Daily Paper Cast

🤗 Upvotes: 130 | cs. CV, cs. CL, cs. MM Authors: Xuyang Liu, Shihao Wang, Xinyu Wei, Minghan Li, Zeqing Wang, Zuhao Yang, Zichen Wen, Lei Zhang Title: OmniCompass: A Holistic Benchmark for Omni-Modal Video Grounding and Understanding Arxiv: http://arxiv.org/abs/2610.18829v1 Abstract: With the rapid development of multimodal large language models (MLLMs), evaluating their video understanding capab...

🤗 Upvotes: 108 | cs. LG, cs. CL Authors: Bingxi Hou, Guochao Jiang, Guofeng Quan, Weiqing Li, Wenfeng Feng, Guohua Liu, Yuewei Zhang Title: Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy Distillation Arxiv: http://arxiv.org/abs/2610.21601v1 Abstract: Policy-shift distillation reuses the difference between a teacher's log probabilities before and after reinforcement learni...

🤗 Upvotes: 68 | cs. AI, cs. CL Authors: Chuyang Wei, Haoxiang Guan, Minghang Zhu, Kexin Zhou, Jiyan He, Xiren Zhou, Huanhuan Chen, Yifei Shen Title: Self-Evolving Agent Harness Arxiv: http://arxiv.org/abs/2610.18813v1 Abstract: Harness self-evolution revises an agent's software scaffolding, including prompts, memory, workflows, and executable components, while keeping model weights fixed. We form...

🤗 Upvotes: 58 | cs. CV Authors: Hyunwook Choi, Seungryong Kim, Dongyoon Han, Sangdoo Yun Title: Speculative Forcing for Efficient Autoregressive Video Generation Arxiv: http://arxiv.org/abs/2610.22347v1 Abstract: Few-step autoregressive video generators enable streaming generation, yet they run every token through the final denoising step, regardless of how much refinement it needs. We find this...

🤗 Upvotes: 45 | cs. SD Authors: Bingzhe Zhang, Kai Li, Xiaolin Hu Title: Learning Where to Split: Dynamic Band Splitting for Audio Source Separation Arxiv: http://arxiv.org/abs/2610.14993v1 Abstract: Audio source separation models must preserve fine spectral structure while keeping the cost of the separator tractable. Band-split architectures reduce this cost by manually partitioning the spectrum...

🤗 Upvotes: 44 | cs. CV Authors: Yufei Jiao, Jinfa Huang, Zeyue Xue, Tianyu Xie, Yuexiao Ma, Rongrong Ji, Xiawu Zheng, Jiebo Luo Title: SCOPE: Semantic Bounded Credit Optimization for Prompt Engineering in Text-to-image Generation Arxiv: http://arxiv.org/abs/2610.22264v1 Abstract: Reinforcement learning (RL) enables language models to improve text-to-image (T2I) generation by rewriting prompts for...

🤗 Upvotes: 36 | cs. CV, cs. AI Authors: Yipeng Gao, Ziyao Zeng, Jinfa Huang, Yijiang Li, Haolin Xiong, Wang Qin, Xinlei Yu, Youheng Yao, Jie Zhang, Kun-Yu Lin, Rong Liu, Yinyuan Zhao, Yajie Zhao, Meiqi Guo, Jiebo Luo, Lei Shu, Yunhao Ge, Xinyu Hu, Laurent Itti Title: 3DCodeVerse: Exploring 3D Worlds as Code Arxiv: http://arxiv.org/abs/2610.22341v1 Abstract: We introduce 3DCodeVerse, a research ec...

🤗 Upvotes: 28 | cs. CV Authors: Dongyue Lu, Guian Fang, Hur Lim, Yonggang Wu, Yanrong Wang, Hongshan Chen, Shengju Qian, Yue Liao, Wei Chow, Lingdong Kong, Tianxin Huang, Yu Yang, Hongji Yang, Junchao Huang, Ao Liang, Yihao Wang, Zihan Wang, Rong Li, Hanlin Chen, Xin Wang, Wei Tsang Ooi, Mike Zheng Shou, Shuicheng Yan Title: Orbia: Do Generated Videos Form a Consistent 3D World? Arxiv: http://arx...

🤗 Upvotes: 26 | cs. CL, cs. AI, cs. LG Authors: Yilong Xu, Minrui Xu, Xinyu Zhou, Xiang Long, Haochen Tan, Zhijiang Guo Title: Mosaic: Multi-Session Agentic Reinforcement Learning Arxiv: http://arxiv.org/abs/2610.22174v1 Abstract: Most existing agentic reinforcement learning methods assume single-session trajectories, where interaction proceeds within one continuously accumulated context. However...

🤗 Upvotes: 25 | cs. CV, cs. AI, cs. HC Authors: Boyuan Sun, Deshui Miao, Wenzhao Gao, Shaoyong Jia, Shaohui Jiao, Chi Liu, Yan Shu, Derek Li, Bryan Dai, Shengsheng Qian, Qibin Hou Title: DeltaStreamer: Towards Human-Like Agentic Streaming Video Understanding Arxiv: http://arxiv.org/abs/2610.19849v1 Abstract: We propose DeltaStreamer, a novel agentic framework for streaming video understanding ins...

🤗 Upvotes: 166 | cs. CV Authors: Weijie Wang, Haomin Zhang, Ye Li, Akide Liu, Zeyu Zhang, Haoyu Zhao, Yefei He, Donny Y. Chen, Feng Chen, Bohan Zhuang Title: Any Time, Any Space: A World-State-Centric Framework for Agentic Video Generation Arxiv: http://arxiv.org/abs/2610.20571v1 Abstract: Long-form agentic video generation must turn a screenplay into shots that remain spatially and temporally co...

🤗 Upvotes: 117 | cs. CV, cs. AI, cs. HC Authors: Mingyu Ouyang, Yiran Wang, Siyuan Hu, Xingyilang Yin, Huiqiang Sun, Guangzhi Wang, Junfu Pu, Cheng Cheng, Honghao Chen, Ying Shan, Mike Zheng Shou, Hwee Tou Ng Title: Player-1: Learning to Play from Synthetic Experience Arxiv: http://arxiv.org/abs/2610.20584v1 Abstract: Computer-use agents have demonstrated strong capabilities in interpreting scree...

🤗 Upvotes: 50 | cs. LG Authors: Zefeng He, Yunkai Chai, Xiaoyang Chen, Tong Zhu, Zongliang Wu, Yun Luo, Yafu Li, Yu Cheng Title: LatentLoop: Scaling Looped MoE Models with Asymmetric Compute Allocation Arxiv: http://arxiv.org/abs/2610.19975v1 Abstract: Looped Mixture-of-Experts (MoE) models combine parameter reuse with sparse computation to improve performance under matched budgets. Prior work de...

🤗 Upvotes: 47 | cs. LG, cs. AI Authors: Kaixiang Zhao, Tianrun Yu, Shangzhe Li, Yuxiao Yang, Amanda Hughes, Weitong Zhang, Taylor W. Killian Title: Recovering Reliable Supervision via Critique-Guided On-Policy Distillation Arxiv: http://arxiv.org/abs/2610.20572v1 Abstract: On-policy distillation (OPD) trains a student model on its own rollouts with token-level supervision from a teacher, assuming...

🤗 Upvotes: 47 | cs. CL, cs. AI, cs. CV, cs. LG, cs. MA Authors: Jintian Zhang, Feng He, Xinnian Liang, Ziwen Xu, Haoming Xu, Haojing Huang, Shuangzhi Wu, Ningyu Zhang Title: Cortex: Consolidating Context into Parametric Memory Arxiv: http://arxiv.org/abs/2610.20399v1 Abstract: Context can improve large language model performance, but these gains may be lost after context removal. Inspired by biol...

🤗 Upvotes: 44 | cs. AI Authors: Huining Yuan, Xiangmin Yi, Zelai Xu, Zixu Wang, Xinlei Chen, Xiao-Ping Zhang, Yu Wang, Chao Yu Title: ActionOPD: Distilling Decisions, Not Tokens Arxiv: http://arxiv.org/abs/2610.19820v1 Abstract: Distilling capable language-model agents into smaller models requires aligning the student's behavior with the teacher's. This raises a fundamental question: at which sta...

🤗 Upvotes: 39 | cs. SE, cs. AI Authors: Yuyao Ge, Shenghua Liu, Peizhong Gao, Chendi Ge, Chenxu Liu, Yiwei Wang, Lingrui Mei, Baolong Bi, Jiayu Yao, Jiafeng Guo, Xueqi Cheng Title: Doc2Harness: Benchmarking LLMs on Long-Horizon Harness Engineering of Coding Agents Arxiv: http://arxiv.org/abs/2610.20344v1 Abstract: Agents driven by large language models are increasingly deployed on long-horizon ta...

🤗 Upvotes: 31 | cs. AI, cs. CL, cs. RO Authors: Minghao Guo, Qingyue Jiao, Xiujin Liu, Jianhong Pang, Ruimeng Yang, Naihao Xu, Liwei Che, Haochen Xue, Ziqi Wang, Xinying Cai, Jiaxuan Luo, Jiaxin Deng, Xuesong Ye, Jintao Qu, Wenyao Gao, Jingwen Gu, Chenyang Zhao, Shichun Liu, Guibin Zhang, Zefang Liu, Siyin Wang, Wujiang Xu, Danrui Li, Zeru Shi, Woongyeong Yeo, Shuang Liu, Wei-Chieh Huang, Weizhi...

🤗 Upvotes: 30 | cs. CV Authors: Yunuo Chen, Xiuyu Wu, Qinhong Yang, Jinjing Zhao, Yanchen Dong, Sirui Zhang, Dong Chen, Fangyun Wei, Lei Shi, Ji Li, Dongdong Chen, Guo Lu, Yan Lu, Baining Guo Title: LensErase: A Foundation Model for Erase Anything Arxiv: http://arxiv.org/abs/2610.18072v1 Abstract: We introduce LensErase, a general-purpose mask-based Erase Anything foundation model designed to cov...

🤗 Upvotes: 25 | cs. LG, cs. AI Authors: Sihwan Park, Sung-Yub Kim, Doohyuk Jang, Eunho Yang Title: KV4-RL: Unlocking Faster Asynchronous Rollouts for RLVR with INT4 KV Cache Quantization Arxiv: http://arxiv.org/abs/2610.14448v1 Abstract: Asynchronous rollout improves the efficiency of reinforcement learning with verifiable rewards (RLVR) by reducing under-utilization during long-tail generation....

🤗 Upvotes: 58 | cs. CV, cs. AI Authors: Yu Wang, Rui Dai, Wei Chen, Yi Wang, Bo Dang, Kaikui Liu, Xiangxiang Chu, Yansheng Li Title: Urbex: Agentic Spatial Grounding for City-Scale 3D Scenes Arxiv: http://arxiv.org/abs/2610.18356v1 Abstract: City-scale 3D grounding is commonly formulated as candidate ranking, which requires precomputed instance proposals and becomes inefficient in large outdoor s...

🤗 Upvotes: 55 | cs. CL, cs. AI Authors: Yang Xiao, Yusong Sun, Haoyi Wu, Wenyang Hui, Weiqiang Hang, Luqing Wang, Zejian Wang, Zirui Tu, Jingsheng Mao, Shaobo Hu, Yang Luo, Jiale Guo, Jin Zhang, Yudan Liu, Zhaokai Luo, Ren Huan, Yao Hu, Chengyue Jiang, Wenjie Li, Chuan Mu Title: Vela: User Behavior as Supervision Sparks LLM Intelligence Arxiv: http://arxiv.org/abs/2610.14709v1 Abstract: Large lan...

🤗 Upvotes: 54 | cs. LG Authors: Fanxu Meng, Taosong Fang Title: DoSA: Dots Sparse Attention Arxiv: http://arxiv.org/abs/2610.13990v1 Abstract: Sparse attention visits few keys, yet finding them can still require scoring every key with every indexer head. We introduce DoSA (Dots Sparse Attention), a lossless method for making this computation conditional while preserving exact token selection. Its...

🤗 Upvotes: 48 | cs. RO, cs. AI Authors: Minghao Fu, Tavis Siebert, Eryk Halicki, Randall Balestriero Title: LeWAM: End-to-End World and Action Modeling with JEPAs Arxiv: http://arxiv.org/abs/2610.12818v1 Abstract: Joint-embedding predictive architectures (JEPAs) learn compact latent world models without pixel reconstruction. Existing JEPA pipelines typically train visual representations and actio...

🤗 Upvotes: 40 | cs. AI Authors: Qi Li, Xinchao Wang Title: Jeva: System One Models from First Principles Arxiv: http://arxiv.org/abs/2610.18773v1 Abstract: System one models enable direct, low-latency decision-making. Yet recent practice of predicting an answer from serialized text does not, by itself, respect the structure of a decision problem. We argue that current designs underemphasize four...