Metraipolis › archive › dc_a25fa336f35b

GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis — arXiv:2609.38923

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.38923
License (as archived)
CC0-1.0
Archived
2026-10-07T13:13:35.865Z
SHA-256 of the archived text
d27275dbe0561ae102af7a3bfe5dc12b8a2e3937f7eedf156dc87311d2bde0f7
Stable URL of this page
https://metraipolis.com/documento/dc_a25fa336f35b

How to cite

GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis — arXiv:2609.38923 — arXiv (en), https://arxiv.org/abs/2609.38923 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a25fa336f35b (Metraipolis archive dc_a25fa336f35b, sha256 d27275dbe0561ae102af7a3bfe5dc12b8a2e3937f7eedf156dc87311d2bde0f7)
BibTeX
@misc{dca25fa336f35b,
  title = {GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis — arXiv:2609.38923},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38923},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a25fa336f35b; sha256 d27275dbe0561ae102af7a3bfe5dc12b8a2e3937f7eedf156dc87311d2bde0f7},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38923. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

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.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.38923 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T12:44:51.775Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).