WizardCoder: Empowering Code Large Language Models with Evol-Instruct
Open paper
Benchmark
Model
Rank
Results
code-generation-on-codecontests
WizardCoder-15B
#8
Test Set pass@1: 1.11
Test Set pass@5: 3.18
Val Set pass@1: 1.98
…
code-generation-on-mbpp
WizardCoder 15B
#47
Accuracy: 51.8
Rank counts only results with a code link.