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

PathVQA

Benchmarks

Task NameDataset NameSOTA ResultTrend
Medical Visual Question AnsweringPathVQA
Overall Accuracy72.61
92
Medical Visual Question AnsweringPathVQA
Accuracy76.8
80
Medical Visual Question AnsweringPathVQA (test)
Accuracy78.2
55
Medical Visual Question AnsweringPathVQA closed-end
Accuracy93.63
35
Vision-Language Medical ReasoningPathVQA
Token Cost (tokens/question)0.7
29
Visual Question AnsweringPathVQA open-ended
Exact Match (EM)4.88
25
Visual Question Answering (Closed-ended)PathVQA closed-ended
Accuracy95
23
Medical Visual Question AnsweringPathVQA Open
Accuracy38.65
22
Hallucination detectionPathVQA
AUC82
20
Visual Question AnsweringPathVQA
Accuracy (Closed)92.9
19
Visual Question AnsweringPathVQA (test)
Overall Accuracy92.7
19
Visual Question AnsweringPathVQA (Open)
Token Recall39
15
Medical Visual Question Answering (Free-text)PathVQA OOD
Accuracy62.3
12
Multi-modal Question AnsweringPathVQA
Accuracy65.9
12
Visual Question AnsweringPathVQA Closed
Token Recall0.93
7
Visual Question AnsweringPathVQA
Accuracy74.4
6
Medical Visual Question AnsweringPathVQA (held-out)
Accuracy59.9
6
Visual Question AnsweringPathVQA
BLEU-163.93
5
Out-of-distribution DetectionPathVQA (PVQA) (test)
FPR6.24
5
Showing 19 of 19 rows