PosterForest employs hierarchical multi-agent collaboration for scientific poster generation, outperforming prior methods without additional training.

Jiho Choi, Seojeong Park, Seongjong Song et al.

SVS-SNG analyzes single-mini-batch natural gradient through sketch-and-project geometry, yielding convergence rates governed by α/γ.

Gil Goldshlager, Jiang Hu, Lin Lin

OneRec-V2 introduces a lazy decoder architecture, reducing 94% computation, scaling to 8B parameters, and improving real-world user engagement.

Guorui Zhou, Hengrui Hu, Hongtao Cheng et al.

OLMoASR achieves comparable speech recognition performance to Whisper using the OLMoASR-Mix dataset.

Huong Ngo, Matt Deitke, Martijn Bartelds et al.

UTRL enhances Qwen3-4B's unit test generation via adversarial reinforcement learning, outperforming GPT-4.1.

Dongjun Lee, Changho Hwang, Kimin Lee

ConspirED dataset reveals cognitive traits of conspiracy theories, assesses LLM safety.

Luke Bates, Max Glockner, Preslav Nakov et al.

This work uncovers security vulnerabilities in approximate caches of text-to-image diffusion models, enabling remote covert channels, prompt stealing, and poisoning attacks.

Desen Sun, Shuncheng Jie, Sihang Liu

MPFormer employs a target-conditioned Transformer with dynamic quota adjustment for multi-objective personalized retrieval, boosting coverage and efficiency.

Yijia Sun, Shanshan Huang, Linxiao Che et al.

Proposes Entropy Area Score (EAS), a sequence-level entropy integral for quantifying LLM uncertainty without external models, improving data selection.

Yongfu Zhu, Lin Sun, Guangxiang Zhao et al.

DeepScholar-bench is a real-time benchmark for generative research synthesis, evaluating knowledge integration, retrieval relevance, and citation verifiability.

Liana Patel, Negar Arabzadeh, Harshit Gupta et al.

Introduces T2R-bench, a benchmark for industrial table-to-report tasks, with models scoring only 62.71%, highlighting room for improvement.

Jie Zhang, Changzai Pan, Kaiwen Wei et al.

Instructional Agents reduce faculty workload by automating course material creation with multi-agent design.

Huaiyuan Yao, Wanpeng Xu, Justin Turnau et al.

VoxHammer is a training-free 3D editing method using inverse trajectory prediction and feature replacement.

Lin Li, Zehuan Huang, Haoran Feng et al.

StepWiser enhances reasoning accuracy via generative judges, showing improved intermediate step accuracy in experiments.

Wei Xiong, Wenting Zhao, Weizhe Yuan et al.

Proposes LSTM-based central path projection to accelerate OPF solving, reducing iterations by 85.5% and solution time by 94%.

Farshad Amani, Amin Kargarian, Ramachandran Vaidyanathan

Proposes a novel framework combining probabilistic crowd modeling and active learning to enhance preference-based reinforcement learning efficiency.

Jonathan Erskine, Taku Yamagata, Raúl Santos-Rodríguez

ZeST uses LLMs for zero-shot traversability navigation in unknown environments, enhancing safety and efficiency.

Shreya Gummadi, Mateus V. Gasparino, Gianluca Capezzuto et al.

This study evaluates LVLMs' knowledge boundary perception using probabilistic, answer consistency, and verbal confidence signals, revealing significant room for improvement.

Zhikai Ding, Shiyu Ni, Keping Bi

Proposes the Experience-driven Lifelong Learning (ELL) framework integrating knowledge abstraction, skill transfer, and internalization, evaluated on the StuLife benchmark for autonomous self-evolving agents.

Yuxuan Cai, Yipeng Hao, Jie Zhou et al.

Drawing2CAD employs sequence-to-sequence learning to convert SVG vector drawings into parametric CAD operation sequences.

Feiwei Qin, Shichao Lu, Junhao Hou et al.