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

HumanEval+

Benchmarks

Task NameDataset NameSOTA ResultTrend
Code GenerationHumanEval+ (test)
Pass@198.1
132
Code GenerationHumanEval+ v1 (test)
Pass Rate87.8
55
Code GenerationHumanEval+ Out-of-Domain (test)
Accuracy81.71
18
Code ReasoningHumanEval+
Pass@1697
15
Code GenerationHumanEval+
Score34.76
11
Unit test generationHumanEval+ (test)
Error Rate1.27
7
Code GenerationHumanEval+ ko
Score92.1
3
Showing 7 of 7 rows