HallusionBench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
HallusionBench 1.0 (test)
22.2fACC
10
HallusionBench
73Accuracy
10
HallusionBench visual questions
65.8Accuracy
10
HallusionBench
37.3fAcc
7
HallusionBench
54.5Avg@8
7
HallusionBench avg
67Score
7
HallusionBench (test)
53.31Simple Accuracy
7
HallusionBench
72.9Answer Accuracy (aAcc)
6
HallusionBench
60.5General Score
6
HallusionBench Image
64.1Average Score
5
HallusionBench
76Accuracy
5
HallusionBench
18.7fACC
4
HallusionBench (test)
17.8Question Pair Accuracy
4
HallusionBench HBI (all)
45.21Score
4
HallusionBench n=920 (held-out)
0.897Per-Class Loss
3
HallusionBench
52.89Simple Accuracy
2