HuggingFace Daily Papers（社区热门论文）· HuggingFace Daily Papers（社区热门论文）·· 2026-04-13精选AI 评分75
ClawGUI：GUI Agent训练、评估与部署的统一开源框架
AI 导读
ClawGUI是面向GUI Agent的开源全栈框架，统一解决训练、评估与部署的基础设施瓶颈。ClawGUI-RL首创支持并行虚拟环境与真实设备的开源强化学习管道，集成GiGPO与过程奖励模型；ClawGUI-Eval在6项基准实现95.8%基线复现率；ClawGUI-Agent支持部署至Android、HarmonyOS、iOS及12个以上聊天平台。经该管道训练的ClawGUI-2B在MobileWorld GUI-Only任务达17.1%成功率，较同规模基线提升6.0%。
推荐理由
这是 GUI agent 方向久等的工程基座，第一次把在线 RL 训练、标准化评估和真实设备部署装进同一个开源框架，2B 模型就能跑赢大尺寸模型，做移动 agent 的团队可以直接上手复现。
正文
Abstract:GUI agents drive applications through their visual interfaces instead of programmatic APIs, interacting with arbitrary software via taps, swipes, and keystrokes, reaching a long tail of applications that CLI-based agents cannot. Yet progress in this area is bottlenecked less by modeling capacity than by the absence of a coherent full-stack infrastructure: online RL training suffers from environment instability and closed pipelines, evaluation protocols drift silently across works, and trained agents rarely reach real users on real devices. We present \textbf{ClawGUI}, an open-source framework addressing these three gaps within a single harness. \textbf{ClawGUI-RL} provides the first open-source GUI agent RL infrastructure with validated support for both parallel virtual environments and real physical devices, integrating GiGPO with a Process Reward Model for dense step-level supervision. \textbf{ClawGUI-Eval} enforces a fully standardized evaluation pipeline across 6 benchmarks and 11+ models, achieving 95.8\% reproduction against official baselines. \textbf{ClawGUI-Agent} brings trained agents to Android, HarmonyOS, and iOS through 12+ chat platforms with hybrid CLI-GUI control and persistent personalized memory. Trained end to end within this pipeline, \textbf{ClawGUI-2B} achieves 17.1\% Success Rate on MobileWorld GUI-Only, outperforming the same-scale MAI-UI-2B baseline by 6.0\%.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2604.11784 [cs.LG] |
| (or arXiv:2604.11784v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2604.11784 arXiv-issued DOI via DataCite |
Submission history
From: Fei Tang [view email]
[v1]
Mon, 13 Apr 2026 17:52:04 UTC (6,573 KB)
来源：HuggingFace Daily Papers（社区热门论文） · arxiv.org