SWE-CI: Evaluating Agent Capabilities in Maintaining Codebases via Continuous Integration
Paper • 2603.03823 • Published • 7
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
This repository contains the official experimental trajectory data for SWE-CI: Evaluating Agent Capabilities in Maintaining Codebases via Continuous Integration (paper).
Since running the full SWE-CI benchmark is quite expensive, we publish all raw experiment trajectories (lightly sanitized only for desensitization) to facilitate comparison and further research.
The default settings used in the paper are SWE-CI-v1, iFlow CLI, and TDD (Test-Driven Development) mode. We will continue to update the results with more settings in the future.