Explore · 每周精选
发现最受关注的研究论文,追踪研究趋势,订阅感兴趣的期刊与关键词。
Qi Li, Xinchao Wang
Jeva is a decision-native architecture that jointly enforces null-awareness, permutation invariance, option scalability, and cross-request caching for structured decision-making.
50 篇论文 · 按点赞排序
Luozheng Qin, Yuhan Zhu, Changhai Zhou +9 authors
MJev-Spatial is a fast multimodal decision model that unifies spatial reasoning tasks into a single-pass classification framework, improving speed and accuracy over autoregressive vision-language models.
Xuyang Liu, Shihao Wang, Xinyu Wei +5 authors
OmniCompass is a holistic benchmark for omni-modal video grounding and understanding that evaluates vision-language and audio-visual reasoning through temporal grounding, spatial grounding, and question answering tasks.
Weijie Wang, Haomin Zhang, Ye Li +7 authors
ATAS improves long-form video consistency by compiling screenplays into referenced shot requests, retrieving spatial and temporal evidence from scene and event memories, and conditionally updating memory after generation.
Mingyu Ouyang, Yiran Wang, Siyuan Hu +9 authors
Player-1 improves game control by combining supervised fine-tuning on gameplay trajectories with reinforcement learning across synthetic environments, achieving substantial gains over demonstration-only baselines.
Yuhan Zhu, Xiangyu Zeng, Qingyi Si +15 authors
VideoJev converts video observations into structured decisions by scoring temporal intervals and text options, enabling efficient reusable inference for visual agents.
Zhenglin Wan, Jiayu Chen, Zhengfeiyang Zhang +6 authors
DiffusionART improves reinforcement learning fine-tuning of diffusion models by removing hidden consistency forces and enabling stable, efficient training without EMA tuning or differentiable rewards.
Zhiyuan Shi, Xiang Li, Xingzhuo Guo +8 authors
Multimodal language models use an intrinsic linear subspace called S-Space to encode object coordinates, supporting spatial perception, reasoning, and response generation, while also revealing bottlenecks in viewpoint transformation.
Dianqiao Lei, Kevin Qinghong Lin, Pan Lu +2 authors
Researchers propose multi-level metrics to evaluate LLM no-thinking behavior, finding that explicit reasoning persists despite controls and varies with task openness, establishing no-thinking as a distinct capability requiring systematic assessment.
Bingxi Hou, Guochao Jiang, Guofeng Quan +4 authors
Waypoint-OPD improves policy-shift distillation by gating endpoint shifts using directional agreement between early and late teacher policy changes, boosting reasoning performance with minimal overhead.
Guanting Dong, Xinyu Yang, Jiajie Jin +9 authors
Agent-Reflex trains interactive agents to reflect and recover from noisy tool feedback and user corrections through skill discovery, supervised fine-tuning, and on-policy optimization.
Taoyang Jia, Weikai Huang, Linxin Song +6 authors
Reflex is a three-stage framework that enables humanoid robots to catch thrown boxes using egocentric vision by decomposing whole-body control, dynamics inference, and visual perception into sequentially trained stages.
Panrong Tong, Quyu Kong, Pengxiang Li +12 authors
MAI-Gym is an open platform that unifies GUI agent evaluation, data collection, and online reinforcement learning training across mobile, computer, and web environments, enabling scalable synthesis of verified tasks and training of a unified 9B agent with state-of-the-art open performance.
Qize Yu, Lianrui Fan, Boyu Chen +23 authors
GroundingPI is a 4B grounding foundation model that improves precise object localization and downstream robotic and autonomous driving performance through multimodal pretraining and reinforcement learning.
Shun Zou, Ziyu Ma, Yi Zou +6 authors
TerminalHorizon automates long-horizon terminal agent data synthesis by reconstructing environments around omitted work and progressively expanding tasks, yielding substantial gains across coding, production, and reasoning benchmarks.
Xin Gao, Yifan Yang, Xin Li +13 authors
AesCode improves code generation for visual artifacts by combining design graphs with cross-modal rewards guided by reference images, enhancing layout, aesthetics, and verifiable correctness.
Yu Wang, Rui Dai, Wei Chen +5 authors
Urbex reframes city-scale 3D grounding as an active search process using a vision-language agent with reinforcement learning, improving accuracy and speed without precomputed proposals.
Yongzhen Hu, Hao Yu, Jiawei Wang +7 authors
InfiniSplat v2 improves feed-forward 3D Gaussian reconstruction by decoupling geometry and attribute prediction, aligning Gaussians with ground-truth poses, and using large-angle supervision to enhance novel-view rendering quality.
Ruiyao Xu, Yanfei Chen, Zhongying CuiZhu +6 authors
Harness² improves agent execution harnesses through recursive task-level refinement and continual harness-level learning, boosting performance across benchmarks without ground-truth verifiers.
Hanzhang Zhou, Long Li, Quyu Kong +12 authors
OpenGCUA introduces an open generalist computer-use agent that integrates visual and programmatic interfaces, harness-to-model decision transfer, and code execution to improve task performance and efficiency across models.
Aryan Das, Surjo Dey, Koushik Biswas +4 authors
SALD is a self-referenced training framework for diffusion models that uses dual noise-level evaluation and adaptive weighting to improve generation quality without external teachers or extra parameters.
Bingxuan Li, Yiwen Song, Xueqing Wu +10 authors
VIS-Ground transforms heterogeneous inputs into executable constraints for interactive video generation by recovering cross-context dependencies and enforcing them through planning, verification, and rendering.
Gennady Sidorov, Valerii Kornilov, Denis Grankin +2 authors
GreenWAM is a unified world foundation model for robotics that uses a shared anatomical action space and joint training to serve as both a predictive world model and a control policy across diverse robot bodies.
Tian Jin, Wenhao Wang, Jing Kang +4 authors
SoftwareGym automates synthesis of diverse, executable GUI applications to provide scalable training data that improves computer-use agent performance on out-of-domain benchmarks.
Chuyang Wei, Haoxiang Guan, Minghang Zhu +5 authors
A framework for self-evolving agent harnesses uses feedback-driven revisions and evolutionary optimization with theoretical generalization guarantees while keeping model weights fixed.
Hyunwook Choi, Seungryong Kim, Dongyoon Han +1 authors
Speculative Forcing accelerates few-step autoregressive video generation by accepting early token drafts and skipping unnecessary denoising steps while preserving output quality.
Shi-Yu Tian, Yong Liu, Ximan Liu +10 authors
Road MOPD reduces gradient conflicts in multi-teacher distillation by using cross-teacher agreement for token-level credit assignment and auxiliary gradients to constrain conflicting updates.
Futing Wang, Chenxi Li, Jianhao Yan +6 authors
Tracking-Value PPO improves reinforcement learning for reasoning by modeling value evolution through cumulative log-odds updates, yielding better accuracy and generalization.
Jiahao Wang, Hongjin Qian, Yuyang Hu +2 authors
AgentSWE evaluates whether coding agents build software that truly meets user requirements across creation, editing, and optimization stages, revealing widespread failures to deliver faithful behavior.
Hang He, Linting Wang, Guanming Liu +14 authors
WorkRover enables iterative refinement of interactive training environments for long-horizon agents via recursive synthesis, compatibility replay, and turn-level credit assignment.
北京市昌平区探索星信息技术及软件开发工作室
京ICP备2026059466号