nus-apr / auto-code-rover

A project structure aware autonomous software engineer aiming for autonomous program improvement. Resolved 30.67% tasks (pass@1) in SWE-bench lite with each task costs less than $0.7.
Other
2.37k stars 236 forks source link

Check that task ids have valid setups before running them. #24

Closed jimwhite closed 2 months ago

jimwhite commented 2 months ago

The environment setup process is complex and problems like unsupported dependency versions for some project versions in some cases can be difficult to resolve. Therefore we should be tolerant when tasks can't be run.

This change checks that there is a task setup and mapping for each task id before trying to run them. If there isn't then it is skipped with a log message about that for the task id.