E-VQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
E-VQA (test)
84.03Accuracy
85
E-VQA Lifelong Sequential
96.67Rel. Score
72
E-VQA
100Reliability
68
E-VQA Single-Hop
59.9Accuracy
52
E-VQA (test)
52.6Accuracy (All)
49
E-VQA
55.9Accuracy (Single-Hop)
34
E-VQA MMEdit 1.0 (test)
100Reliability
24
E-VQA All
39.5Accuracy
23
E-VQA
12.5Final Fidelity Rate
18
E-VQA
70.27Single-Hop Accuracy
16
E-VQA All
45.8Accuracy
15
E-VQA
53.7Accuracy
15
E-VQA
43.6Accuracy
12
E-VQA (test)
93.4MCC
12
E-VQA 5
98.12Reliability Score
11
E-VQA
44.9Recall@1
11
E-VQA Lifelong Editing 5
96.85Relational Score
10
E-VQA
100Reliability
10
E-VQA
90CondR@1
10
E-VQA
44.7Recall@1
10
E-VQA M2KR setup (test)
70.3BEM
8
E-VQA standard (test)
44.1R@1
8
E-VQA (test)
0.428Recall@1
7
E-VQA (Overall)
47.4Top-1 Accuracy
6
E-VQA 1.0 (test)
0.449MRR@5
6