Proposes an LLM-based agent framework for automatic SCIP plugin generation, improving MIP solving by identifying global constraints and propagators.

Liding Xu, Yugeng Zhou, Sebastian Pokutta

Introducing sparse MoE layers in looped transformers enhances scalability and early-exit efficiency, outperforming dense models at fixed compute levels.

Ryan Lee, Jacob Biloki, Edward J. Hu et al.

FORTIS benchmark quantifies over-privilege in large models' skill selection and execution, with failure rates exceeding 62.5%, revealing systemic permission control issues.

Shawn Li, Chenxiao Yu, Han Wang et al.

Proposes 'Imagining in 360°' framework with probabilistic spatial priors, boosting humanoid visual search success by over 30%.

Jingdong Zhang, Yizhou Wang, Zhengzhong Tu et al.

IMPACT uses augmented Lagrangian with active-set strategy to solve MPCCs, achieving up to 70× speedup in contact-implicit trajectory optimization.

Jiayun Li, Dejian Gong, Georgia Chalvatzaki

Proposes a non-parametric rehearsal learning framework using conditional mean embeddings and kernel ridge regression to optimize AUF probability without model assumptions.

Wen-Bo Du, Tian-Zuo Wang, Han-Jia Ye et al.

Budgeted Subset Refinement improves execution-aware LLM research ideation, with MMR-k showing best results.

Micah Zhang

Introduces anisotropic SPDE-based score models, enhancing geometric structure preservation over longer timescales in image generation.

Sascha Holl, Jente Vandersanden, Gurprit Singh et al.

Transforming Alpamayo 1 into a single-reasoning architecture reduces inference latency by 69.23%, maintaining trajectory diversity.

Yunseong Jeon, Namcheol Lee, Yoonsu Lee et al.

DocScope benchmarks trustworthy reasoning in long-document understanding, finding only 29% complete evidence chains.

Xiang Feng, Jiawei Zhou, Zhangfeng Huang et al.

Proposes a structural pruning method for SO(3)-equivariant atomistic models, reducing parameters by 1.5-4×, with improved accuracy and efficiency.

Chen Wang, Siyu Hu, Guangming Tan et al.

HyRo uses hyperbolic rotation to decouple hierarchical and semantic alignment, achieving state-of-the-art open-vocabulary semantic segmentation.

Hoang M. Truong, Hai Nguyen-Truong, Dang Huynh

Proposes CoDistill-GRPO, jointly training large and small models, boosting small model accuracy by 11.6%, with 18% speedup.

Soo Min Kwon, Ziteng Sun, Ananda Theertha Suresh et al.

Introduced Compressed Video Aggregator (CVA), improving recommendation performance on MicroLens with significant training time reduction.

Yang Xiao, Huiyuan Chen, Kaiyuan Deng et al.

ColGraphRAG enhances image retrieval precision in multimodal QA using MaxSim multi-vector scoring.

Seonok Kim

EvoMAS framework enhances complex task success by constructing execution-time multi-agent workflows.

Chengdong Xu, Kaiqiang Ke, Ziheng Liu et al.

LAQuant uses layer-wise lookahead loss for overhead-free quantization, improving AIME25 Pass@1 by 15.11 percentage points.

Euntae Choi, Sumin Song, Sungjoo Yoo

This work systematically analyzes large-scale MoE model pruning and distillation, showing pruning outperforms training from scratch, with combined KD and progressive schedules yielding best results.

Shengkun Tang, Zekun Wang, Bo Zheng et al.

SoftGAC introduces path regularization with short Gaussian bridges, enabling efficient single-pass maximum entropy RL with improved performance.

Ke He, Le He, Shunpu Tang et al.

Sycamore framework compares ungrounded, grounded synthetic personas and real experts in genomics visualization evaluation, revealing biases and limitations.

Huyen N. Nguyen, Astrid van den Brandt, Nils Gehlenborg