SWE-agent takes a GitHub issue and tries to automatically fix it, using GPT-4, or your LM of choice. It solves 12.47% of bugs in the SWE-bench evaluation set and takes just 1 minute to run.
I was just wondering how SWE-agent parses the file tree of github repositories. Does it place emphasis on file names? How does it efficiently parse through files (especially long files)?
I would also appreciate it if you could show me which file contains the code to do said action.
Thanks!
SWE-agent uses relatively simple find and grep-style commands. See our docs for commands. You can also check the trajectories from SWE-agent here to see it in action without running it yourself.
Describe the issue
I was just wondering how SWE-agent parses the file tree of github repositories. Does it place emphasis on file names? How does it efficiently parse through files (especially long files)? I would also appreciate it if you could show me which file contains the code to do said action. Thanks!
Optional: Relevant documentation page
No response