AgentCPM-Report: 交错起草与深入以促进开放性深度研究
AgentCPM-Report: Interleaving Drafting and Deepening for Open-Ended Deep Research
AI总结：
AgentCPM-Report通过交错起草与深入的方法，提升开放性深度研究的性能，优于现有闭源系统。
摘要：
生成深度研究报告需要大规模信息获取和基于洞察的分析综合，这对当前语言模型构成了重大挑战。大多数现有方法遵循计划后写作的范式，其性能严重依赖于初始大纲的质量。然而，构建一个全面的大纲本身需要强大的推理能力，导致当前深度研究系统几乎完全依赖闭源或在线大模型。这种依赖性提高了部署的实际障碍，并引入了用户生成数据的安全性和隐私问题。在本工作中，我们提出了AgentCPM-Report，一个轻量但高性能的本地解决方案，由一个模仿人类写作过程的框架和一个8B参数的深度研究代理组成。我们的框架使用写作作为推理策略（WARP），使模型能够在报告生成过程中动态修订大纲。在该策略下，代理在基于证据的起草和基于推理的深入之间交替，共同支持信息获取、知识细化和迭代大纲进化。为了有效使小模型具备这一能力，我们引入了多阶段代理训练策略，包括冷启动、原子技能强化学习和整体流程强化学习。在DeepResearch Bench、DeepConsult和DeepResearch Gym上的实验表明，AgentCPM-Report优于领先的闭源系统，在洞察方面有显著提升。
英文摘要：
Generating deep research reports requires large-scale information acquisition and the synthesis of insight-driven analysis, posing a significant challenge for current language models. Most existing approaches follow a plan-then-write paradigm, whose performance heavily depends on the quality of the initial outline. However, constructing a comprehensive outline itself demands strong reasoning ability, causing current deep research systems to rely almost exclusively on closed-source or online large models. This reliance raises practical barriers to deployment and introduces safety and privacy concerns for user-authored data. In this work, we present AgentCPM-Report, a lightweight yet high-performing local solution composed of a framework that mirrors the human writing process and an 8B-parameter deep research agent. Our framework uses a Writing As Reasoning Policy (WARP), which enables models to dynamically revise outlines during report generation. Under this policy, the agent alternates between Evidence-Based Drafting and Reasoning-Driven Deepening, jointly supporting information acquisition, knowledge refinement, and iterative outline evolution. To effectively equip small models with this capability, we introduce a Multi-Stage Agentic Training strategy, consisting of cold-start, atomic skill RL, and holistic pipeline RL. Experiments on DeepResearch Bench, DeepConsult, and DeepResearch Gym demonstrate that AgentCPM-Report outperforms leading closed-source systems, with substantial gains in Insight.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
