Iron：用于增强通用虚拟智能体的意图对齐与回溯双学习框架
Iron: Intent-Aligned and Retrospective Dual Learning Framework for Enhancing Generalist Virtual Agents
浏览论文内容
中文总结 AI 辅助
Iron是用于训练GUI智能体的意图对齐与回溯双学习框架，通过逐步循环一致奖励和事后回放机制提升性能，在跨环境等任务中优于三倍数据训练的模型，未见过的网页任务获25.06%相对提升。
中文摘要 AI 辅助
实现能在各类数字环境中自动执行任务的虚拟智能体，仍是具身智能领域的关键挑战。多模态大语言模型（MLLM）虽具备更强的视觉感知与推理能力，但其智能体部署面临三大难题：数据标注成本高昂、动作-意图对齐不精确、废弃失败轨迹导致探索效率低下。为解决这些问题，我们提出Iron——一种用于训练GUI智能体的意图对齐、自改进且标注高效的框架。Iron采用新颖的双学习策略，利用逐步循环一致（SCC）奖励实现低层动作与高层意图的细粒度对齐，从而提升指令落地与意图理解能力；同时，Iron引入事后回放机制，将失败轨迹重新用于训练，提高学习效率与任务多样性。大量实验表明，经Iron训练的通用智能体在跨环境、跨设备任务中性能持续提升，表现优于用三倍数据训练的模型；Iron在未见过的网页任务上实现了25.06%的相对提升，在固有复杂任务上也观察到进一步增益，证明了构建更强大虚拟智能体的可行性。
英文摘要
Achieving virtual agents capable of automating tasks across diverse digital environments remains a pivotal challenge in Embodied AI. While Multimodal Large Language Models (MLLMs) offer enhanced visual perception and reasoning, their agentic deployment faces three challenges: costly data annotation, imprecise action-intent alignment, and inefficient exploration from discarded failed trajectories. To address these, we introduce Iron, an intent-aligned, self-improved, and annotation-efficient framework for training GUI agents. Iron employs a novel dual learning strategy that utilizes a stepwise cycle-consistent (SCC) reward to achieve fine-grained alignment between low-level actions and high-level intents, thereby improving instruction grounding and intent understanding. Concurrently, Iron introduces a hindsight reproduction mechanism to repurpose failed trajectories for training, improving both learning efficiency and task diversity. Extensive experiments demonstrate that Iron-trained generalist agents consistently improve performance on cross-environment and cross-device tasks, outperforming models trained with three times more data. Iron also achieves a substantial 25.06% relative improvement on unseen web tasks, with further gains observed on inherently complex tasks, demonstrating the feasibility of building more capable virtual agents.
发表机构
- Fudan University（复旦大学）
- Zhejiang University（浙江大学）
机构由 AI 辅助整理，请以论文原文为准。
