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

GQA

Benchmarks

Task NameDataset NameSOTA ResultTrend
Visual Question AnsweringGQA
Accuracy83.2
1,445
Visual Question AnsweringGQA
Accuracy74.9
524
Visual Question AnsweringGQA (test-dev)
Accuracy73.9
236
Visual Question AnsweringGQA
Accuracy77.5
218
Visual Question AnsweringGQA (test)
Accuracy89.3
204
Visual Question AnsweringGQA
Mean Accuracy65.8
196
Visual Question AnsweringGQA
Score71.3
193
Performance EstimationGQA
MAE0
184
Visual Question AnsweringGQA
GQA Score64.83
152
Visual ReasoningGQA
Accuracy73.01
96
Visual Question AnsweringGQA
GQA Score63.4
75
Visual Question AnsweringGQA (test-std)
Accuracy65.65
74
Visual Question AnsweringGQA
Accuracy65.34
59
Object Hallucination ProbingGQA Adversarial
Accuracy83.27
55
Multi-modal Vision-Language UnderstandingGQA
Accuracy64.2
51
Object Hallucination ProbingGQA POPE Popular
Accuracy86.07
49
Object Hallucination EvaluationGQA (Random)
Accuracy90.13
43
Object Hallucination ProbingGQA POPE Random
Accuracy (GQA POPE)89.93
42
Visual Question AnsweringGQA
Accuracy75.77
36
General Visual Question AnsweringGQA
Accuracy61.94
35
Multi-turn Visual Question AnsweringMT-GQA
Acc165.45
33
Visual Question AnsweringGQA
Exact Match63.3
32
Visual Question AnsweringGQA balanced (test-dev)
Accuracy77.4
32
Visual Question AnsweringGQA (val)
Accuracy83.39
32
Visual Question AnsweringGQA
Accuracy61.97
31
Showing 25 of 148 rows