Search papers, labs, and topics across Lattice.
The paper introduces LAMO, a framework for creating lightweight GUI agents based on MLLMs that can participate in multi-agent systems. LAMO uses role-oriented data synthesis and a two-stage training process involving supervised fine-tuning with Perplexity-Weighted Cross-Entropy and reinforcement learning for cooperative exploration. Experiments demonstrate that LAMO-3B, a 3B parameter agent built with this framework, achieves task scalability and can be effectively orchestrated in multi-agent GUI automation scenarios.
Lightweight MLLMs can punch far above their weight in GUI automation when given the right training diet and a little help from their friends.
Autonomous Graphical User Interface (GUI) agents powered by Multimodal Large Language Models (MLLMs) enable digital automation on end-user devices. While scaling both parameters and data has yielded substantial gains, advanced methods still suffer from prohibitive deployment costs on resource-constrained devices. When facing complex in-the-wild scenarios, lightweight GUI agents are bottlenecked by limited capacity and poor task scalability under end-to-end episodic learning, impeding adaptation to multi-agent systems (MAS), while training multiple skill-specific experts remains costly. Can we strike an effective trade-off in this cost-scalability dilemma, enabling lightweight MLLMs to participate in realistic GUI workflows? To address these challenges, we propose the LAMO framework, which endows a lightweight MLLM with GUI-specific knowledge and task scalability, allowing multi-role orchestration to expand its capability boundary for GUI automation. LAMO combines role-oriented data synthesis with a two-stage training recipe: (i) supervised fine-tuning with Perplexity-Weighted Cross-Entropy optimization for knowledge distillation and visual perception enhancement, and (ii) reinforcement learning for role-oriented cooperative exploration. With LAMO, we develop a task-scalable native GUI agent, LAMO-3B, supporting monolithic execution and MAS-style orchestration. When paired with advanced planners as a plug-and-play policy executor, LAMO-3B can continuously benefit from planner advances, enabling a higher performance ceiling. Extensive static and online evaluations validate the effectiveness of our design.
