ResearchBenchmarksOpen-ended Video Question Answering on iVQA (test)Follow60.9AccuracyMoReVQA13.16425.55737.9550.343Apr 9, 2024Evaluation ResultsMethodMethodLinksAccuracyMoReVQAVLM=PALI-3 (5B), LLM=P...VLM=PALI-3 (5B), LLM=PaLM-2, Video context=16 uniformly sampled frames2024.0460.9JCEFVLM=PALI-3 (5B), LLM=P...VLM=PALI-3 (5B), LLM=PaLM-2, Video context=1 frame per second2024.0456.9ViperGPT+VLM=PALI-3 (5B), LLM=P...VLM=PALI-3 (5B), LLM=PaLM-22024.0446.6LLM-onlyLLM=PaLM-2LLM=PaLM-22024.0415