Training-free Uncertainty Guidance for Complex Visual Tasks with MLLMs
Open paper
Benchmark
Model
Rank
Results
video-question-answering-on-mlvu
UG-Sample (InternVL2.5-8B)
#44
Accuracy: 67.6
moment-retrieval-on-charades-sta
UG-Ground (InternVideo2.5-8B)
#55
R@1 IoU=0.5: 52.5
R@1 IoU=0.7: 32.6
R@1 IoU=0.3: 74.7
mIoU: 51.0
Rank counts only results with a code link.