ScribeAgent: Towards Specialized Web Agents Using Production-Scale Workflow Data
This addresses the need for more effective web agents in specialized domains, though it is incremental as it builds on existing fine-tuning methods with new data.
The paper tackled the problem of LLM agents struggling with specialized web contexts and long-horizon planning by fine-tuning open-source LLMs on production-scale workflow data, achieving state-of-the-art performance on Mind2Web and a 7.3% improvement in task success rate on WebArena.
Large Language Model (LLM) agents are rapidly improving to handle increasingly complex web-based tasks. Most of these agents rely on general-purpose, proprietary models like GPT-4 and focus on designing better prompts to improve their planning abilities. However, general-purpose LLMs are not specifically trained to understand specialized web contexts such as HTML, and they often struggle with long-horizon planning. We explore an alternative approach that fine-tunes open-source LLMs using production-scale workflow data collected from over 250 domains corresponding to 6 billion tokens. This simple yet effective approach shows substantial gains over prompting-based agents on existing benchmarks -- ScribeAgent achieves state-of-the-art direct generation performance on Mind2Web and improves the task success rate by 7.3% over the previous best text-only web agents on WebArena. We further perform detailed ablation studies on various fine-tuning design choices and provide insights into LLM selection, training recipes, context window optimization, and effect of dataset sizes.