MSVBench introduces hierarchical scripts and reference images, combining LMM and expert models for human-level multi-shot video evaluation.

Haoyuan Shi, Yunxin Li, Nanhao Deng et al.

SPLARE introduces a sparse latent retrieval model using pretrained SAEs, outperforming vocabulary-based methods in multilingual and out-of-domain tasks.

Thibault Formal, Maxime Louis, Hervé Dejean et al.

OmniXtreme uses flow matching and physics-aware residual fine-tuning to overcome the scalability and fidelity limits in high-dynamic humanoid control.

Yunshen Wang, Shaohang Zhu, Peiyuan Zhi et al.

Fourier Angle Alignment (FAA) improves rotated object detection accuracy in remote sensing, achieving 78.72% mAP on DOTA-v1.0 with frequency-based orientation estimation.

Changyu Gu, Linwei Chen, Lin Gu et al.

FuXi-Linear model leverages linear attention for efficient long-sequence recommendation, achieving up to 21x decoding speedup.

Yufei Ye, Wei Guo, Hao Wang et al.

Introduces LFQA-HP-1M dataset with nine scoring metrics, achieving model performance comparable to human preferences in long-form QA evaluation.

Rafid Ishrak Jahan, Fahmid Shahriar Iqbal, Sagnik Ray Choudhury

Evidential NeRF introduces a probabilistic framework for joint aleatoric and epistemic uncertainty estimation from a single forward pass, enhancing scene reconstruction fidelity and reliability.

Ruxiao Duan, Alex Wong

SOTAlign leverages minimal image-text pairs and optimal transport to achieve semi-supervised vision-language alignment, outperforming existing methods significantly.

Simon Roschmann, Paul Krzakala, Sonia Mazelet et al.

EvoX employs meta-evolution of search strategies, outperforming AlphaEvolve on nearly 200 tasks with significant efficiency gains.

Shu Liu, Shubham Agarwal, Monishwaran Maheswaran et al.

Proposes RaWMPC, a risk-aware predictive control framework that learns safety without expert supervision, improving autonomous driving generalization.

Jiangxin Sun, Feng Xue, Teng Long et al.

SPARR combines simulation-trained and real-world residual policies, improving assembly success by 38.4% and reducing cycle time by 29.7%.

Yijie Guo, Iretiayo Akinola, Lars Johannsmeier et al.

Applying Takeuchi Information Criterion (TIC) to evaluate DNN generalization near NTK regime; experiments show high correlation with actual gaps.

Hiroki Naganuma, Taiji Suzuki, Rio Yokota et al.

PATRA employs pattern-aware alignment and balanced reinforcement learning to enhance time series question answering, achieving significant performance gains.

Junkai Lu, Peng Chen, Xingjian Wu et al.

SkillNet builds a unified skill ontology with multi-source creation and multi-dimensional evaluation, enhancing scalable AI skill management.

Yuan Liang, Ruobin Zhong, Haoming Xu et al.

ToProVAR achieves 3.4x acceleration via tri-dimensional entropy-aware semantic analysis and sparsity optimization.

Jiayu Chen, Ruoyu Lin, Zihao Zheng et al.

Operation-R1 trains lightweight LLMs via verifiable reinforcement learning to generate entire data preprocessing pipelines in one step, boosting table QA accuracy by 8.83%.

Fengyu Li, Junhao Zhu, Kaishi Song et al.

Enhanced persuasive agents using cross-disciplinary strategies, achieving higher success rates, especially for low-intent users.

Shinnosuke Nozue, Yuto Nakano, Yotaro Watanabe et al.

Personalized LLM agents integrate user signals for cross-component interaction, enhancing long-term adaptability.

Yue Xu, Qian Chen, Zizhan Ma et al.

LegoOcc uses 3D language-embedded Gaussians for monocular open-vocabulary indoor occupancy, achieving 59.50 IoU and 21.05 mIoU, surpassing previous methods.

Changqing Zhou, Yueru Luo, Han Zhang et al.

STATIC vectorizes Trie decoding on TPU/GPU, reaching 0.033 ms per step with strict constraints.

Zhengyang Su, Isay Katsman, Yueqi Wang et al.