Metraipolis › archive › dc_0d5ca02ed509
SFT or RL for Tool-Calling Agents? A Controlled Study Across Data, Method, and Scale — arXiv:2609.17848
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.17848
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T13:49:16.919Z
- SHA-256 of the archived text
57d91d169ee01d2f4f6ea192e41e37015fb3eeb5e9224517a4268e6e8d87693d- Stable URL of this page
https://metraipolis.com/documento/dc_0d5ca02ed509
How to cite
SFT or RL for Tool-Calling Agents? A Controlled Study Across Data, Method, and Scale — arXiv:2609.17848 — arXiv (en), https://arxiv.org/abs/2609.17848 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0d5ca02ed509 (Metraipolis archive dc_0d5ca02ed509, sha256 57d91d169ee01d2f4f6ea192e41e37015fb3eeb5e9224517a4268e6e8d87693d)
BibTeX
@misc{dc0d5ca02ed509,
title = {SFT or RL for Tool-Calling Agents? A Controlled Study Across Data, Method, and Scale — arXiv:2609.17848},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.17848},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0d5ca02ed509; sha256 57d91d169ee01d2f4f6ea192e41e37015fb3eeb5e9224517a4268e6e8d87693d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.17848. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Limited controlled evidence exists on how training data, adaptation method, and model scale jointly affect tool-calling performance in language-model agents. We evaluate supervised fine-tuning (SFT) with LoRA, reinforcement learning (RL) via Group Relative Policy Optimization (GRPO), and SFT followed by GRPO across six Qwen3 models from 0.6B to 32B parameters, covering both in-distribution performance and cross-dataset transfer. SFT with LoRA is the strongest in-distribution method throughout the 0.6B-32B range and best in 15 out of 18 experimental settings. On cross-dataset transfer, the methods are closer: GRPO wins 29 out of 54 settings where training and test datasets differ, but its margin over SFT averages under one point, and SFT->GRPO is rarely strongest in either comparison. Dataset mixing gives consistently strong transfer while staying close to specialized in-distribution training, regardless of method. Additional analysis further confirms that LoRA outperforms full-parameter fine-tuning, demonstrating that LoRA better preserves pretrained agentic behavior. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-17 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.17848 — 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 2026-10-15T13:48:24.281Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-17?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).