Ir directamente a la navegación principal Ir directamente a la búsqueda Ir directamente al contenido principal

Grounded Language Agent for Product Search via IntelligentWeb Interactions

Producción científica: Conference contribution › revisión exhaustiva

4 Citas (Scopus)

Resumen

The development of agents powered by large language models (LLMs) to accomplish complex high-level user intents, has attracted significant attention recently. However, employing LLMs with billions of parameters (e.g., GPT-4) may incur substantial costs on top of handcrafting extensive prompts. To address this, we introduce a Grounded Language Agent for Intelligent Web Interactions, named GLAINTEL. GLAINTEL employs Flan-T5 as its backbone and is flexible in training in various settings: unsupervised learning, supervised learning, and unsupervised domain adaptation. Specifically, we tackle both the challenge of learning without human demonstrations and the opportunity to leverage human demonstrations effectively when those are available. Additionally, we explore unsupervised domain adaptation for cases where demonstrations are limited to a specific domain. Experimental evaluations across diverse setups demonstrate the effectiveness of GLAINTEL in unsupervised settings, outperforming in-context learning-based approaches that employ larger models with up to 540 billion parameters. Surprisingly, behavioral cloning-based methods that straightforwardly use human demonstrations do not outperform unsupervised variants of GLAINTEL. Additionally, we show that combining human demonstrations with reinforcement learningbased training yields results comparable to methods utilizing GPT-4. The code is available at: https://github.com/MultifacetedNLP/Web-Agents-Unsupervised.

Idioma originalEnglish
Título de la publicación alojada1st Workshop on Customizable NLP
Subtítulo de la publicación alojadaProgress and Challenges in Customizing NLP for a Domain, Application, Group, or Individual, CustomNLP4U 2024 - Proceedings of the Workshop
EditoresSachin Kumar, Vidhisha Balachandran, Chan Young Park, Weijia Shi, Shirley Anugrah Hayati, Yulia Tsvetkov, Noah A. Smith, Hannaneh Hajishirzi, Dongyeop Kang, David Jurgens
Páginas63-75
Número de páginas13
ISBN (versión digital)9798891761803
EstadoPublished - 2024
Evento1st Workshop on Customizable NLP: Progress and Challenges in Customizing NLP for a Domain, Application, Group, or Individual, CustomNLP4U 2024 - Miami, United States
Duración: nov 16 2024 → …

Serie de la publicación

Nombre1st Workshop on Customizable NLP: Progress and Challenges in Customizing NLP for a Domain, Application, Group, or Individual, CustomNLP4U 2024 - Proceedings of the Workshop

Conference

Conference1st Workshop on Customizable NLP: Progress and Challenges in Customizing NLP for a Domain, Application, Group, or Individual, CustomNLP4U 2024
País/TerritorioUnited States
CiudadMiami
Período11/16/24 → …

Nota bibliográfica

Publisher Copyright:
©2024 Association for Computational Linguistics.

Financiación

This work is supported in part by the National Science Foundation (NSF) under grant IIS-2401685.

FinanciadoresNúmero del financiador
National Science Foundation Arctic Social Science ProgramIIS-2401685

    ASJC Scopus subject areas

    • Computational Theory and Mathematics
    • Computer Science Applications
    • Software
    • Library and Information Sciences
    • Linguistics and Language

    Huella

    Profundice en los temas de investigación de 'Grounded Language Agent for Product Search via IntelligentWeb Interactions'. En conjunto forman una huella única.

    Citar esto