WorkFactory：将专业意图转化为可执行工作以扩展智能体训练
原标题：Paper page - WorkFactory: Turning Professional Intents into Executable Work for Scalable Agent Training
AI 摘要
研究团队提出 WorkFactory 框架，通过信息规划与修复从专业意图构建连贯可执行的专业任务，并据此构建了覆盖 612 种职业的 112,686 个任务。团队用收集到的教师轨迹微调 Qwen 模型（4B、27B、35B-A3B 规模），得到 WorkPilot 系列，三个模型在四个专业工作基准上均优于各自基座模型，其中 WorkPilot-35B-A3B 在 GDPval-AA v2.1 上从 879 Elo 提升至 1565 Elo。数据和模型已在 GitHub 开源。
正文节选
WorkFactory: Turning Professional Intents into Executable Work for Scalable Agent Training Abstract WorkFactory builds coherent professional tasks via information planning and repair, yielding WorkPilot models that improve expert-level agent performance across benchmarks. Scaling supervised training for working agents requires diverse professional tasks that yield reliable expert demonstrations. However, missing or inconsistent task information can obstruct teacher execution and weaken the resul