ResearchDatasetsEval-OODEval-OODFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsVideo-LLM performance calibrationEval-OOD (EgoSchema & LongVideoBench)5.8ECE45Video Question AnsweringEval-OOD (test)63Accuracy14