Proposes a unified architecture-lifecycle framework to enhance CUA reliability in real environments, analyzing perception, decision, execution layers and stages of creation, deployment, operation, maintenance.

Zejian Chen, Zhanyuan Liu, Chaozhuo Li et al.

Proposed Residual Latent Action (RLA) method using DINO residuals to improve world model prediction accuracy and reduce computational costs in multi-task robotics.

Xinyu Zhang, Zhengtong Xu, Yutian Tao et al.

ModelLens learns a performance-aware latent space from leaderboard interactions to recommend models without running them, outperforming metadata-based baselines on 162K records.

Rui Cai, Weijie Jacky Mo, Xiaofei Wen et al.

Behavior Cue Reasoning improves efficiency and safety in LLMs, doubling success rate from 46% to 96%.

Christopher Z. Cui, Taylor W. Killian, Prithviraj Ammanabrolu

STABLE uses RLPR to achieve rollback-free stable brick generation, improving inference speed by 94%.

Chenhui Xu, Ziyue Bai, Fuxun Yu et al.

Introduces PostEDA-Bench with 145 tasks to evaluate LLMs in DRC fixing and PPA optimization.

Pengju Liu, Nuo Xu, Jinwei Tang et al.

SAFE challenge employs deep learning with pre-trained vision backbones and autoencoding to detect synthetic videos, achieving an average AUC of 0.82 across 13 models.

Kirill Trapeznikov, Gabriel Mancino-Ball, Jonathan Li et al.

MIND uses sliced Wasserstein for efficient, robust evaluation of generative models, requiring fewer samples and being faster than FID.

Quentin Berthet, Yu-Han Wu, Clement Crepy et al.

ROSE uses cooperative elasticity on serving GPUs for Agentic RL, boosting throughput by 1.3-3.3x.

Wei Gao, Yuheng Zhao, Dilxat Muhtar et al.

Proposes 'Binding Constraint Thesis', showing scheduler configuration impacts long-horizon LLM performance more than model upgrades, advocating for disclosure.

Yunbei Zhang, Janet Wang, Yingqiang Ge et al.

HumanNet dataset enables Qwen model to substitute human video for robot data, enhancing learning efficiency.

Yufan Deng, Daquan Zhou

AOPD replaces negative reinforcement with local divergence minimization, reducing variance and exploration issues in on-policy distillation, boosting math reasoning by 4.09/8.34 points.

Nan Jia, Haojin Yang, Xing Ma et al.

Introduces bridge interface theory, separating prediction, compression, and empowerment; proves perfect prediction requires identifying latent quotients or control capacity.

Richard Csaky

InciteResearch framework transforms tacit understanding into explicit research proposals, enhancing novelty and impact.

Jie Yu, Song Qiu

Proposes a robust linear semantic segmentation model for low-resource spoken dialects, outperforming baselines with over 1000 multi-genre samples.

Kirill Chirkunov, Younes Samih, Abed Alhakim Freihat et al.

Introduces the Adaptive Coordinate Transform (ACT) module, significantly improving neural operators' ability to track evolving structures in PDE modeling.

Chaoyu Liu, Zhonghao Li, Gaohang Chen et al.

Prologue introduces prologue tokens to decouple generation and reconstruction, reducing gFID from 21.01 to 10.75 on ImageNet 256×256.

Bowen Zheng, Weijian Luo, Guang Yang et al.

Introducing TEBench, a project-level test evolution benchmark with 314 instances, achieving 45.7%-49.4% F1 in test identification, highlighting current model limitations.

Ye Shang, Quanjun Zhang, Haichuan Hu et al.

SPEED method reduces long-context inference cost via shallow prefill and deep decode, enhancing efficiency.

Jungsuk Oh, Hyeseo Jeon, Hyunjune Ji et al.

VibeServe employs multi-agent loops to automatically generate bespoke LLM serving systems, matching or exceeding hand-tuned performance.

Keisuke Kamahori, Shihang Li, Simon Peter et al.