arXiv cs.CL· Qisheng Su, Hanchen Wang, Guanru Zhu, Huicheng Jiang, Qiuyinzhe Zhang, Kou Shi, Zhen Fang, Ziao Zhang, Qingnan Ren, Honglin Guo, Jiazheng Zhang, Zehui Chen, Tao Gui, Feng Zhao·· 13 小时前AI 评分43
GraphForge：用图锚定工作区合成训练工作型智能体
GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis
AI 导读
GraphForge 是一个基于证据图的框架，从职业种子出发为每个种子组装真实文件工作区并构建证据图，任务描述与评分标准均由图派生，使每条标准都锚定到可验证的文件。
正文 · 原文
Authors:Qisheng Su, Hanchen Wang, Guanru Zhu, Huicheng Jiang, Qiuyinzhe Zhang, Kou Shi, Zhen Fang, Ziao Zhang, Qingnan Ren, Honglin Guo, Jiazheng Zhang, Zehui Chen, Tao Gui, Feng Zhao
Abstract:Working agents need to read diverse files, coordinate tools, and produce deliverables. Training such agents requires tasks built on many real files with verifiable results, but few pipelines exist to synthesize this kind of data. Existing pipelines either generate files with models, which lack realism and diversity, or build tasks on real files without task-specific verifiers, leaving result quality unchecked. We introduce GraphForge, an evidence-graph based framework that grounds both the task and its verification in real files. Starting from occupation-grounded seeds for controlled diversity, GraphForge assembles a workspace of real files for each seed and builds an evidence graph over their relations. Since the task statement and rubrics are both derived from this graph, task requirements are backed by the workspace files and each criterion is anchored to the files needed to verify it. An initial rollout further tests executability, and a revision agent repairs the task and rubrics against the original files before trajectories are collected. Fine-tuning Qwen3.6-27B on 2,169 GraphForge trajectories brings GDPVal to 1445.7 (+65.7) under OpenHands, and Workspace-Bench-Lite and SpreadsheetBench II to 63.7 (+7.7) and 24.0 (+13.7) under Claude Code. Rejection fine-tuning on the SFT model's own rollouts, with candidates selected by the evidence-anchored rubrics, yields further improvements on all three benchmarks, suggesting that the rubrics provide a useful selection signal.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2609.38923 [cs.CL] |
| (or arXiv:2609.38923v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2609.38923 arXiv-issued DOI via DataCite |
Submission history
From: Qisheng Su [view email]
[v1]
Wed, 30 Sep 2026 04:07:17 UTC (340 KB)
[v2]
Sun, 4 Oct 2026 14:53:13 UTC (340 KB)
[v3]
Thu, 22 Oct 2026 05:33:22 UTC (340 KB)
来源：arXiv cs.CL · arxiv.org