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

MixEval

Benchmarks

Task NameDataset NameSOTA ResultTrend
Knowledge-focused evaluationMixEval Hard
Accuracy21.8
8
Knowledge-focused evaluationMixEval Standard
Accuracy33
8
AlignmentMixEval
Score86.7
5
AlignmentMixEval v1 (test)
Accuracy76.5
4
Showing 4 of 4 rows