A-OKVQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
A-OKVQA
92.68Acc
240
A-OKVQA (test)
90.56Accuracy
103
A-OKVQA (val)
79.5Accuracy
92
A-OKVQA POPE (Popular)
90.3Accuracy
91
A-OKVQA POPE (Random)
92.1Accuracy
75
A-OKVQA POPE SEEM-annotated
91.6Accuracy
60
A-OKVQA POPE (Adversarial)
0.836Accuracy
57
A-OKVQA
82.71Accuracy
49
A-OKVQA 2022 (test)
100Accuracy
48
A-OKVQA
5.4ECE
32
A-OKVQA POPE (test)
90.13Accuracy (Random)
29
A-OKVQA (test)
87.2Accuracy
28
A-OKVQA (Adversarial split)
79.1Accuracy
27
A-OKVQA (test)
69Accuracy
22
A-OKVQA POPE
92.37Random Accuracy
21
A-OKVQA
68.7Accuracy
18
A-OKVQA POPE Evaluation (Adversarial)
82Accuracy
16
A-OKVQA POPE (Popular)
89.77Accuracy
16
A-OKVQA POPE Evaluation (Random)
90.03Accuracy
16
A-OKVQA
93.76Accuracy (Random)
15
A-OKVQA Open-Ended
72.14Accuracy
15
A-OKVQA v1.0 (test)
53.36Accuracy
14
A-OKVQA (Random split)
90.83Accuracy
12
A-OKVQA 1.0 (test)
68Accuracy
12
A-OKVQA POPE (Adversarial)
81.94Accuracy
12