AndroidLab: Training and Systematic Benchmarking of Android Autonomous Agents

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xu, Yifan, Liu, Xiao, Sun, Xueqiao, Cheng, Siyi, Yu, Hao, Lai, Hanyu, Zhang, Shudan, Zhang, Dan, Tang, Jie, Dong, Yuxiao
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929573817483264
author Xu, Yifan
Liu, Xiao
Sun, Xueqiao
Cheng, Siyi
Yu, Hao
Lai, Hanyu
Zhang, Shudan
Zhang, Dan
Tang, Jie
Dong, Yuxiao
author_facet Xu, Yifan
Liu, Xiao
Sun, Xueqiao
Cheng, Siyi
Yu, Hao
Lai, Hanyu
Zhang, Shudan
Zhang, Dan
Tang, Jie
Dong, Yuxiao
contents Autonomous agents have become increasingly important for interacting with the real world. Android agents, in particular, have been recently a frequently-mentioned interaction method. However, existing studies for training and evaluating Android agents lack systematic research on both open-source and closed-source models. In this work, we propose AndroidLab as a systematic Android agent framework. It includes an operation environment with different modalities, action space, and a reproducible benchmark. It supports both large language models (LLMs) and multimodal models (LMMs) in the same action space. AndroidLab benchmark includes predefined Android virtual devices and 138 tasks across nine apps built on these devices. By using the AndroidLab environment, we develop an Android Instruction dataset and train six open-source LLMs and LMMs, lifting the average success rates from 4.59% to 21.50% for LLMs and from 1.93% to 13.28% for LMMs. AndroidLab is open-sourced and publicly available at https://github.com/THUDM/Android-Lab.
format Preprint
id arxiv_https___arxiv_org_abs_2410_24024
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle AndroidLab: Training and Systematic Benchmarking of Android Autonomous Agents
Xu, Yifan
Liu, Xiao
Sun, Xueqiao
Cheng, Siyi
Yu, Hao
Lai, Hanyu
Zhang, Shudan
Zhang, Dan
Tang, Jie
Dong, Yuxiao
Artificial Intelligence
Autonomous agents have become increasingly important for interacting with the real world. Android agents, in particular, have been recently a frequently-mentioned interaction method. However, existing studies for training and evaluating Android agents lack systematic research on both open-source and closed-source models. In this work, we propose AndroidLab as a systematic Android agent framework. It includes an operation environment with different modalities, action space, and a reproducible benchmark. It supports both large language models (LLMs) and multimodal models (LMMs) in the same action space. AndroidLab benchmark includes predefined Android virtual devices and 138 tasks across nine apps built on these devices. By using the AndroidLab environment, we develop an Android Instruction dataset and train six open-source LLMs and LMMs, lifting the average success rates from 4.59% to 21.50% for LLMs and from 1.93% to 13.28% for LMMs. AndroidLab is open-sourced and publicly available at https://github.com/THUDM/Android-Lab.
title AndroidLab: Training and Systematic Benchmarking of Android Autonomous Agents
topic Artificial Intelligence
url https://arxiv.org/abs/2410.24024