Auto-Intent: Automated Intent Discovery and Self-Exploration for Large Language Model Web Agents
Jaekyeom Kim, Dong Ki Kim, Lajanugen Logeswaran, Sungryull Sohn, Honglak Lee · 2024
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 decisionmaking capabilities.Auto-Intent substantially improves the performance of GPT-{3.5, 4} and Llama-3.1-{70B,405B} agents on the largescale real-website navigation benchmarks from Mind2Web and online navigation tasks from WebArena with its cross-benchmark generalization from Mind2Web.Decision-Making with Self-Exploration Task: Find a permanent job in Logistics within 20 miles of New York, zip 11005, in the middle-income … CLICK L. Salary Candidate elements: A. B. … … L. Salary …