LiteGUI: Distilling Compact GUI Agents with Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wu, Yubin, Cai, Zicheng, Ning, Liping, Wang, Hua, Chen, Zhi, Tang, Yaohua, Chen, Hao
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910200880955392
author Wu, Yubin
Cai, Zicheng
Ning, Liping
Wang, Hua
Chen, Zhi
Tang, Yaohua
Chen, Hao
author_facet Wu, Yubin
Cai, Zicheng
Ning, Liping
Wang, Hua
Chen, Zhi
Tang, Yaohua
Chen, Hao
contents Developing lightweight, on-device vision-language GUI agents is essential for efficient cross-platform automated interaction. However, current on-device agents are constrained by limited model capacity, and further performance improvements remain urgently needed. Traditional Supervised Fine-Tuning (SFT) for small-scale models often leads to overfitting, catastrophic forgetting and policy rigidity, and thus fails to fully address these challenges. In this work, we propose a novel SFT-free training paradigm that significantly enhances the performance of small-scale models. We first present the initial systematic integration of generalized knowledge distillation into the GUI agent domain via Guided On-policy Distillation. By incorporating oracle reference trajectories together with a dynamic retrieval mechanism, our method reduces hallucinations and mitigates the cognitive misalignment inherent in multi-solution GUI tasks. Building on this foundation, we further introduce a Multi-solution Dual-level GRPO framework that jointly aligns macro-level subtask planning with micro-level execution matching, thereby improving exploration in long-horizon GUI agent scenarios. In addition, we construct an automated data generation pipeline to synthesize GUI task trajectories with rich multi-solution annotations. Extensive experiments show that our method achieves state-of-the-art performance among lightweight models while remaining competitive with substantially larger-scale models across all benchmarks. Ablation studies further demonstrate that structured on-policy distillation and multi-solution dual-level exploration can fully unlock the capabilities of 2B/3B scale agents, surpassing the performance limits of conventional imitation learning.
format Preprint
id arxiv_https___arxiv_org_abs_2605_07505
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle LiteGUI: Distilling Compact GUI Agents with Reinforcement Learning
Wu, Yubin
Cai, Zicheng
Ning, Liping
Wang, Hua
Chen, Zhi
Tang, Yaohua
Chen, Hao
Artificial Intelligence
Machine Learning
Developing lightweight, on-device vision-language GUI agents is essential for efficient cross-platform automated interaction. However, current on-device agents are constrained by limited model capacity, and further performance improvements remain urgently needed. Traditional Supervised Fine-Tuning (SFT) for small-scale models often leads to overfitting, catastrophic forgetting and policy rigidity, and thus fails to fully address these challenges. In this work, we propose a novel SFT-free training paradigm that significantly enhances the performance of small-scale models. We first present the initial systematic integration of generalized knowledge distillation into the GUI agent domain via Guided On-policy Distillation. By incorporating oracle reference trajectories together with a dynamic retrieval mechanism, our method reduces hallucinations and mitigates the cognitive misalignment inherent in multi-solution GUI tasks. Building on this foundation, we further introduce a Multi-solution Dual-level GRPO framework that jointly aligns macro-level subtask planning with micro-level execution matching, thereby improving exploration in long-horizon GUI agent scenarios. In addition, we construct an automated data generation pipeline to synthesize GUI task trajectories with rich multi-solution annotations. Extensive experiments show that our method achieves state-of-the-art performance among lightweight models while remaining competitive with substantially larger-scale models across all benchmarks. Ablation studies further demonstrate that structured on-policy distillation and multi-solution dual-level exploration can fully unlock the capabilities of 2B/3B scale agents, surpassing the performance limits of conventional imitation learning.
title LiteGUI: Distilling Compact GUI Agents with Reinforcement Learning
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2605.07505