arxiv:2407.16741

OpenDevin: An Open Platform for AI Software Developers as Generalist Agents

Published on Jul 23

· Submitted by

akhaliq on Jul 25

#1 Paper of the day

Upvote

Authors:

Xingyao Wang ,

Boxuan Li ,

Yufan Song ,

Frank F. Xu ,

Xiangru Tang ,

Mingchen Zhuge ,

Jiayi Pan ,

Yueqi Song ,

Bowen Li ,

Jaskirat Singh ,

Hoang H. Tran ,

Bill Qian ,

Yanjun Shao ,

Niklas Muennighoff ,

Binyuan Hui ,

Junyang Lin ,

Hao Peng

Abstract

Software is one of the most powerful tools that we humans have at our disposal; it allows a skilled programmer to interact with the world in complex and profound ways. At the same time, thanks to improvements in large language models (LLMs), there has also been a rapid development in AI agents that interact with and affect change in their surrounding environments. In this paper, we introduce OpenDevin, a platform for the development of powerful and flexible AI agents that interact with the world in similar ways to those of a human developer: by writing code, interacting with a command line, and browsing the web. We describe how the platform allows for the implementation of new agents, safe interaction with sandboxed environments for code execution, coordination between multiple agents, and incorporation of evaluation benchmarks. Based on our currently incorporated benchmarks, we perform an evaluation of agents over 15 challenging tasks, including software engineering (e.g., SWE-Bench) and web browsing (e.g., WebArena), among others. Released under the permissive MIT license, OpenDevin is a community project spanning academia and industry with more than 1.3K contributions from over 160 contributors and will improve going forward.

View arXiv page View PDF Add to collection

Community

akhaliq

Paper submitter Jul 25

https://github.com/OpenDevin/OpenDevin

Taylor658

Jul 25

Thanks for taking time to publish a paper on OpenDevin.

What improvements are planned for OpenDevin's evaluation framework, particularly when it comes to extending benchmarks in future versions?

gneubig

Paper author Jul 25

We're always happy to add more evaluation benchmarks! If there are any that you'd like to see, please look at our evaluation harness https://github.com/OpenDevin/OpenDevin/tree/main/evaluation

and come connect with us on Slack, github, etc: https://github.com/OpenDevin/OpenDevin/?tab=readme-ov-file#-join-our-community