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

SciQA

Benchmarks

Task NameDataset NameSOTA ResultTrend
Visual Question AnsweringSciQA IMG
Accuracy87.1
71
Uncertainty EstimationSciQA
AUROC0.8269
56
Science Question AnsweringSciQA-IMG
SciQA-IMG Accuracy89
53
Scientific Question AnsweringSciQA
Accuracy91.4
35
Question AnsweringSciQA (test)
Accuracy80.6
30
Step-level correctness assessmentSciQA (test)
PR-AUC55.3
22
Step-level reasoning verificationSciQA
PR-AUC44
19
Multimodal ReasoningSciQA
Accuracy93.7
14
Question AnsweringSciQA
Normalized Accuracy76.2
10
Clarifying QuestionsSciQA (test)
Accuracy26
6
Showing 10 of 10 rows