无需微调，让大模型自动发现网页导航意图并自我探索提升决策能力。
Auto-Intent: Automated Intent Discovery and Self-Exploration for Large Language Model Web Agents
- 通过无监督提取目标领域中的紧凑意图（最多三词）
- 在Mind2Web和WebArena上显著提升GPT和Llama模型的导航性能
- 引入自探索机制，用候选意图提示增强模型决策
本文提出Auto-Intent，一种将预训练大语言模型（LLM）适配到目标领域（聚焦网页导航任务）的方法，无需直接微调。该方法首先无监督地从目标领域示范中提取底层意图，形式高度紧凑（最多三个词）。基于提取的意图，训练意图预测器以根据代理的历史观测和动作预测下一个意图。特别地，我们提出一种自探索方法：将前k个可能的意图预测作为提示提供给预训练的LLM代理，从而增强其决策能力。Auto-Intent在大型真实网站导航基准（Mind2Web）和WebArena在线导航任务上显著提升了GPT-{3.5, 4}及Llama-3.1-{70B, 405B}代理的表现，并实现了从Mind2Web到其他任务的跨基准泛化能力。
原文摘要 · Abstract (English)
In this paper, we introduce Auto-Intent, a method to adapt a pre-trained large language model (LLM) as an agent for a target domain without direct fine-tuning, where we empirically focus on web navigation tasks. Our approach first discovers the underlying intents from target domain demonstrations unsupervisedly, in a highly compact form (up to three words). With the extracted intents, we train our intent predictor to predict the next intent given the agent's past observations and actions. In particular, we propose a self-exploration approach where top-k probable intent predictions are provided as a hint to the pre-trained LLM agent, which leads to enhanced decision-making capabilities. Auto-Intent substantially improves the performance of GPT-{3.5, 4} and Llama-3.1-{70B, 405B} agents on the large-scale real-website navigation benchmarks from Mind2Web and online navigation tasks from WebArena with its cross-benchmark generalization from Mind2Web.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。