HumanEval and MBPP hardened with far more test cases, exposing solutions that only pass the originals by luck.
Read the original source — github.com
benchmark · Shared by tscosj
0 comments
No comments yet.
EvalPlus (HumanEval+ / MBPP+)
HumanEval and MBPP hardened with far more test cases, exposing solutions that only pass the originals by luck.