Q-Bench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Q-Bench
87.23Attack Success Rate
37
Q-Bench LLVisionQA 1.0 (dev)
74.5Overall Score
29
Q-Bench-Video (dev)
76.78Yes-or-No Acc
14
Q-Bench subset (dev)
85.82Yes/No Accuracy
14
Q-Bench (val)
80.7Score
12
Q-Bench (test)
63.6Overall Score
11
Q-Bench2
74.2Overall Score
9
Q-Bench (test)
62.9Overall Score
8
Q-Bench-Video
62.2Score
7
Q-Bench
74.4Accuracy
2