Metraipolis › archive › dc_7c9ea548e9d5
GRADS: Graph-Regulated Agentic Data Synthesis with Controllable Difficulty — arXiv:2610.21167
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.21167
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T08:59:46.788Z
- SHA-256 of the archived text
5d96c08733d878bbb047face7231dc6f7bace666fed135fd91cfb5ba0c93fcc3- Stable URL of this page
https://metraipolis.com/documento/dc_7c9ea548e9d5
How to cite
GRADS: Graph-Regulated Agentic Data Synthesis with Controllable Difficulty — arXiv:2610.21167 — arXiv (en), https://arxiv.org/abs/2610.21167 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7c9ea548e9d5 (Metraipolis archive dc_7c9ea548e9d5, sha256 5d96c08733d878bbb047face7231dc6f7bace666fed135fd91cfb5ba0c93fcc3)
BibTeX
@misc{dc7c9ea548e9d5,
title = {GRADS: Graph-Regulated Agentic Data Synthesis with Controllable Difficulty — arXiv:2610.21167},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.21167},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7c9ea548e9d5; sha256 5d96c08733d878bbb047face7231dc6f7bace666fed135fd91cfb5ba0c93fcc3},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.21167. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Deep research agents are increasingly deployed to retrieve precise information from large corpora, yet multi-hop, long-horizon tasks remain challenging. Synthetic data generation has emerged as a scalable solution, with existing pipelines typically expanding seed tasks to increase their apparent depth and breadth. However, this expansion does not reliably translate into necessary reasoning: newly introduced evidence may be only weakly coupled to the answer, creating a mismatch between intended and empirical difficulty. To address this problem, we introduce GRADS, a graph-guided agentic data synthesis framework with explicit difficulty control. GRADS runs a closed loop of seed selection, evidence expansion, question--answer construction, and verification while maintaining a graph of the evidence and inference dependencies required to derive each answer. At each step, it conditions generation on a relevant subgraph, preserving indispensable inference without accumulating irrelevant context. Experiments across security, software-engineering, and scientific-document domains show that GRADS consistently generates verified tasks with empirically separated difficulty distributions, as measured by independent solvers. Moreover, under matched training budgets, post-training on this data delivers consistent gains at every level: SFT raises the solution-termination rate to over 95% while cutting tool calls and turns by up to 75%, RL trims tool use by up to a further 19.5%, and on our held-out, domain-specific benchmarks, the resulting 4B model matches Sonnet~4 on software engineering and surpasses it on the scientific and security domains. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.21167 — 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 2026-10-16T08:44:17.708Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).