A-OKVQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
A-OKVQA POPE Popular
0.881Accuracy
12
A-OKVQA POPE Random
89.6Accuracy
12
A-OKVQA Speech-converted
0.2Accuracy
12
A-OKVQA 1.0 (val)
0.683Accuracy
11
A-OKVQA
74.93Accuracy
9
A-OKVQA 1.0 (test)
86.7Accuracy
9
A-OKVQA 1.0 (val)
87.7Accuracy
9
A-OKVQA
73Detection Accuracy (DACC)
8
A-OKVQA
73.48Accuracy
8
A-OKVQA (val)
68.9Accuracy
8
A-OKVQA (test)
59.5Accuracy
7
A-OKVQA 60 (test)
0.972AUROC
6
A-OKVQA Adversarial
80.2Accuracy
6
A-OKVQA (Random)
89.57Accuracy
6
A-OKVQA (val)
65.3DA
6
A-OKVQA (test)
7.83Faithfulness Score
6
A-OKVQA
83Accuracy
4
A-OKVQA n=1145 (held-out)
0.381Per-Class Loss
3
A-OKVQA (val)
91.8Score
3
A-OKVQA MC (val)
88.2Accuracy
2
A-OKVQA DA (val)
70.2Accuracy
2