FlexiVideo: Variation-Aware Temporal Dynamics Modeling for Efficient Video Understanding

Benchmark Model Rank Results
video-question-answering-on-lvbenchFlexiVideo-3B (extra data)#19Accuracy: 57.3
video-question-answering-on-video-mme-with-subtitlesFlexiVideo-3B (extra data)#29Accuracy: 67.7
video-question-answering-on-mlvuFlexiVideo-3B (extra data)#37Accuracy: 69.6
zero-shot-video-question-answer-on-video-mme-w-o-subsFlexiVideo-3B (extra data)#41Accuracy (%): 62.5