Open sourced predictions, execution logs, trajectories, and results from model inference + evaluation runs on the SWE-bench task.
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).