaoxi Jiang, Guanjun JiangRead
This paper learns when to trust weak labels so a stronger model can turn imperfect supervision into nearly lossless training and even bootstrap itself repeatedly.
Arda Uzunoglu, Alvin Zhang, Daniel KhashabiRead
Load more videos
412
Efficiency
This paper introduces latent context models that compress long prompts into compact representations, making long-context language models faster and more memory-efficient without sacrificing as much quality.
Ang Li, Sean McLeish, Haozhe Chen, Nimit Kalra, Zaiqian Chen, Artem Gazizov, Venkata Anoop Suhas Kumar Morisetty, Bhavya Kailkhura, Harshitha Menon, Zhuang Liu, Brian R. Bartoldson, Tom Goldstein, Sanae Lotfi, Micah Goldblum, Pavel IzmailovRead
MobileMoE shows that sparse mixture-of-experts language models can run efficiently on phones, delivering better speed-memory tradeoffs than dense on-device LLMs.
Yanbei Chen, Hanxian Huang, Ernie Chang, Jacob Szwejbka, Digant Desai, Zechun Liu, Vikas Chandra, Raghuraman KrishnamoorthiRead
This paper shows how to use scaling laws plus architecture choices like attention design to build language models that are both cheaper to run and more accurate.
Song Bian, Tao Yu, Shivaram Venkataraman, Youngsuk ParkRead
413
Diffusion
This paper finds that early diffusion steps can preserve better physics than later ones, and introduces a lightweight method to lock in that motion prior while still producing high-quality video.
Woojung Han, Seil Kang, Youngjun Jun, Min-Hung Chen, Fu-En Yang, Seong Jae HwangRead
RT-Lynx speeds up diffusion transformers by sparsifying activations instead of weights, preserving image quality while making generation faster.
Xing Cong, Hanlin Tang, Kan Liu, Lan Tao, Lin Qu, Chenhao XieRead
This paper argues that the way diffusion transformers pass information across layers is suboptimal, and introduces a new residual-routing method that can speed up training and improve image quality.
Chao Xu, Maohua Li, Qirui Li, Yixuan Xu, Yanke Zhou, Yunhe Li, Cuifeng Shen, Hanlin Tang, Kan Liu, Tao Lan, Lin Qu, Shao-Qun ZhangRead
414
Generative Model
Dream-Cubed turns Minecraft worlds into a massive training ground for learning controllable 3D generative models that can fill in, extend, and create block-based environments interactively.
Tim Merino, Sam Earle, Ryunosuke Iwai, Julian Togelius, Edoardo CetinRead
StreamForce is a real-time video generation system that can instantly react to continuous force inputs, aiming to make generated motion both physically responsive and visually realistic.
Hanhui Wang, Yiming Xie, Haiwen Feng, Zhaoyang Lv, Shenlong Wang, Huaizu JiangRead
03
VibeToken: Scaling 1D Image Tokenizers and Autoregressive Models for Dynamic Resolution Generations
VibeToken is a new image tokenization and generation approach that lets autoregressive models create high-resolution images at arbitrary aspect ratios with far less compute.
Maitreya Patel, Jingtao Li, Weiming Zhuang, Yezhou Yang, Lingjuan LvRead
Load more videos
415
Llm
A comprehensive primer that maps the emerging landscape of post-training reasoning data and explains what makes it work for modern large language models.
Yaoming Li, Guangxiang Zhao, Qilong Shi, Lin Sun, Xiangzheng Zhang, Tong YangRead
This paper teaches an LLM to think more efficiently by steering its reasoning step-by-step with a controller that balances accuracy and token budget.
Yu Xia, Zhouhang Xie, Xin Xu, Byungkyu Kang, Prarit Lamba, Xiang Gao, Julian McAuleyRead
This paper treats LLM reasoning like an economy, using shadow prices to decide which queries deserve more compute so limited budget yields the best overall accuracy.
Xu Wan, Speed Zhu, Jianwei Cai, Guang Chen, XiMing Huang, Wiggin Zhou, Mingyang SunRead
416
Transformer
SISA is a new attention design that lets state-space models guide which tokens matter inside softmax attention, aiming to combine global retrieval with better prioritization in language models.
Soohyeong Shin, Yeongwook YangRead
This paper turns attention from pairwise token matching into a functional map over continuous fields, aiming to make transformer-style operator learning more compact, resolution-invariant, and better suited for PDEs and other scientific tasks.
Jiefang Xiao, Maolin Gao, Simon Weber, Guandao Yang, Daniel CremersRead
This paper makes transformer feedforward layers both smaller and more interpretable by replacing dense expert blocks with sparsely selected single-neuron linear experts.
Simon SchugRead
417
Multimodal
M^3Eval is a new benchmark that tests whether multimodal models can actually remember what they see and hear in long videos, revealing surprising weaknesses in how they store and separate information.
Jie Huang, Ruixun Liu, Sirui Sun, Xinyi Yang, Yin Li, Yixin Zhu, Yiwu ZhongRead
This work brings 3D vision-language models into real-time streaming settings, letting them understand changing scenes on the fly while staying efficient and grounded in geometry.
Hanxun Yu, Xuan Qu, Lei Ke, Boqiang Zhang, Yuxin Wang, Jianke Zhu, Dong YuRead
Video2LoRA compresses a video into a tiny adapter so a vision-language model can answer later without carrying any visual tokens at query time.
Manan Suri, Sarvesh Baskar, Dinesh ManochaRead
Load more videos
418
Cv
This paper shows how to train image and video editors without paired before-and-after data by bootstrapping guidance from the model itself and using flow matching to preserve structure.
Yoad Tewel, Yuval Atzmon, Gal Chechik, Lior WolfRead
This paper uses multimodal foundation models to fabricate realistic ground-truth images from degraded photos, creating a 100K-pair dataset that helps restoration models work better on real-world images.
Xiangtao Kong, Jixin Zhao, Lingchen Sun, Rongyuan Wu, Lei ZhangRead
This paper tackles a harder version of video grounding where one text query can match multiple separate video segments, and it builds a benchmark plus training method to make multimodal models much better at finding all of them.
Qi Xu, Yue Tan, Shihao Chen, Jiahao Meng, Anna Wang, Shunping Ji, Hao Fei, Jason LiRead
419
Nlp
This paper builds a benchmark and extractive QA system that answers research questions by copying exact text from ACL papers, aiming to reduce hallucinations in AI-assisted literature review.
Gábor Recski, Szilveszter Tóth, Nadia Verdha, István Boros, Ádám KovácsRead
This paper shows how to cheaply adapt a multilingual embedding model to Turkish by redesigning its tokenizer and distilling it offline, achieving strong retrieval-quality gains at low training cost.
M. Ali Bayram, Banu Diri, Savaş YıldırımRead
This paper builds a large, carefully translated German pretraining corpus and shows that better curated data can noticeably improve German language models.
Maurice Kraus, Ruben Härle, Sebastian Sztwiertnia, Abbas Goher Khan, Mehdi Ali, Michael Fromm, Kristian KerstingRead
420
Efficiency
Domino speeds up LLM generation by using a parallel draft model plus a lightweight causal refinement step, delivering faster speculative decoding without sacrificing too much draft quality.
Jianuo Huang, Yaojie Zhang, Qituan Zhang, Hao Lin, Hanlin Xu, Linfeng ZhangRead
Draft-OPD improves speculative decoding by training draft models on the kinds of errors they actually make at inference time, yielding faster LLM generation with better acceptance rates.
Haodi Lei, Yafu Li, Haoran Zhang, Shunkai Zhang, Qianjia Cheng, Xiaoye Qu, Ganqu Cui, Bowen Zhou, Ning Ding, Yun Luo, Yu ChengRead
PassNet reframes LLM-based compiler optimization as graph pass generation, backing it with a large dataset and benchmark that show LLMs can meaningfully speed up real-world workloads.
Yiqun Liu, Yingsheng Wu, Ruqi Yang, Enrong Zheng, Honglei Qiu, Sijun He, Tai Liang, Jingjing Wu, Yuhan Zhou, Yiwei Zhang, Dongyan Chen, Weihan Yi, Xinqi Li, Siqi BaoRead
Load more videos
421
Diffusion
DEMON turns a diffusion music generator into a live, playable instrument by speeding up and restructuring how denoising controls take effect in real time.
Ryan FosdickRead
This paper improves image-to-image translation by splitting diffusion into two stages so one part aligns domains with noise and the other learns the actual semantic change more efficiently.
Ziyue Lin, Jiahe Hou, Hongyu Xia, Xinrui Xie, Feifei Wang, Yuyin Zhou, Wei Wang, Jiawei Liu, Liangqiong QuRead
This paper makes real-time video editing practical on a consumer GPU by pairing a diffusion-transformer model with a new consistency training trick and aggressive Blackwell-specific optimization.
Yuyang Zhao, Yicheng Pan, Qiyuan He, Jincheng Yu, Junsong Chen, Tian Ye, Haozhe Liu, Enze Xie, Song HanRead
422
Generative Model
This paper introduces a new memory system that lets video generators keep producing coherent frames for extremely long durations in real time, pushing toward truly infinite video generation.
Yuxuan Bian, Zeyue Xue, Songchun Zhang, Shiyi Zhang, Weiyang Jin, Yaowei Li, Junhao Zhuang, Haoran Li, Jie Huang, Haoyang Huang, Nan Duan, Qiang XuRead
Lens is a more compute-efficient text-to-image model that uses richer data, smarter batching, and post-training tricks to match or beat larger systems while generating images faster.
Dong Chen, Fangyun Wei, Ziyu Wan, Dongdong Chen, Jiawei Zhang, Jinjing Zhao, Sirui Zhang, Yang Yue, Zhiyang Liang, Baining Guo, Chong Luo, Jianmin Bao, Ji Li, Lei Shi, Qinhong Yang, Xiuyu Wu, Xuelu Feng, Yan Lu, Yanchen Dong, Yitong Wang, Yunuo ChenRead
LoomVideo is a faster, smaller video generation-and-editing model that uses a clever conditioning trick to avoid expensive token concatenation while still handling multimodal inputs.
Jianzong Wu, Hao Lian, Jiongfan Yang, Dachao Hao, Ye Tian, Yunhai Tong, Jingyuan Zhu, Biaolong Chen, Qiaosong Qi, Aixi Zhang, Wanggui He, Mushui Liu, Jinlong Liu, Hao JiangRead
423
Transformer
This paper asks whether transformers really need separate query, key, and value projections—and finds that tying some of them can preserve much of the quality while greatly reducing memory use.
Ali Kayyam, Anusha Madan Gopal, M Anthony LewisRead
This paper shows that the tiny scale vectors inside LLM normalization layers matter a lot for training, and proposes simple changes that make large language models learn better with almost no extra cost.
Mingze Wang, Shuchen Zhu, Yuxin Fang, Binghui Li, Kai Shen, Shu ZhongRead
Parallax is a new attention design for language models that makes local linear attention practical at scale and claims better training quality and faster decoding than FlashAttention-style baselines.
Yifei Zuo, Dhruv Pai, Zhichen Zeng, Alec Dewulf, Shuming Hu, Zhaoran WangRead
Load more videos
424
Llm
This paper shows how to compress expensive multi-agent debate into one language model, making reasoning cheaper and exposing interpretable internal agent-like directions that can also be steered for safety control.
John Seon Keun Yi, Aaron Mueller, Dokyun LeeRead
This paper shows that averaging outputs from a few LLMs can wash away text watermarks, exposing a major weakness in current AI-generated text detection methods.
Zhihao Wu, Gracia Gong, Qinglin Zhu, Yudong Chen, Runcong ZhaoRead
PEFT-Arena reframes LLM finetuning as a balance between learning new tasks and preserving old abilities, revealing which parameter-efficient methods best avoid forgetting.
Yangyi Huang, Ruotian Peng, Zeju Qiu, Jiale Kang, Yandong Wen, Bernhard Schölkopf, Weiyang LiuRead
425
Foundation Model
This paper maps the fast-growing landscape of AI in dentistry, showing how general-purpose and dental-specific foundation models can work together while highlighting key safety and data gaps.
Sema Helali, Lina Abu Nada, Sausan Al Kawas, Alaa Abd-Alrazaq, Faleh Tamimi, Rafat DamsehRead
The paper asks whether vision-language models or video generation models better support spatial intelligence, and finds they encode complementary strengths that can be combined for stronger representations.
Haozhan Shen, Tiancheng Zhao, Kangjia Zhao, Jianwei YinRead
The paper explains why bigger models can learn rare, complex tasks that smaller ones miss: they have enough capacity to avoid overwriting those features while learning the common ones.
Jing Huang, Daniel Wurgaft, Rachit Bansal, Laura Ruis, Naomi Saphra, David Alvarez-Melis, Andrew Kyle Lampinen, Christopher Potts, Ekdeep Singh LubanaRead
426
Foundation Model
NITP tweaks LLM pretraining by making models predict not just the next token, but its hidden semantic representation too, improving downstream performance with little extra cost.
Xiangdong Zhang, Debing Zhang, Shaofeng Zhang, Xiaohan Qin, Yu Cheng, Junchi YanRead
This paper argues that small adapters on top of foundation models could become persistent personal models at massive scale, enabling individualized behavior without retraining huge models.
Mind Lab, :, Song Cao, Vic Cao, Kaijie Chen, Bunny Fan, Hera Feng, Huan Feng, Arthur Fu, Jun Gao, Hongquan Gu, Aaron Guan, Mutian Hong, Hailee Hou, Peixuan Hua, Charles Huang, Miles Jiang, Nora Jiang, Yuyi Jiang, Autumn Jin, Fancy Kong, Kyrie Lei, Alexy Li, Dawn Li, Ray Li, Theo Li, Wenhao Li, Jiayi Lin, Domini Liu, Heshan Liu, Kairus Liu, Logan Liu, Maeve Luo, Runism Lv, Pony Ma, Verity Niu, Anson Qiu, Vincent Wang, Maxwell Yao, Regis Ye, Wenlin Ye, Yanying Ye, Josh Ying, Danney Zeng, Salmon Zhan, Anya Zhang, Ruijia Zhang, Shiyang Zhang, Sueky Zhang, Ya Zhang, Wei Zhao, Ada Zhou, Sizer Zhou, Xinyue Zhu, Murphy ZhuangRead
This paper proposes a single scaling-law formula that better predicts how neural networks behave as model size, data, compute, and other factors all change at once.
Ethan Caballero, Priyank Jaini, David Krueger, Irina RishRead
Load more videos
427
Hardware
This paper presents CXL-ClusterSim, a new simulation framework for studying how pooled, disaggregated memory over CXL could make large-scale AI and cloud systems more efficient.
Kaustav Goswami, Maryam Babaie, Hoa Nguyen, Venkatesh Akella, Jason Lowe-PowerRead
This paper shows that language models can act like smart GPU performance predictors, helping search systems test more kernel ideas while spending less time on expensive hardware runs.
Zaid Khan, Justin Chih-Yao Chen, Jaemin Cho, Elias Stengel-Eskin, Mohit BansalRead
This paper introduces a cheaper, scalable datacenter network design that uses quasi-random graph routing and novel cable-shuffling hardware to beat or match fat-tree performance in production.
Giacomo Bernardi, Ratul Mahajan, C. Seshadhri, Enrico Carlesso, Chinchu Merine Joseph, Saurabh Kumar, Pavan Manikonda, Luiza Popa, Randy Ram, Steven Robinson, Elizabeth TennentRead
428
Safety
This paper argues that AI agents can power a new kind of self-sustaining worm that adapts to each target, spreads across real networks, and uses stolen compute to keep attacking at near-zero marginal cost.
Jonas Guan, Tom Blanchard, Hanna Foerster, Hengrui Jia, Gabriel Huang, Nicolas PapernotRead
This paper shows how persistent LLM agents can be turned into self-spreading worms across platforms, and proposes a formal defense to stop them.
Mingming Zha, Xiaofeng WangRead
This paper shows that chain-of-thought monitoring can look reliable in English but break badly across many languages, exposing a major weakness in current AI safety oversight.
Eric Onyame, Runtao Zhou, Kowshik Thopalli, Bhavya Kailkhura, Chirag AgarwalRead
429
Continual Learning
This paper argues that language models should 'sleep' to consolidate memories and generate synthetic practice, helping them keep learning over time without forgetting.
Ali Behrouz, Farnoosh Hashemi, Vahab MirrokniRead
This paper asks whether LLMs can get better at making synthetic data over time by learning from a stream of past synthesis tasks instead of treating each one in isolation.
Zhenlin Hu, Yan Wang, Zhen Bi, Zihao Xue, Bingyu Zhu, Longtao Huang, Xiongtao Zhang, Zeyu Yang, Zhixuan Chu, Jungang LouRead
This paper shows that the order of pre-training data matters: training large language models on time-ordered text can make them better at knowing which facts were true when, without hurting general language ability.
Pilchen Hippolyte, Fabre Romain, Signe Talla Franck, Perez Patrick, Grave EdouardRead
Load more videos
430
World Model
This paper tests whether LLMs can build spatial mental maps from text across languages, and finds a consistent reasoning cliff that suggests current models struggle to form robust world models.
Zhikai Pan, Chih-Ting Liao, Chunrui Liu, Xi Xiao, Yitong Qiao, Chunlei Meng, Zhangquan Chen, Xin CaoRead
GE-Sim 2.0 is a faster, more useful robotic world simulator that not only predicts video from actions but also helps train and evaluate manipulation policies in a closed loop.
Boxiang Qiu, Liliang Chen, Yue Liao, Nan Wang, Lintao Wang, Jiayi Luo, Wenzhi Zhao, Shengcong Chen, Di Chen, Ye Li, Chen Gao, Shuicheng Yan, Si Liu, Maoqing Yao, Guanghui RenRead
OmniDreams is a real-time AI driving simulator that generates photorealistic future scenes from actions, aiming to make autonomous vehicle testing safer, broader, and more realistic.
NVIDIA, :, Aarti Basant, Amlan Kar, Despoina Paschalidou, Fangyin Wei, Francesco Ferroni, Guillermo Garcia Cobo, Haithem Turki, Huan Ling, Jaewoo Seo, James Lucas, Jay Zhangjie Wu, Jialiang Wang, Jonathan Lorraine, Jun Gao, Kai He, Katarina Tothova, Kevin Xie, Michał Tyszkiewicz, Qi Wu, Riccardo de Lutio, Ruilong Li, Sanja Fidler, Seung Wook Kim, Tianchang Shen, Tianshi Cao, Tobias Pfaff, William Lew, Xindi Wu, Xuanchi Ren, Yifan Lu, Yuxuan Zhang, Zan Gojcic, Zian WangRead
431
Embodied Ai
01
Qwen-VLA: Unifying Vision-Language-Action Modeling across Tasks, Environments, and Robot Embodiments
Qwen-VLA is a single robot brain that can see, reason, and act across different tasks and robot bodies by training one vision-language model to handle navigation, manipulation, and trajectory prediction together.
Qiuyue Wang, Mingsheng Li, Jian Guan, Jinhui Ye, Sicheng Xie, Yitao Liu, Junhao Chen, Zhixuan Liang, Jie Zhang, Xintong Hu, Xuhong Huang, Pei Lin, Junyang Lin, Dayiheng Liu, Shuai Bai, Jingren Zhou, Jiazhao Zhang, Haoqi Yuan, Gengze Zhou, Hang Yin, Ye Wang, Yiyang Huang, Zixing Lei, Wujian Peng, Delin Chen, Yingming Zheng, Jingyang Fan, Xianwei Zhuang, Xin Zhou, Haoyang Li, Anzhe Chen, Tong Zhang, Xuejing Liu, Yuchong Sun, Ruizhe Chen, Zhaohai Li, Chenxu Lü, Zhibo Yang, Tao Yu, Xionghui ChenRead
This work asks whether foundation models can actively move to match a target view, and shows that new training methods can dramatically improve their ability to navigate in 3D indoor environments.
Liyang Li, Muzhi Zhu, Zhiyue Zhao, Hengyu Zhao, Ke Liu, Linhao Zhong, Hao Chen, Chunhua ShenRead
432
Speech
This paper builds a privacy-preserving speech translation system that sends only compressed features from device to cloud, cutting bandwidth while scaling translation across 45 languages.
Yexing Du, Kaiyuan Liu, Youcheng Pan, Bo Yang, Ming Liu, Bing Qin, Yang XiangRead
HoliTok is a new speech tokenization approach that compresses audio into a compact latent sequence usable for both speech generation and understanding, helping unify speech AI systems.
Bohan Li, Shi Lian, Hankun Wang, Yiwei Guo, Yu Xi, Zhihan Li, Da Zheng, Colin Zhang, Kai YuRead
This paper proposes a new way to generate combined speech and sound directly from natural language using an LLM-based planner and a new benchmark for composite audio.
Yuyue Wang, Xihua Wang, Xin Cheng, Yijing Chen, Ruihua SongRead
Load more videos
433
Foundation Model
LLMSurgeon tries to infer what data an LLM was trained on just by reading its outputs, offering a new way to audit foundation models without access to their training sets.
Yaxin Luo, Jiacheng Cui, Xiaohan Zhao, Xinyi Shang, Jiacheng Liu, Xinyue Bi, Zhaoyi Li, Zhiqiang ShenRead
This paper treats LLM training like information flowing through a noisy channel, offering a new scaling law that explains why bigger models and more data can sometimes make performance worse.
Xu Ouyang, Deyi Liu, Yuhang Cai, Jing Liu, Yuan Yang, Chen Zheng, Thomas Hartvigsen, Yiyuan MaRead
434
Multimodal
LLaVA-OneVision-2 is a next-generation multimodal model that compresses video more intelligently to better understand, ground, and reason over long visual content.
Xiang An, Yin Xie, Feilong Tang, Yunyao Yan, Huajie Tan, Didi Zhu, Changrui Chen, Xiuwei Zhao, Bin Qin, Kaicheng Yang, Yifei Shen, Yuanhan Zhang, Kaichen Zhang, Wenkang Zhang, Zheng Cheng, Nansen Zhang, Chunsheng Wu, Chunjiang Ge, Zimin Ran, Dehua Song, Chunyuan Li, Shikun Feng, Ming Hu, Zhangquan Chen, Junbo Niu, Bo Li, Ziyong Feng, Ziwei Liu, Zongyuan Ge, Jiankang DengRead
A CLIP-like foundation model for ECGs learns from millions of ECG-report pairs and can help detect not only common rhythm problems but also rare and subtle cardiovascular diseases.
Ziqing Yu, Yuhui Tao, Jiayu Huo, Lei Pan, Zilong Xiao, Juecheng Chen, Xiao Li, Jianxuan Li, You Zhou, Zhixing Li, Cong Wang, Beijian Zhang, Chen Chen, Hongyang Lu, Konstantinos Patlatzoglou, Daniel B. Kramer, Jonathan W. Waks, Yangang Su, Fu Siong Ng, Shuo Wang, Yixiu Liang, Junbo GeRead
This paper shows that adding simple procedurally generated geometric tasks can significantly improve multimodal language models’ ability to understand spatial and fine-grained visual details.
Rim Assouel, Amir Bar, Michal Drozdzal, Adriana Romero-SorianoRead
435
Benchmark
This work builds the largest real-world 9D pose dataset to date by mining millions of object-centric videos, enabling better training and evaluation of pose models for everyday objects.
Leonhard Sommer, Emil Akopyan, Adam KortylewskiRead
HorizonStream is a new streaming Transformer that helps 3D reconstruction stay stable over extremely long video sequences by separating short-range matching from long-range geometric memory.
Chong Cheng, Peilin Tao, Nanjie Yao, Guanzhi Ding, Xianda Chen, Yuansen Du, Xiaoyang Guo, Wei Yin, Weiqiang Ren, Qian Zhang, Zhengqing Chen, Hao WangRead
VideoMLA makes minute-scale autoregressive video diffusion much cheaper in memory by compressing the KV cache with a low-rank latent representation, enabling longer video rollouts with better throughput and strong quality.
Hidir Yesiltepe, Jiazhen Hu, Tuna Han Salih Meral, Adil Kaan Akan, Kaan Oktay, Hoda Eldardiry, Pinar YanardagRead
Load more videos
436
Scientific Ai
This work turns thousands of AI agents loose on real math textbooks to build a massive, machine-checked Lean library, pushing automated formalization from a niche tool toward practical scale.
Ahmad Rammal, Niket Patel, Fabian Gloeckle, Amaury Hayat, Julia Kempe, Remi Munos, Charles Arnal, Vivien CabannesRead
This paper separates what Transformers know about meaning and word order into distinct streams, revealing how positional structure is stored and showing that the split can improve representations.
Pierre-Antoine Lequeu, Camille Barboule, Benjamin PiwowarskiRead
ObsCast shows that high-resolution weather forecasts can be learned directly from observations, bypassing traditional numerical weather models while still delivering state-of-the-art regional predictions.
Pengcheng Zhao, Siqi Xiang, Weixin Jin, Zekun Ni, Jiang Bian, Zuliang Fang, Hongyu Sun, Bin Zhang, Richard E. Turner, Jonathan Weyn, Haiyu Dong, Kit Thambiratnam, Qi ZhangRead
437
Neural Network
This paper develops a new theory for how wide Bayesian neural networks generalize, showing that finite-width effects can be captured with a renormalized kernel model that matches experiments reasonably well.
Paolo Baglioni, Christian Keup, Vincenzo Zimbardo, Rosalba Pacelli, Alessandro Vezzani, Raffaella Burioni, Pietro RotondoRead
Oryx is a hybrid language model that dynamically alternates between attention and recurrent mixing within a sequence, aiming to get both strong long-context performance and efficient generation.
Kevin Y. Li, Asher Trockman, Ananda Theertha Suresh, Ziteng SunRead
This paper shows that transformers can go beyond point estimates and learn full Bayesian predictive distributions in context, with theory explaining why architecture choices like normalization and attention depth matter.
Gyeonghun Kang, Changwoo J. Lee, Xiang ChengRead
438
Nlp
This paper argues that children may learn subject-verb agreement by exploiting word-pair statistics in their input, and shows neural networks can learn the same way under the right data conditions.
Claire Hobbs, R. Thomas McCoyRead
EmoTrack uses LLM-derived clinical signals plus structured transcript embeddings to track depression severity across counseling sessions more robustly, especially when longitudinal context is available.
Zhaomin Wu, Jiayi Li, Bingsheng HeRead
ICICLE lets generative retrievers add new documents on the fly by using in-context document IDs instead of retraining the model each time.
Yu-Chen Den, Yung-Yu Shih, Zhi Rui Tam, Kuan-Yu Chen, Pu-Jen Cheng, Yun-Nung Chen, Eugene YangRead
Load more videos
439
Efficiency
This paper makes attention much cheaper to run by reducing memory traffic, bringing it closer to the theoretical limits of efficiency.
Pál András Papp, Aleksandros Sobczyk, Anastasios ZouziasRead
IndexMem makes long-context LLMs more efficient by learning what to keep in memory and compressing what gets evicted so the model can still recall important information later.
Xintong Yang, Hao Gu, Binxing Xu, Lujun Li, Bei Liu, Jiacheng Liu, Qiyuan Zhu, Sirui Han, Yike GuoRead
This paper shows how to turn large mixture-of-experts language models into smaller dense models that keep much of the performance while being easier and cheaper to deploy.
Junhyuck Kim, Jihun Yun, Haechan Kim, Gyeongman Kim, Joonghyun Bae, Jaewoong ChoRead
440
Diffusion
This paper speeds up reward-guided sampling in discrete diffusion models by learning a cheap twist function, making aligned generation more practical for text, DNA, proteins, and large language models.
Jaihoon Kim, Taehoon Yoon, Prin Phunyaphibarn, Seungjun Kim, Morteza Mardani, Minhyuk SungRead
This paper speeds up discrete diffusion sampling by using Gibbs-style corrections, making text and music generation much faster without extra training.
Yuchen Liang, Ness Shroff, Yingbin LiangRead
This paper makes fast image generators better at following human preferences by blending distillation and reinforcement learning into a new two-stage training recipe.
Yushi Huang, Xiangxin Zhou, Ruoyu Wang, Chi Zhang, Jun Zhang, Tianyu PangRead
441
Generative Model
This paper gives a mathematical guarantee for a new one-step generative modeling method by showing how particle-based drifting can converge and how its errors can be explicitly bounded.
Krishnakumar BalasubramanianRead
This paper shows how to infer personalized time evolution from sparse snapshot data, potentially replacing the need for dense longitudinal tracking in science and medicine.
Christian Lagemann, Kai Lagemann, Steven L. Brunton, Sach MukherjeeRead
This paper uses diffusion models to generate survival times and censoring outcomes, aiming to make continuous-time survival analysis more flexible and accurate without discretizing time.
Stanislav R. Kirpichenko, Andrei V. Konstantinov, Lev V. UtkinRead
Load more videos
442
Efficiency
BigMac is a new multimodal LLM training pipeline that claims to break the usual tradeoff between speed and memory, making training both faster and more memory-friendly.
Zili Zhang, Chengxu Yang, Shenglong Zhang, Chenyu Wang, Yufan Zhang, Tuo Dai, Zhouyang Li, Yuhong Ge, Chao Jin, Xin Jin, Yuliang LiuRead
This paper shows that multilingual reasoning in LLMs can be improved by swapping in stronger mid-layers from an English model, revealing a shared reasoning core beneath language-specific outer layers.
Maxence Lasbordes, Amélie Chatelain, Djamé SeddahRead
This paper shows that teaching a model to verify its own answers better can improve both test-time reasoning loops and training-time self-improvement, leading to stronger performance on hard reasoning tasks.
Chen Henry Wu, Aditi RaghunathanRead
443
Self Supervised
This paper teaches ultrasound models to learn from anatomy, not just pixels, by using self-supervision tied to clinically meaningful structures for more robust medical imaging representations.
Chunzheng Zhu, Yijun Wang, Jianxin Lin, Feng Wang, Hongwei Wang, Lei Zhao, Shengli Li, Kenli LiRead
This paper improves unsupervised anomaly detection in multivariate time series by using dynamic graph contrastive learning to model changing relationships instead of forcing them to stay fixed.
Yunhua Pei, Zixing Song, Jin Zheng, John CartlidgeRead
This paper explains how depth alone can steer neural networks away from neural collapse and toward low-rank softmax-like solutions, revealing a new theoretical bias in training dynamics.
Connall Garrod, Jonathan P. Keating, Christos ThrampoulidisRead
444
Optimizer
This paper shows that a smarter preconditioned optimizer can make physics-informed neural networks far more stable when solving tightly coupled multiphysics problems.
Youngjae Park, Jaemin Kim, Junghwa HongRead
This paper explains why the popular SVRG optimizer generalizes well, not just why it converges, and extends the same theory to related variance-reduced methods.
Yunwen Lei, Zimeng Wang, Xiaoming YuanRead
This paper gives Muon optimizer a physics-style mathematical makeover, showing it as a damped Hamiltonian flow and deriving convergence and mean-field guarantees.
Aratrika Mustafi, Soumya Mukherjee, Bharath K. SriperumbudurRead
Load more videos
445
Neural Network
This paper shows that norm-constrained deep neural networks can learn sparse hierarchical functions efficiently, avoiding the curse of dimensionality by matching the structure of the target.
Shuo Huang, Lorenzo Fiorito, Lorenzo Rosasco, Tomaso PoggioRead
This paper proves that small two-layer ReLU networks can learn one neuron at a time in a saddle-to-saddle process, and it quantifies how close the resulting solution is to the simplest possible interpolator.
James Town, Etienne Boursier, Ben Lewis, Matthias Englert, Ranko LazicRead
By training racing drones against each other in multi-agent self-play, the system learns safer, faster, and more human-compatible maneuvers than single-agent methods.
Ismail Geles, Leonard Bauersfeld, Markus Wulfmeier, Davide ScaramuzzaRead
446
Foundation Model
This paper peels back the internals of tabular foundation models to show how different architectures make predictions, why they become permutation-invariant, and how they fail under targeted attacks.
Marin Biloš, James T. Wilson, Anderson Schneider, Yuriy NevmyvakaRead
This paper shows that several popular sparse autoencoder benchmarks may be unreliable, and argues that the field needs better ways to tell good SAEs from bad ones.
David ChaninRead
This paper introduces a biologically inspired attention mechanism that turns attention scores into stochastic variables to produce uncertainty-aware predictions across several continuous-time tasks.
Waleed Razzaq, Yun-Bo ZhaoRead
447
Code Generation
CoSPlay lets an LLM improve its own code and tests at inference time, using self-generated unit tests and execution signals to boost code generation without ground-truth labels.
Zhangyi Hu, Chenhui Liu, Tian Huang, Jindong Li, Yang Yang, Jiemin Wu, Zining Zhong, Menglin Yang, Yutao YueRead
MobileGym is a scalable, deterministic mobile app simulator and benchmark designed to train and evaluate GUI agents more reliably and at much larger scale.
Dingbang Wu, Rui Hao, Haiyang Wang, Shuzhe Wu, Han Xiao, Zhenghong Li, Bojiang Zhou, Zheng Ju, Zichen Liu, Lue Fan, Zhaoxiang ZhangRead
This paper replaces softmax attention with a hysteresis-based memory model that tracks extrema, aiming to give transformers a more efficient way to remember long histories.
Piotr FrydrychRead
Load more videos
448
Foundation Model
This work introduces a clinically validated lung pathology foundation model that not only performs well across many diagnostic tasks, but also improves pathologist accuracy, speed, and consistency in real-world trials.
Zhengrui Guo, Zhengyu Zhang, Jiabo Ma, Yihui Wang, Fengtao Zhou, Yingxue Xu, Ling Liang, Chenglong Zhao, Qi Xie, Jinbang Li, Shujing Guo, Fangyi Han, Zhijian Cen, Ziyi Liu, Cheng Jin, Junlin Hou, Zhixuan Chen, Yu Cai, Lijuan Qu, Shifu Chen, Yueping Liu, Zhe Wang, Xiuming Zhang, Muyan Cai, Li Liang, Hao ChenRead
This paper shows that a function can be understood three equivalent ways: by how well it can be reconstructed from random samples, by a compact polynomial representation, or by a small neural network.
Balazs SzegedyRead
This study shows that top AI reviewers can already outperform some human reviewers on specific critique quality, but they still have important weaknesses and work best as complements to humans.
Seungone Kim, Dongkeun Yoon, Kiril Gashteovski, Juyoung Suk, Jinheon Baek, Pranjal Aggarwal, Ian Wu, Viktor Zaverkin, Spase Petkoski, Daniel R. Schrider, Ilija Dukovski, Francesco Santini, Biljana Mitreska, Yong Jeong, Kyeongha Kwon, Young Min Sim, Dragana Manasova, Arthur Porto, Biljana Mojsoska, Makoto Takamoto, Marko Shuntov, Ruoqi Liu, Hyunjoo Jenny Lee, Niyazi Ulas Dinç, Yehhyun Jo, Sunkyu Han, Chungwoo Lee, Huishan Li, Esther H. R. Tsai, Ergun Simsek, Khushboo Shafi, Yeonseung Chung, Jihye Park, Aleksandar Shulevski, Henrik Christiansen, Yoosang Son, Elly Knight, Amanda Montoya, Jeongyoun Ahn, Christian Langkammer, Heera Moon, Changwon Yoon, Nikola Stikov, Mooseok Jang, Edward Choi, Junhan Kim, Yeon Sik Jung, Woo Youn Kim, Jae Kyoung Kim, Ishraq Md Anjum, Hyun Uk Kim, Drew Bridges, Carolin Lawrence, Xiang Yue, Alice Oh, Akari Asai, Sean Welleck, Graham NeubigRead
449
Cv
SetCon improves referring segmentation by having the model predict shared semantic concepts for whole target sets, then refining them into subgroups for stronger image and video grounding.
Zhixiong Zhang, Yizhuo Li, Shuangrui Ding, Yuhang Zang, Shengyuan Ding, Long Xing, Yibin Wang, Qiaosheng Zhang, Jiaqi WangRead
This work builds a large multimodal Earth observation foundation model that learns from hyperspectral, multispectral, radar, and thermal data together, aiming to improve both hyperspectral and general geospatial AI tasks.
Nassim Ait Ali Braham, Aaron Banze, Conrad M. Albrecht, Julien Mairal, Jocelyn Chanussot, Xiao Xiang ZhuRead
This paper turns video editing into a fast, few-step generation problem, using streaming diffusion-style models and clever attention mechanisms to make edits both quicker and more controllable.
Guanlong Jiao, Chenyangguang Zhang, Jia Jun Cheng Xian, Zewei Zhang, Renjie LiaoRead
450
Nlp
FAME uses an LLM only once offline to help build a lightweight on-premise expert system that pinpoints anomalous log messages with far less labeling effort.
Huanchi Wang, Zihang Huang, Yifang Tian, Kristina Dzeparoska, Hans-Arno Jacobsen, Alberto Leon-GarciaRead
This paper turns text-model interpretability into an auditability test: a feature should be both understandable to humans and genuinely separate from the prediction label.
Tong Wang, Yiqing Xu, Leo Yang YangRead
This paper turns tokenization from a greedy heuristic into a convex optimization problem, producing a tokenizer that can be near-optimal and often slightly better for language models.
Jan Tempus, Philip Whittington, Craig W. Schmidt, Dennis Komm, Tiago PimentelRead
Load more videos
451
Llm
This paper shows how to run large language models efficiently in the browser with WebGPU, cutting memory use and boosting speed across many devices and quantization formats.
Reese Levine, Rithik Sharma, Nikhil Jain, Abhijit Ramesh, Zheyuan Chen, Neha Abbas, James Contini, Tyler SorensenRead
Lumberjack makes private random forests much more practical by using a clever heavy-hitter pruning trick that keeps deeper trees useful without leaking sensitive data.
Christian Janos Lebeda, David Erb, Tudor Cebere, Aurélien BelletRead
VeriCache makes compressed LLM caching fast without changing the output, by drafting with a tiny cache and verifying against the full one only when needed.
Jiayi Yao, Samuel Shen, Kuntai Du, Shaoting Feng, Dongjoo Seo, Rui Zhang, Yuyang Huang, Yuhan Liu, Shan Lu, Junchen JiangRead
452
Llm
This paper argues that model collapse in LLMs is really a kind of cultural evolution problem, and shows that self-training can first improve then damage compositional language structure before eventually degrading it.
Dongxin Guo, Jikun Wu, Siu Ming YiuRead
This paper argues that many LLM simulation experiments are really observational studies in disguise, and it proposes causal tools to detect and reduce the hidden bias from shifting synthetic users.
Victoria Lin, Taedong Yun, Maja Matarić, John Canny, Arthur Gretton, Alexander D'AmourRead
This paper shows that a subtle mismatch in how uniform diffusion models are trained and sampled can be fixed with a leave-one-out view, leading to better language generation and a cleaner alternative formulation.
Samson Gourevitch, Yazid Janati, Dario Shariatian, Umut Simsekli, Eric Moulines, Eric P. Xing, Alain DurmusRead
453
Multimodal
Bernini uses a multimodal language model to plan video semantics and a diffusion model to turn that plan into realistic video, aiming to improve generation and editing with strong generalization.
Bernini Team, Chenchen Liu, Junyi Chen, Lei Li, Lu Chi, Mingzhen Sun, Zhuoying Li, Yi Fu, Ruoyu Guo, Yiheng Wu, Ge Bai, Zehuan YuanRead
This paper speeds up diffusion-based image and video generation by growing resolution gradually in the frequency domain, so models spend computation where it actually matters.
Howard Xiao, Brian Chao, Lior Yariv, Gordon WetzsteinRead
Stable Audio 3 is a fast, consumer-friendly audio generation and editing system that can create or modify music and sounds in seconds, even for variable-length clips.
Zach Evans, Julian D. Parker, Matthew Rice, CJ Carr, Zack Zukowski, Josiah Taylor, Jordi PonsRead
Load more videos
454
Optimizer
This paper shows that popular optimizer updates for language models are subtly biased on small batches, and it introduces a practical correction that can make AdamW, Sophia, and Shampoo train a bit better.
Nikhil Nayak, Julia White, Urchade Zaratiana, Kelton Zhang, Henrijs Princis, Dhruv Atreja, Henry Fawcett, Matthew Thomas, George Hurn-Maloney, Ash LewisRead
This paper uses heavy-tail theory to automatically give different Transformer layers different learning rates, helping large language models train faster and perform better.
Di He, Songjun Tu, Keyu Wang, Lu Yin, Shiwei LiuRead
This paper argues that SGD should be viewed less like Brownian motion and more like motion in a randomly changing landscape, revealing when training trajectories diffuse versus stay trapped.
Igor Ignashin, Anna Radovskaya, Andrew Semenov, Egor Lopatin, Stanislav Potapov, Aleksandr Kovalenko, Andrey Veprikov, Aleksandr Shestakov, Andrey Leonidov, Aleksandr BeznosikovRead
455
Generative Model
This paper turns generative modeling into a control problem and uses value functions to learn fast, straight transport paths for sampling data efficiently.
Pablo Moreno-Muñoz, Adrian Müller, Gergely NeuRead
This paper teaches a transformer to generate EEG as a continuous signal, using flow matching and signal-aware constraints to better preserve the brain-wave patterns that traditional denoising methods often miss.
Yifan Wang, Yijia Ma, Wen Li, Chenyu YouRead
This paper shows that generative models can look converged without truly learning the underlying factors, separating surface-level matching from deeper representation recovery.
Antoine Maillard, Sebastian GoldtRead
456
Self Supervised
This paper introduces a faster, augmentation-free way to learn strong time-series representations by contrasting meaningful chunks of each sequence instead of individual timesteps.
Abdul-Kazeem Shamba, Kerstin Bach, Gavin TaylorRead
This paper claims many common robustness tricks are really different ways of estimating the same nuisance-covariance object, and builds a geometric theory showing how to match regularization to that nuisance structure.
Vishal RajputRead
This paper turns massive wearable sensor data into a foundation model that can predict health outcomes, estimate daily metrics, and power a personalized health assistant.
Girish Narayanswamy, Maxwell A. Xu, A. Ali Heydari, Samy Abdel-Ghaffar, Marius Guerard, Kara Vaillancourt, Zhihan Zhang, Jake Garrison, Levi Albuquerque, Dimitris Spathis, Hong Yu, Hamid Palangi, Xuhai "Orson" Xu, David G. T. Barrett, Joseph Breda, Jed McGiffin, Yubin Kim, Yuwei Zhang, Naghmeh Rezaei, Samuel Solomon, Karan Ahuja, Tim Althoff, Jake Sunshine, Ming-Zher Poh, Benjamin Yetton, Ari Winbush, Nicholas B. Allen, James M. Rehg, Isaac Galatzer-Levy, Yun Liu, John Hernandez, Anupam Pathak, Conor Heneghan, Yuzhe Yang, Ahmed A. Metwally, Pushmeet Kohli, Mark Malhotra, Shwetak Patel, Xin Liu, Daniel McDuffRead
Load more videos
457
Transformer
This paper shows that transformers can only generate a surprisingly limited set of outputs, with the reachable sequence space growing linearly with prompt length and shrinking sharply beyond a threshold.
Maxime Meyer, Mario Michelessa, Caroline Chaux, Vincent Y. F. TanRead
This paper shows that, in a simplified transformer model, large learning rates can make training settle into cycles or chaos instead of cleanly learning, revealing new stability thresholds for transformer dynamics.
Krishnakumar BalasubramanianRead
03
Represented Is Not Computed: A Causal Test of Candidate Algorithmic Intermediates in a Transformer
This paper shows that a Transformer can represent the right intermediate arithmetic steps without actually using them in the way probes suggest, highlighting the gap between what models encode and what they compute.
Ishita Darade, Sushrut ThoratRead
458
Transformer
This paper makes looped transformers much cheaper by swapping in linear or sparse attention, showing you can keep quality while dramatically cutting compute.
Chunyuan Deng, Yizhe Zhang, Rui-Jie Zhu, Yuanyuan Xu, Jiarui Liu, T. S. Eugene Ng, Hanjie ChenRead
02
Neural Flow Operators can Approximate any Operator: Abstract Frameworks and Universal Approcimations
This paper builds a unified theory showing that neural flow models can approximate both functions and operators, connecting modern continuous-depth networks to residual and plain architectures.
Shuang Chen, Juncai He, Xue-Cheng TaiRead
This paper argues that GLU layers work better because they make the training dynamics better conditioned, helping models learn faster rather than necessarily generalizing better.
Xingyu Lyu, Qianqian Xu, Zhiyong Yang, Peisong Wen, Qingming HuangRead
459
Ai Research
This paper shows how an on-device AI assistant can store less, remember better, and answer in ways that match your preferences using a compact preference-aligned memory system.
Changmin Lee, Jaemin Kim, Taesik GongRead
This paper introduces a new dataset of real user chats paired with what people were actually thinking, aiming to help AI systems better infer hidden goals, preferences, and reactions.
Chuanyang Jin, Binze Li, Haopeng Xie, Cathy Mengying Fang, Tianjian Li, Shayne Longpre, Hongxiang Gu, Maximillian Chen, Tianmin ShuRead
TierCheck makes LLM training more resilient by keeping quick-to-recover checkpoints close to the GPUs while asynchronously backing up full states to durable storage.
Shujie Han, Feng Jiang, Patrick P. C. Lee, Xiao Zhang, Zhijie Huang, Nannan Zhao, Xiaonan Zhao, Lichen PanRead
Load more videos
460
Ai Research
This paper brings formal verification ideas to LLM safety, showing how to audit, monitor, and even intervene at runtime to keep AI systems within rules and constraints.
Parand A. Alamdari, Toryn Q. Klassen, Sheila A. McIlraithRead
The paper shows that LLM agents with memory can become less safe over time, even when each individual task looks fine, and introduces a way to measure that hidden risk.
Ahmad Al-Tawaha, Shangding Gu, Peizhi Niu, Ruoxi Jia, Ming JinRead
03
Your Neighbors Know: Leveraging Local Neighborhoods for Backdoor Detection in Decentralized Learning
Argus helps decentralized learners spot poisoned model updates by comparing suspicious triggers across neighbors, cutting backdoor attacks without needing a central server.
Sayan Biswas, Antoine Boutet, Davide Frey, Romaric Gaudel, Rachid Guerraoui, Maxime Jacovella, Anne-Marie Kermarrec, Dimitri Lerévérend, François Taïani, Martijn de VosRead
461
Ai Research
This paper shows how tiny wireless IoT devices can team up to run transformer models far too large for any one device by co-designing communication, partitioning, and loss-tolerant training.
Alexander Gräfe, Ding Huo, Johannes Berger, Marco Zimmerling, Sebastian TrimpeRead
OSCAR is a new way to shrink LLM KV caches to 2 bits by learning a better rotation from attention statistics, making long-context inference much faster and more memory-efficient without collapsing accuracy.
Zhongzhu Zhou, Donglin Zhuang, Jisen Li, Ziyan Chen, Shuaiwen Leon Song, Ben Athiwaratkun, Xiaoxia WuRead
This paper shows how to shrink slow tabular foundation models into fast CPU-friendly decision-tree ensembles, preserving most of the accuracy while cutting inference to milliseconds.
Aditya Tanna, Nassim Bouarour, Mohamed Bouadi, Vinay kumar Sankarapu, Pratinav SethRead
462
Ai Research
This paper shows that making a multimodal model imagine visual details first can sometimes help it understand images better, revealing a surprising generation-to-understanding loop.
Yujun Tong, Dongliang Chang, Zijin Yin, Xintong Liu, Yuanchen Fang, Zhanyu MaRead
This paper argues that vision-language-action models forget because one encoder is doing two jobs, and fixes it by adding a second brain-like pathway that preserves language understanding while improving robot control.
Jianke Zhang, Yuanfei Luo, Yucheng Hu, Xiaoyu Chen, Yanjiang Guo, Ziyang Liu, Hongbin Xu, Tian Lan, Jianyu ChenRead
This work turns a vision-language model into a single-pass depth predictor that can output dense geometry and text together, aiming to make VLMs more useful for 3D understanding.
Hanxun Yu, Xuan Qu, Yuxin Wang, Jianke Zhu, Lei keRead
Load more videos
463
Ai Research
This paper shows that RL systems trained on rubric-style rewards can learn to game the scoring rules without truly getting better, even when the verifiers are stronger.
Anas Mahmoud, MohammadHossein Rezaei, Zihao Wang, Anisha Gunjal, Bing Liu, Yunzhong HeRead
RubricEM trains research agents to think in rubric-defined stages, turning feedback from long, messy tasks into reusable experience for better long-form reasoning.
Gaotang Li, Bhavana Dalvi Mishra, Zifeng Wang, Jun Yan, Yanfei Chen, Chun-Liang Li, Long T. Le, Rujun Han, George Lee, Hanghang Tong, Chen-Yu Lee, Tomas PfisterRead
This paper shows that asking a medical AI to check its own answers can create a dangerous illusion of reliability, where the verifier often just agrees with the original mistake instead of catching it.
Ruinan Jin, Beidi Zhao, Myeongkyun Kang, Qiong Zhang, Xiaoxiao LiRead
464
Ai Research
This paper introduces a realistic benchmark that tests whether multiple AI agents can actually work together inside enterprise-style organizations with roles, permissions, and policy approvals.
Tao Yu, Hao Wang, Changyu Li, Shenghua Chai, Minghui Zhang, Zhongtian Luo, Yuxuan Zhou, Haopeng Jin, Zhaolu Kang, Jiabing Yang, YiFan Zhang, Xinming Wang, Hongzhu Yi, Zheqi He, Jing-Shu Zheng, Xi Yang, Yan Huang, Liang WangRead
The paper shows that you can train set-based neural posterior estimators on tiny sets of two and still make them work on huge observation sets, cutting compute while keeping accuracy strong.
Antoine Wehenkel, Michael Kagan, Lukas Heinrich, Chris PollardRead
This paper brings software-style unit testing to Text-to-CAD, creating a benchmark that checks whether generated CAD models actually satisfy the prompt and even helps improve generation methods.
Dimitrios Mallis, Marco Wang, Ahmet Serdar Karadeniz, Elisa Ricci, Anis Kacem, Djamila AouadaRead
465
Ai Research
This paper introduces Neo, an AI-assisted program analysis system that hunts for privilege-escalation bugs across large polyglot microservice codebases and has already found dozens of real-world vulnerabilities.
Penghui Li, Hong Yau Chong, Yinzhi Cao, Junfeng YangRead
This paper introduces a benchmark for testing whether multimodal AI can remember and use a user’s personal visual history, and proposes a memory-bank baseline that helps models answer personalized queries better.
Zihui Xue, Ami Baid, Sangho Kim, Mi Luo, Kristen GraumanRead
This paper proposes a way for AI vendors to trace a harmful autonomous agent back to the account that launched it, helping close a major accountability gap.
Ruben Chocron, Doron Jonathan Ben Chayim, Eyal Lenga, Gilad Gressel, Alina Oprea, Yisroel MirskyRead
Load more videos
466
Ai Research
PhyGround is a new benchmark and judging system for testing whether AI-generated videos really obey physical laws, with detailed human annotations and an open physics-aware evaluator.
Juyi Lin, Arash Akbari, Yumei He, Lin Zhao, Haichao Zhang, Arman Akbari, Xingchen Xu, Zoe Y. Lu, Enfu Nan, Hokin Deng, Edmund Yeh, Sarah Ostadabbas, Yun Fu, Jennifer Dy, Pu Zhao, Yanzhi WangRead
RoadmapBench is a new benchmark showing that today’s top coding agents still struggle badly with realistic, large-scale software version upgrades spanning many files and months of work.
Xinbo Xu, Ruihan Yang, Haiyang Shen, Wendong Xu, Bofei Gao, Ruoyu Wu, Kean Shi, Weichu Xie, Xuanzhong Chen, Ming Wu, Jason Zeng, Michael Heinrich, Elvis Zhang, Liang Chen, Kuan Li, Baobao ChangRead
03
WorldReasonBench: Human-Aligned Stress Testing of Video Generators as Future World-State Predictors
This paper introduces a benchmark to test whether video generators can actually reason about how the world should change over time, not just produce realistic-looking footage.
Keming Wu, Yijing Cui, Wenhan Xue, Qijie Wang, Xuan Luo, Zhiyuan Feng, Zuhao Yang, Sudong Wang, Sicong Jiang, Haowei Zhu, Zihan Wang, Ping Nie, Wenhu Chen, Bin WangRead