This dataset collates three class-balanced code classification datasets in the wild, where splits have been stratified by label. The sources are:
CodeXGLUE defect detection (sourced from: semeru/code-code-DefectDetection)
BigCloneBench clone detection (sourced from: nchen909/bigclonebench-processed)
CodeComplex code runtime complexity prediction (sourced from: codeparrot/codecomplex)