CGL: Advancing Continual GUI Learning via Reinforcement Fine-Tuning

TL;DR

This paper introduces CGL, a framework that combines supervised fine-tuning and reinforcement learning to improve continual GUI learning, effectively balancing adaptation and retention in GUI agents.

Contribution

It proposes a novel CGL framework with dynamic SFT-RL balancing, gradient surgery, and a new AndroidControl-CL benchmark for evaluating continual GUI learning.

Findings

CGL outperforms baseline methods in continual learning scenarios.

The gradient surgery strategy reduces gradient interference.

The AndroidControl-CL benchmark effectively evaluates GUI learning performance.

Abstract

Graphical User Interface (GUI) Agents, benefiting from recent advances in multimodal large language models (MLLM), have achieved significant development. However, due to the frequent updates of GUI applications, adapting to new tasks without forgetting old tasks in GUI continual learning remains an open problem. In this work, we reveal that while Supervised Fine-Tuning (SFT) facilitates fast adaptation, it often triggers knowledge overwriting, whereas Reinforcement Learning (RL) demonstrates an inherent resilience that shields prior interaction logic from erasure. Based on this insight, we propose a \textbf{C}ontinual \textbf{G}UI \textbf{L}earning (CGL) framework that dynamically balances adaptation efficiency and skill retention by enhancing the synergy between SFT and RL. Specifically, we introduce an SFT proportion adjustment mechanism guided by policy entropy to dynamically control the weight allocation between the SFT and RL training phases. To resolve explicit gradient interference, we further develop a specialized gradient surgery strategy. By projecting exploratory SFT gradients onto GRPO-based anchor gradients, our method explicitly clips the components of SFT gradients that conflict with GRPO. On top of that, we establish an AndroidControl-CL benchmark, which divides GUI applications into distinct task groups to effectively simulate and evaluate the performance of continual GUI learning. Experimental results demonstrate the effectiveness of our proposed CGL framework across continual learning scenarios. The benchmark, code, and model will be made publicly available.