Rapid advances in cowork agents are shifting human-AI collaboration toward real-world automation, delivering software, designs, and other artifacts. However, such productivity work demands both high-quality delivery on complex productivity tasks and continued monitoring and decision-making over long horizons, a combina...
Jiajie Jin, Guan-Ting Dong, Cheng-Jun Pan et al.· 0 citations
Large language models are deployed as long-horizon agents that compose diverse tools across MCP services into executable workflows. Existing MCP environment-scaling methods are data-centric: they first synthesize a database and scenario, then derive tools and tasks from them. As a result, tool interfaces may not match...
Guan-Ting Dong, Zhi-Chao Sun, Xin-Yu Yang et al.· 0 citations
Large language model (LLM)-based agents have demonstrated strong capabilities on complex tasks. They typically perform reasoning before each action throughout an interaction trajectory. However, reasoning may not be necessary at every turn, as reasoning produced earlier can continue to support subsequent actions. A key...
Yi-Ruo Cheng, Shen Huang, Xiaoshuai Song et al.· 0 citations
Results show that long-horizon reflective data is an effective route toward self-improving agents, and synthesize long-horizon improvement trajectories from machine learning and algorithmic programming tasks, two domains that offer verifiable feedback and reward sustained iteration.
Hong-Jin Qian, Chao-Fan Li, Kun Luo et al.· 0 citations
StepLearn is introduced, a nonparametric framework that separates immediate use from persistent trust, while requiring prospective validation before reuse across episodes, and turns informative transitions into hypotheses that can guide the next step.
Experiments on three diverse long-horizon data-analysis benchmarks show that StateGuard consistently improves data-agent performance while reducing dependency-induced downstream error propagation, demonstrating the advantages of explicit analytical-state management for reliable long-horizon data analysis.
Wen-Le Liao, Zhao Wang, Jing-Chao Zhang et al.· 0 citations
This work introduces EvolveScaler, a code-driven framework that defines information evolution before rendering it as natural language, and shows that code-driven IE synthesis provides both challenging evaluation and transferable training supervision.
Ziliang Zhao, Ze-Nan Xu, Shuting Wang et al.· 0 citations
DisCo is presented, a skill-powered research agent that creates skills and uses them during research, and yields the AREX-Skill Library, with 5,000+ verified skills distilled from 1,000 widely used ML repositories and organized into 20 areas and 178 capability families.
Jianlyu Chen, Yuyang Hu, Hong-Jin Qian et al.· 1 citation
LOFA combines reinforcement learning over verifiable purchase outcomes with feedback-aware on-policy distillation, which identifies users' in-dialogue directives and converts them into dense token-level supervision, which captures both collaborative behavioral patterns and user-specific preferences.
Haobo Zhang, Ke-Long Mao, Su-Long Xu et al.· 1 citation
This work introduces AREX, a family of Recursively Self-Improving (RSI) deep research agents that substantially outperforms comparable-scale baselines and remains competitive with models using substantially more activated parameters.
Shuqi Lu, Chaofan Li, Kun Luo et al.· arXiv.org· 2 citations· ⚡1
On-Policy Omni Distillation (OPOD), which consolidates text, image, and audio teachers into one omni model, and surpasses the base model and pooled RL training on all twelve benchmarks, and ranks first or second on eleven even when the teachers are included.
Tong Zhao, Yuyang Hu, Reed Li et al.· arXiv.org· 0 citations
This work proposes WebSwarm, a progressive recursive delegation framework that jointly constructs task decomposition, recursive expansion, and agent collaboration during inference and consistently outperforms single-agent and multi-agent baselines on deep, wide, and interleaved deep-and-wide tasks.