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

MISBELIEF

Benchmarks

Task NameDataset NameSOTA ResultTrend
Belief assessment on misinformationMISBELIEF Sports domain (Hard split)
Belief Score4.28
28
Belief assessment on misinformationMISBELIEF Sports domain Medium
Belief Score1.79
28
Belief assessment on misinformationMISBELIEF Sports domain Easy split
Belief Score3.32
28
Belief assessment on misinformationMISBELIEF Health domain Hard
Belief Score3.63
28
Belief assessment on misinformationMISBELIEF Health domain (Medium)
Belief Score3.48
28
Belief assessment on misinformationMISBELIEF Health domain Easy split
Belief Score2.87
28
Belief assessment on misinformationMISBELIEF Overall (Hard)
Belief Score4.07
28
Belief assessment on misinformationMISBELIEF Overall (Medium)
Belief Score3.88
28
Belief assessment on misinformationMISBELIEF Overall Easy
Belief Score3.18
28
Misinformation Belief EvaluationMISBELIEF hard misinformation with third round evidence
Performance4.05
5
Showing 10 of 10 rows