One STEP at a time: Language Agents are Stepwise Planners

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Nguyen, Minh, Shareghi, Ehsan
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917835941347328
author Nguyen, Minh
Shareghi, Ehsan
author_facet Nguyen, Minh
Shareghi, Ehsan
contents Language agents have shown promising adaptability in dynamic environments to perform complex tasks. However, despite the versatile knowledge embedded in large language models, these agents still fall short when it comes to tasks that require planning. We introduce STEP, a novel framework designed to efficiently learn from previous experiences to enhance the planning capabilities of language agents in future steps. Concretely, STEP functions through four interconnected components. First, the Planner takes on the task, breaks it down into subtasks and provides relevant insights. Then the Executor generates action candidates, while the Evaluator ensures the actions align with learned rules from previous experiences. Lastly, Memory stores experiences to inform future decisions. In the ScienceWorld benchmark, our results show that STEP consistently outperforms state-of-the-art models, achieving an overall score of 67.4 and successfully completing 12 out of 18 tasks. These findings highlight STEP's potential as a framework for enhancing planning capabilities in language agents, paving the way for more sophisticated task-solving in dynamic environments.
format Preprint
id arxiv_https___arxiv_org_abs_2411_08432
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle One STEP at a time: Language Agents are Stepwise Planners
Nguyen, Minh
Shareghi, Ehsan
Computation and Language
Artificial Intelligence
Machine Learning
Language agents have shown promising adaptability in dynamic environments to perform complex tasks. However, despite the versatile knowledge embedded in large language models, these agents still fall short when it comes to tasks that require planning. We introduce STEP, a novel framework designed to efficiently learn from previous experiences to enhance the planning capabilities of language agents in future steps. Concretely, STEP functions through four interconnected components. First, the Planner takes on the task, breaks it down into subtasks and provides relevant insights. Then the Executor generates action candidates, while the Evaluator ensures the actions align with learned rules from previous experiences. Lastly, Memory stores experiences to inform future decisions. In the ScienceWorld benchmark, our results show that STEP consistently outperforms state-of-the-art models, achieving an overall score of 67.4 and successfully completing 12 out of 18 tasks. These findings highlight STEP's potential as a framework for enhancing planning capabilities in language agents, paving the way for more sophisticated task-solving in dynamic environments.
title One STEP at a time: Language Agents are Stepwise Planners
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2411.08432