HumanEval
34 results · 11 sources
13,812 benchmarks in 32 GPU combinations documented
Evaluation scores organized by test family. Versions, test sets and harnesses remain separate.
34 results · 11 sources
37 results · 8 sources
412 results · 6 sources
31 results · 5 sources
15 results · 5 sources
5 results · 5 sources
45 results · 4 sources
14 results · 4 sources
3 results · 3 sources
3 results · 3 sources
24 results · 2 sources
24 results · 2 sources
5 results · 2 sources
Pages require at least three documented tests from two sources. Every collected result remains available in the explorer.