ResearchDatasetsSWE-GymFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsSoftware Engineering Task ResolutionSWE-Gym (Holdout)19Task Resolution Rate12Failure predictionSWE-Gym (Transfer held-out)60ROC-AUC4Failure predictionSWE-Gym (Within corpus)85.3ROC-AUC4Leave-one-corpus-out transferSWE-Gym59.9AUC3