SynthAgent: Adapting Web Agents with Synthetic Supervision

TL;DR

SynthAgent introduces a dual refinement framework for synthetic supervision, significantly improving web agent adaptation by generating high-quality synthetic data through task and trajectory refinement.

Contribution

It presents a novel synthetic supervision method that refines tasks and trajectories to enhance web agent adaptation in new environments.

Findings

SynthAgent outperforms existing synthetic data methods.

Refined synthetic data improves web agent performance.

Dual refinement reduces hallucinations and noise in data.

Abstract

Web agents struggle to adapt to new websites due to the scarcity of environment specific tasks and demonstrations. Recent works have explored synthetic data generation to address this challenge, however, they suffer from data quality issues where synthesized tasks contain hallucinations that cannot be executed, and collected trajectories are noisy with redundant or misaligned actions. In this paper, we propose SynthAgent, a fully synthetic supervision framework that aims at improving synthetic data quality via dual refinement of both tasks and trajectories. Our approach begins by synthesizing diverse tasks through categorized exploration of web elements, ensuring efficient coverage of the target environment. During trajectory collection, tasks are refined only when conflicts with observations are detected, which mitigates hallucinations while preserving task consistency. After collection, we conduct trajectory refinement with global context to mitigate potential noise or misalignments. Finally, we fine-tune open-source web agents on the refined synthetic data to adapt them to the target environment. Experimental results demonstrate that SynthAgent outperforms existing synthetic data methods, validating the importance of high-quality synthetic supervision. The code is publicly available at https://github.com/aiming-lab/SynthAgent.