LiteGUI: Distilling Compact GUI Agents with Reinforcement Learning

• URL: http://arxiv.org/abs/2605.07505v1
• Date: Fri, 08 May 2026 09:38:29 GMT
• Title: LiteGUI: Distilling Compact GUI Agents with Reinforcement Learning
• Abstract summary: Current on-device vision-language GUI agents are constrained by limited model capacity.<n>TraditionalSupervised Fine-Tuning for small-scale models often leads to overfitting, catastrophic forgetting and policy rigidity.<n>We propose a novel SFT-free training paradigm that significantly enhances the performance of small-scale models.
• Score: 11.35341244051162
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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.

Related papers

• GUI Agents with Reinforcement Learning: Toward Digital Inhabitants [21.438394192355407]
  We present the first comprehensive overview of the intersection between RL and GUI agents.<n>We propose a principled taxonomy that organizes existing methods into Offline RL, Online RL, and Hybrid Strategies.<n>We distill these findings into a roadmap covering process rewards, continual RL, cognitive architectures, and safe deployment.
  arXiv  Detail & Related papers  (2026-04-30T14:51:49Z)
• Towards Scalable Lightweight GUI Agents via Multi-role Orchestration [26.13785805540596]
  We present the LAMO framework, which endows a lightweight MLLM with GUI-specific knowledge and task scalability.<n>We develop a task-scalable native GUI agent, LAMO-3B, supporting monolithic execution and MAS-style orchestration.
  arXiv  Detail & Related papers  (2026-04-15T05:23:04Z)
• UI-Voyager: A Self-Evolving GUI Agent Learning via Failed Experience [56.50613398808361]
  We propose a novel two-stage self-evolving mobile GUI agent.<n>In the first stage, we employ Rejection Fine-Tuning (RFT), which enables the continuous co-evolution of data and models in a fully autonomous loop.<n>The second stage introduces Group Relative Self-Distillation (GRSD), which identifies critical fork points in group rollouts and constructs dense step-level supervision from successful trajectories to correct failed ones.
  arXiv  Detail & Related papers  (2026-03-25T17:10:29Z)
• Insight-V++: Towards Advanced Long-Chain Visual Reasoning with Multimodal Large Language Models [65.4947731385794]
  We present a unified multi-agent visual reasoning framework that evolves from our foundational image-centric model, Insight-V.<n>We introduce two novel algorithms, ST-GRPO and J-GRPO, which enhance spatial-temporal reasoning and improve evaluative robustness.<n>Experiments on base models like LLaVA-NeXT and Qwen2.5-VL demonstrate significant performance gains across challenging image and video reasoning benchmarks.
  arXiv  Detail & Related papers  (2026-03-18T15:28:07Z)
• CGL: Advancing Continual GUI Learning via Reinforcement Fine-Tuning [67.78566256784404]
  Supervised Fine-Tuning (SFT) facilitates fast adaptation, it often triggers knowledge overwriting.<n>Reinforcement Learning (RL) demonstrates an inherent resilience that shields prior interaction logic from erasure.<n>We propose a textbfContinual textbfGUI textbfLearning framework that balances efficiency and skill retention.
  arXiv  Detail & Related papers  (2026-03-03T13:02:20Z)
• AR-MOT: Autoregressive Multi-object Tracking [56.09738000988466]
  We propose a novel autoregressive paradigm that formulates MOT as a sequence generation task within a large language model (LLM) framework.<n>This design enables the model to output structured results through flexible sequence construction, without requiring any task-specific heads.<n>To enhance region-level visual perception, we introduce an Object Tokenizer based on a pretrained detector.
  arXiv  Detail & Related papers  (2026-01-05T09:17:28Z)
• Mano Technical Report [29.551514304095296]
  Mano is a robust GUI agent built upon a multi-modal foundation model pre-trained on extensive web and computer system data.<n>Mano demonstrates state-of-the-art performance on multiple GUI benchmarks, including Mind2Web and OSWorld.
  arXiv  Detail & Related papers  (2025-09-22T03:13:58Z)
• OneCAT: Decoder-Only Auto-Regressive Model for Unified Understanding and Generation [91.45421429922506]
  OneCAT is a unified multimodal model that seamlessly integrates understanding, generation, and editing.<n>Our framework eliminates the need for external components such as Vision Transformers (ViT) or vision tokenizer during inference.
  arXiv  Detail & Related papers  (2025-09-03T17:29:50Z)
• Learning, Reasoning, Refinement: A Framework for Kahneman's Dual-System Intelligence in GUI Agents [15.303188467166752]
  We present CogniGUI, a cognitive framework developed to overcome limitations by enabling adaptive learning for GUI automation resembling human-like behavior.<n>To assess the generalization and adaptability of agent systems, we introduce ScreenSeek, a comprehensive benchmark that includes multi application navigation, dynamic state transitions, and cross interface coherence.<n> Experimental results demonstrate that CogniGUI surpasses state-of-the-art methods in both the current GUI grounding benchmarks and our newly proposed benchmark.
  arXiv  Detail & Related papers  (2025-06-22T06:30:52Z)
• When Parameter-efficient Tuning Meets General-purpose Vision-language Models [65.19127815275307]
  PETAL revolutionizes the training process by requiring only 0.5% of the total parameters, achieved through a unique mode approximation technique. Our experiments reveal that PETAL not only outperforms current state-of-the-art methods in most scenarios but also surpasses full fine-tuning models in effectiveness.
  arXiv  Detail & Related papers  (2023-12-16T17:13:08Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.