Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

QBench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Image Quality AssessmentQBench
Accuracy77.5
81
Multi-image understandingQBench2
Accuracy81.7
30
Multi-image reasoningQBench2 (val)
Accuracy79.3
21
Image Quality AssessmentQBench (test)
Accuracy74.1
17
Visual Quality AssessmentQBench
QBench Score60.6
12
Showing 5 of 5 rows