Saved in:
Bibliographic Details
Main Authors: Pan, Jiayi, Zhang, Yichi, Tomlin, Nicholas, Zhou, Yifei, Levine, Sergey, Suhr, Alane
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2404.06474
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909337996230656
author Pan, Jiayi
Zhang, Yichi
Tomlin, Nicholas
Zhou, Yifei
Levine, Sergey
Suhr, Alane
author_facet Pan, Jiayi
Zhang, Yichi
Tomlin, Nicholas
Zhou, Yifei
Levine, Sergey
Suhr, Alane
contents We show that domain-general automatic evaluators can significantly improve the performance of agents for web navigation and device control. We experiment with multiple evaluation models that trade off between inference cost, modularity of design, and accuracy. We validate the performance of these models in several popular benchmarks for digital agents, finding between 74.4 and 92.9% agreement with oracle evaluation metrics. Finally, we use these evaluators to improve the performance of existing agents via fine-tuning and inference-time guidance. Without any additional supervision, we improve state-of-the-art performance by 29% on the popular benchmark WebArena, and achieve around 75% relative improvement in device control settings.
format Preprint
id arxiv_https___arxiv_org_abs_2404_06474
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Autonomous Evaluation and Refinement of Digital Agents
Pan, Jiayi
Zhang, Yichi
Tomlin, Nicholas
Zhou, Yifei
Levine, Sergey
Suhr, Alane
Artificial Intelligence
We show that domain-general automatic evaluators can significantly improve the performance of agents for web navigation and device control. We experiment with multiple evaluation models that trade off between inference cost, modularity of design, and accuracy. We validate the performance of these models in several popular benchmarks for digital agents, finding between 74.4 and 92.9% agreement with oracle evaluation metrics. Finally, we use these evaluators to improve the performance of existing agents via fine-tuning and inference-time guidance. Without any additional supervision, we improve state-of-the-art performance by 29% on the popular benchmark WebArena, and achieve around 75% relative improvement in device control settings.
title Autonomous Evaluation and Refinement of Digital Agents
topic Artificial Intelligence
url https://arxiv.org/abs/2404.06474