SFT or RL for Tool-Calling Agents? A Controlled Study Across Data, Method, and Scale

arXiv:2609.17848 · cs.CL · Submitted 2026-09-15 · Read on arXiv

cs.CL

Submitted: 2026-09-15

Updated: 2026-09-15

Comments: Accepted to the REALM Workshop at EMNLP 2026

Code: https://github.com/talkiq/dialpad-ai-research

License: http://creativecommons.org/licenses/by/4.0/

The gist: Limited controlled evidence exists on how training data, adaptation method, and model scale jointly affect tool-calling performance in language-model agents.

Terminology

Abstract

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.

Sources

Related papers