Explore · 每周精选
发现最受关注的研究论文,追踪研究趋势,订阅感兴趣的期刊与关键词。
Tianxin Wei, Ting-Wei Li, Zhining Liu +26 authors
Agentic reasoning redefines large language models as autonomous agents capable of planning, acting, and learning through continuous interaction in dynamic environments across single-agent and multi-agent frameworks.
50 篇论文 · 按点赞排序
Fengkai Yang, Zherui Chen, Xiaohan Wang +10 authors
Group-based reinforcement learning from verifier rewards suffers from biased advantage estimation that underestimates hard prompts and overestimates easy prompts, which is addressed through a history-aware adaptive difficulty weighting method that improves performance on mathematical reasoning benchmarks.
Taofeng Xue, Chong Peng, Mianqiu Huang +13 authors
EvoCUA introduces an evolutionary approach to computer-use agents that combines autonomous task generation with policy optimization to achieve superior performance in complex, long-horizon tasks.
Daixuan Cheng, Shaohan Huang, Yuxian Gu +6 authors
LLM-in-Sandbox enables large language models to perform general intelligence tasks across diverse domains by allowing them to explore a code sandbox environment, achieving robust generalization without additional training.
Hangrui Hu, Xinfa Zhu, Ting He +13 authors
The Qwen3-TTS series presents advanced multilingual text-to-speech models with voice cloning and controllable speech generation capabilities, utilizing dual-track LM architecture and specialized speech tokenizers for efficient streaming synthesis.
Hao Luo, Ye Wang, Wanpeng Zhang +9 authors
Being-H0.5 is a Vision-Language-Action model that enables robust cross-embodiment generalization through human-centric learning and a Mixture-of-Transformers architecture with specialized embodiment handling.
Zanlin Ni, Shenzhi Wang, Yang Yue +8 authors
Arbitrary order generation in diffusion large language models limits reasoning capability by causing premature solution space collapse, making standard policy optimization more effective.
Haowei Zhang, Shudong Yang, Jinlan Fu +2 authors
HERMES is a training-free architecture that enables real-time video stream understanding by utilizing a hierarchical memory framework based on KV cache reuse, achieving faster response times and maintained accuracy even with reduced video token input.
Jie Yang, Honglin Guo, Li Ji +11 authors
ABC-Bench evaluates LLM agents on realistic backend coding tasks requiring full development lifecycle management from repository exploration to containerized service deployment and API testing.
Caihua Li, Lianghong Guo, Yanlin Wang +9 authors
Large language models face significant challenges in software issue resolution, prompting the development of autonomous coding agents through various training-free and training-based methodologies.
Sunzhu Li, Jiale Zhao, Miteto Wei +6 authors
RLVR has advanced reasoning capabilities but struggles with open-ended generation due to lack of ground truth; this work proposes an automated rubric generation framework and dataset to improve performance in health reasoning benchmarks.
Xiaofang Yang, Lijun Li, Heng Zhou +13 authors
Efficiency in agentic systems is examined across memory, tool learning, and planning components, analyzing trade-offs between effectiveness and computational costs through various optimization strategies and benchmarks.
Chenghao Fan, Wen Heng, Bo Li +6 authors
Stable-DiffCoder demonstrates superior code modeling performance compared to autoregressive baselines through block diffusion continual pretraining and efficient training mechanisms.
Shengbang Tong, Boyang Zheng, Ziteng Wang +7 authors
Representation Autoencoders (RAEs) demonstrate superior performance over VAEs in large-scale text-to-image generation, showing improved stability, faster convergence, and better quality while enabling unified multimodal reasoning in shared representation spaces.
Shijie Lian, Bin Yu, Xiaopeng Lin +6 authors
BayesianVLA addresses language-action grounding issues in robot manipulation by using Bayesian decomposition to prevent information collapse and improve out-of-distribution generalization.
Qianli Ma, Chang Guo, Zhiheng Tian +4 authors
RebuttalAgent is a multi-agent framework that reframes rebuttal generation as an evidence-centric planning task, improving coverage, faithfulness, and strategic coherence in academic peer review.
Peizhou Huang, Zixuan Zhong, Zhongwei Wan +12 authors
MMDeepResearch-Bench evaluates multimodal research agents on report generation with visual evidence, revealing trade-offs between prose quality, citation accuracy, and visual grounding.
Hengyuan Zhang, Zhihao Zhang, Mingyang Wang +25 authors
Mechanistic interpretability is presented as an actionable framework for understanding and optimizing large language models through systematic localization, steering, and improvement methods.
Pengze Zhang, Yanze Wu, Mengtian Li +8 authors
OmniTransfer presents a unified framework for spatio-temporal video transfer that enhances appearance consistency and temporal control through multi-view information and multimodal semantic guidance.
Zaibin Zhang, Yuhan Wu, Lianjie Jia +9 authors
Think3D enhances vision-language models' 3D reasoning capabilities by enabling interactive spatial exploration through 3D reconstruction and camera-based operations, improving performance without additional training.
Eilam Shapira, Roi Reichart, Moshe Tennenholtz
The integration of AI agents into economic markets fundamentally alters the landscape of strategic interaction. We investigate the economic implications of expanding the set of available technologies in three canonical game-theoretic settings: bargaining (resource division), negotiation (asymmetric information trade), and persuasion (strategic information transmission). We find that simply increasing the choice of AI delegates can drastically shift equilibrium payoffs and regulatory outcomes, often creating incentives for regulators to proactively develop and release technologies. Conversely, we identify a strategic phenomenon termed the "Poisoned Apple" effect: an agent may release a new technology, which neither they nor their opponent ultimately uses, solely to manipulate the regulator's choice of market design in their favor. This strategic release improves the releaser's welfare at the expense of their opponent and the regulator's fairness objectives. Our findings demonstrate that static regulatory frameworks are vulnerable to manipulation via technology expansion, necessitating dynamic market designs that adapt to the evolving landscape of AI capabilities.
Yufan Deng, Zilin Pan, Hongyu Zhang +6 authors
A comprehensive robotics benchmark evaluates video generation models across multiple task domains and robot embodiments, revealing significant gaps in physical realism and introducing a large-scale dataset to address training data limitations.
Mert Yuksekgonul, Daniel Koceja, Xinhao Li +8 authors
Test-time training enables AI systems to discover optimal solutions for specific scientific problems through continual learning focused on individual challenges rather than generalization.
Yikang Zhou, Tao Zhang, Dengxian Gong +13 authors
SAMTok enables pixel-wise capabilities in multi-modal large language models through discrete mask tokenization and standard training methods, achieving state-of-the-art performance on various vision-language tasks.
Zhihao Xu, Rumei Li, Jiahuan Li +4 authors
A text-based data synthesis approach generates multi-turn tool-use trajectories for large language models, achieving improved performance and reduced computational costs through a specialized trajectory synthesizer.
Yao Tang, Li Dong, Yaru Hao +3 authors
Multiplex Thinking introduces a stochastic soft reasoning mechanism that samples multiple candidate tokens at each step to optimize reasoning trajectories with reinforcement learning while maintaining shorter sequences than traditional chain-of-thought methods.
Hunter Heidenreich, Ben Elliott, Olivia Dinica +1 authors
GutenOCR enhances vision-language models for document understanding by enabling unified reading, detection, and grounding through prompt-based interfaces trained on diverse document types.
Mike A. Merrill, Alexander G. Shaw, Nicholas Carlini +82 authors
Terminal-Bench 2.0 presents a challenging benchmark with 89 terminal-based tasks to evaluate AI agents' capabilities in real-world scenarios.
Qian Chen, Jinlan Fu, Changsong Li +2 authors
FutureOmni presents the first benchmark for evaluating multimodal models' ability to forecast future events from audio-visual data, revealing current limitations and proposing an improved training strategy for better performance.
Keyu Li, Junhao Shi, Yang Xiao +10 authors
AgencyBench presents a comprehensive benchmark for evaluating autonomous agents across real-world scenarios, enabling automated evaluation through user simulation and sandbox environments while revealing performance gaps between closed-source and open-source models.
北京市昌平区探索星信息技术及软件开发工作室
京ICP备2026059466号