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

VisualProbe

Benchmarks

Task NameDataset NameSOTA ResultTrend
Visual ReasoningVisualProbe Hard
Accuracy0.434
18
Visual ReasoningVisualProbe Medium
Accuracy40.6
18
Visual ReasoningVisualProbe Easy
Accuracy65.4
18
Visual ReasoningVisualProbe
Score42.7
17
Visual ReasoningVisualProbe (VP) cross-domain (test)
Accuracy0.4357
15
Visual PerceptionVisualProbe
Score53.9
10
Visual Question AnsweringVisualProbe (Overall)
Accuracy49.7
9
Visual Question AnsweringVisualProbe Hard
Accuracy47.2
9
Visual Question AnsweringVisualProbe (Medium)
Accuracy41.4
9
Visual Question AnsweringVisualProbe Easy
Accuracy67.4
9
Visual ReasoningVisualProbe (test)
Accuracy (Hard)47.2
7
Showing 11 of 11 rows