Unable to Reproduce the Reported 80.0% Performance — Any Updates?

[M-Myron]

Hi authors,

First of all, thanks for the exciting work and the impressive result reported (80.0% on SWE-Bench). I’m opening this issue because reproduction is still not possible based on the current repo.

I’m opening this issue because it’s now Nov 30, and I’m still unable to reproduce the reported performance using the current version of the repo. In the previous discussion (Issue #9), you mentioned that the full steps and scripts would be shared after internal review, and that a leaderboard submission was planned. I just wanted to check in on the status of those.

Would it be possible to share an update on:
• The exact steps or scripts you used to get the 80% result
• The specific configs
• Whether the SWE-Bench leaderboard submission has been made yet

If there are internal changes that haven’t been pushed to the repo yet, it would also be helpful to know.

Really appreciate your work and your time — just hoping to reproduce the results accurately. Thanks!