RepoST : Scalable Repository-Level Coding Environment Construction with Sandbox Testing

Yiqing Xie, Alex Xie, Divyanshu Sheth, Pengfei Liu, Daniel Fried, Carolyn Rosé · 2025

Read RepoST : Scalable Repository-Level Coding Environment Construction with Sandbox Testing by Yiqing Xie, Alex Xie, Divyanshu Sheth, Pengfei Liu, Daniel Fried, Carolyn Rosé (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-07358.txt.

Description

Yiqing Xie and co-authors published a computing paper in 2025 on building testable coding environments at the level of whole code repositories. Checking generated code needs running feedback, but building whole repositories is hard for humans and models alike. Their method, RepoST, isolates a target function and its needs into a separate script for sandbox testing. That cut in complexity allows scale.

With it they built a training set of thousands of functions from hundreds of repositories. Training on its feedback raised scores on two code benchmarks. They also built an evaluation set and benchmarked a dozen code models. The authors offer both sets as resources for repository-level generation.

Catalog record

CREATORYiqing Xie, Alex Xie, Divyanshu Sheth, Pengfei Liu, Daniel Fried, Carolyn Rosé
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE52 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-07358

Browse the full library · About the Tower of Babel