Tag: General

Chungpa Lee, Jy-yong Sohn, Kangwook Lee · Feb 26, 2026 · Citations: 0

Xiaoxi Li, Wenxiang Jiao, Jiarui Jin, Shijian Wang, Guanting Dong, Jiajie Jin · Feb 26, 2026 · Citations: 0

• Human intelligence naturally intertwines omni-modal perception -- spanning vision, audio, and language -- with complex reasoning and tool usage to interact with the world.
• To bridge this gap, we introduce OmniGAIA, a comprehensive benchmark designed to evaluate omni-modal agents on tasks necessitating deep reasoning and multi-turn tool execution across video, audio, and image modalities.

Phil Blandfort, Tushar Karayil, Urja Pawar, Robert Graham, Alex McKenzie, Dmitrii Krasheninnikov · Feb 26, 2026 · Citations: 0

• Moral benchmarks for LLMs typically use context-free prompts, implicitly assuming stable preferences.
• We introduce a pilot evaluation harness for directed contextual influence in trolley-problem-style moral triage: for each demographic factor, we apply matched, direction-flipped contextual influences that differ only in which group they fav

Zhe Yang, Yudong Wang, Rang Li, Zhifang Sui · Feb 26, 2026 · Citations: 0

Abhay Sheshadri, Aidan Ewart, Kai Fronsdal, Isha Gupta, Samuel R. Bowman, Sara Price · Feb 26, 2026 · Citations: 0

• We introduce AuditBench, an alignment auditing benchmark.
• To demonstrate AuditBench's utility, we develop an investigator agent that autonomously employs a configurable set of auditing tools.

Fengyu Li, Junhao Zhu, Kaishi Song, Lu Chen, Zhongming Yao, Tianyi Li · Feb 26, 2026 · Citations: 0

• Experiments on two benchmark datasets show that, with the same LLM backbone, Operation-R1 achieves average absolute accuracy gains of 9.55 and 6.08 percentage points over multi-step preparation baselines, with 79\% table compression and a 2

Qianben Chen, Tianrui Qin, King Zhu, Qiexiang Wang, Chengjun Yu, Shu Xu · Feb 26, 2026 · Citations: 0

• Recent deep research agents primarily improve performance by scaling reasoning depth, but this leads to high inference cost and latency in search-intensive scenarios.
• In this work, we propose \emph{Search More, Think Less} (SMTL), a framework for long-horizon agentic search that targets both efficiency and generalization.

Tianle Xia, Ming Xu, Lingxiang Hu, Yiding Sun, Wenwei Li, Linfang Shang · Feb 26, 2026 · Citations: 0

• Agentic RAG addresses this by enabling LLMs to dynamically decide when and what to retrieve, but current RL-based training methods suffer from sparse outcome rewards that discard intermediate signals and low sample efficiency where failed s
• We propose Search-P1, a framework that introduces path-centric reward shaping for agentic RAG training, comprising two key components: (1) Path-Centric Reward, which evaluates the structural quality of reasoning trajectories through order-a

Bo-Wei Chen, Chung-Chi Chen, An-Zi Yen · Feb 25, 2026 · Citations: 0

• Experiments on the Massive Multitask Language Understanding (MMLU) benchmark show that our approach achieves accuracy comparable to the largest model while reducing computational costs by 20\% to 40\%.

Yanbin Wei, Jiangyue Yan, Chun Kang, Yang Chen, Hua Liu, James Kwok · Feb 25, 2026 · Citations: 0

• This ``one-size-fits-all'' strategy often neglects model-specific and task-specific preferences, resulting in inaccurate or over-lengthy responses to graph-related queries.

Jabez Magomere, Elena Kochkina, Samuel Mensah, Simerjot Kaur, Fernando Acero, Arturo Oncevay · Feb 25, 2026 · Citations: 0

• Across six evaluation settings, our trained 8B decomposer improves downstream verification performance to (71.75%) macro-F1, outperforming prompt-based approaches ((+1.99), (+6.24)) and existing RL methods ((+5.84)).
• Human evaluation confirms the high quality of the generated subclaims.

Mustafa Dogan, Ilker Kesen, Iacer Calixto, Aykut Erdem, Erkut Erdem · Feb 25, 2026 · Citations: 0

• In this paper, we introduce FewMMBench, a comprehensive benchmark designed to evaluate MLLMs under few-shot conditions, with a focus on In-Context Learning (ICL) and Chain-of-Thought (CoT) prompting.

Shunsuke Ubukata · Feb 25, 2026 · Citations: 0

Hyo Jin Kim · Feb 25, 2026 · Citations: 0

• Although initially formulated for human truth-telling under asymmetric stakes, the same phase-dynamic architecture extends to AI systems operating under policy constraints and alignment filters.
• The framework therefore provides a unified structural account of both human silence under pressure and AI preference-driven distortion.

Shiqi Yan, Yubo Chen, Ruiqi Zhou, Zhengxi Yao, Shuai Chen, Tianyi Zhang · Feb 25, 2026 · Citations: 0

• Extensive experiments on five KGQA benchmark datasets demonstrate that, to the best of our knowledge, our method achieves state-of-the-art performance, outperforming not only open-source but also even closed-source LLMs.

Tomoya Kawabe, Rin Takano · Feb 25, 2026 · Citations: 0

• We present a hierarchical multi-agent LLM-based planner with prompt optimization: an upper layer decomposes tasks and assigns them to lower-layer agents, which generate PDDL problems solved by a classical planner.
• When plans fail, the system applies TextGrad-inspired textual-gradient updates to optimize each agent's prompt and thereby improve planning accuracy.

Zhijiang Tang, Linhua Wang, Jiaxin Qi, Weihao Jiang, Peng Hou, Anxiang Zeng · Feb 25, 2026 · Citations: 0

• Image captioning remains a fundamental task for vision language understanding, yet ground-truth supervision still relies predominantly on human-annotated references.
• Because human annotations reflect subjective preferences and expertise, ground-truth captions are often incomplete or even incorrect, which in turn limits caption models.

Yukun Chen, Jiaming Li, Longze Chen, Ze Gong, Jingpeng Li, Zhen Qin · Feb 25, 2026 · Citations: 0

• Extensive experiments on various visual reasoning benchmarks show that RuCL yields a remarkable +7.83% average improvement over the Qwen2.5-VL-7B model, achieving a state-of-the-art accuracy of 60.06%.

Xiaoxuan Wang, Han Zhang, Haixin Wang, Yidan Shi, Ruoyan Li, Kaiqiao Han · Feb 25, 2026 · Citations: 0

• Agentic reinforcement learning (ARL) has rapidly gained attention as a promising paradigm for training agents to solve complex, multi-step interactive tasks.
• Through this fine-grained analysis, we distill a unified perspective on ARL and propose SAMPO, a stable agentic policy optimization method designed to mitigate the dominant sources of instability in ARL.

Yue Yang, Shuo Cheng, Yu Fang, Homanga Bharadhwaj, Mingyu Ding, Gedas Bertasius · Feb 25, 2026 · Citations: 0

• We introduce a 21-task simulation benchmark consisting of two challenging suites: LIBERO-Long++ and Ultra-Long.
• Furthermore, real-world evaluations across 8 long-horizon tasks demonstrate an average success rate of 85%.