FutureOmni: Evaluating Future Forecasting from Omni-Modal Context for Multimodal LLMs

Benchmark Model Rank Results
audio-visual-question-answering-on-futureomniOla (7B) +OFF#3Accuracy: 50.19
audio-visual-question-answering-on-omnivideobenchOla (7B) +OFF#6Accuracy: 36.9
audio-visual-question-answering-on-worldsensevideo-SALMONN 2 (7B) +OFF#10Accuracy: 48.77
audio-visual-question-answering-on-dailyomnivideo-SALMONN 2 (7B) +OFF#14Accuracy: 65.8
video-question-answering-on-mlvuvideo-SALMONN 2 (7B) +OFF#43Accuracy: 67.86