Open caohch-1 opened 3 months ago
Hi @caohch-1 the preprocess_data.py
and postprocess_data.py
are not specific for swe bench data. In there we provide some utility functions to support various repair and localization procedures that we do (they are geared for python projects)
if you want to look at how the preprocessing is done, we mainly just turn the original repo into a structure format containing files, classes, and functions for ease of operating (see get_repo_structure.py
) for the processing.
The main scripts of repair.py
and localize.py
are definitely geared towards swe-bench-lite, since we directly operate on individual issues.
Hello, agentless is a very impressive work.
I want to apply agentless on some issues outside of the swe dataset.
I see that there are
preprocess_data.py
andpostprocess_data.py
that seem to contain preprocessing of the swe dataset, but there is no main entry in the scripts.Can you provide the complete preprocessing script or how to use the functions in the existing scripts (
preprocess_data.py
andpostprocess_data.py
) to do the processing, which would be helpful in applying agentless to more issues.