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

CREAK

Benchmarks

Task NameDataset NameSOTA ResultTrend
Fact VerificationCreak WikiData (test)
Hits@197.45
17
Knowledge Base Question AnsweringCreak WikiData (test)
Hits@191.82
14
False Premise DetectionCREAK
TP (%)92.8
12
Fact VerificationCreak
Accuracy0.956
8
Commonsense Fact VerificationCREAK (test)
Accuracy88.6
5
Commonsense Fact VerificationCREAK Contrast Set (contra)
Accuracy92.2
4
Showing 6 of 6 rows