New framework enhances GUI agent reward modeling accuracy

Researchers have developed AdaptRubric, a new framework designed to improve the accuracy of reward modeling for graphical user interface (GUI) agents. This framework constructs task-adaptive judging criteria by first identifying the task family and retrieving relevant criteria, then refining these criteria to match the specific instruction's constraints. AdaptRubric has demonstrated superior performance over existing methods in both offline reward evaluation and online reinforcement learning, showing significant gains in F1 score and task success rate. AI

影响 This framework could lead to more capable and reliable GUI agents by improving their ability to understand and execute user instructions.

排序理由 The cluster contains an academic paper detailing a new framework and its experimental results. [lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

• AdaptRubric
• arXiv

• 论文
• 模型发布

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →