WizardCoder: Empowering Code Large Language Models with Evol-Instruct

Code Large Language Models (Code LLMs), such as StarCoder, have demonstrated exceptional performance in code-related tasks. However, most existing models are solely pre-trained on extensive raw code data without instruction fine-tuning. In this paper, we introduce WizardCoder, which empowers Code LLMs with complex instruction fine-tuning, by adapting the Evol-Instruct method to the domain of code. Through comprehensive experiments on four prominent code generation benchmarks, namely HumanEval, HumanEval+, MBPP, and DS-1000, we unveil the exceptional capabilities of our model. It surpasses all other open-source Code LLMs by a substantial margin. Moreover, our model even outperforms the largest closed LLMs, Anthropic's Claude and Google's Bard, on HumanEval and HumanEval+. Our code, model weights, and data are public at https://github.com/nlpxucan/WizardLM

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Benchmark
Code Generation CodeContests WizardCoder-15B Test Set pass@1 1.11 # 4
Test Set pass@5 3.18 # 4
Val Set pass@1 1.98 # 4
Val Set pass@5 3.27 # 3
Code Generation HumanEval WizardCoder 15B Pass@1 57.30 # 41
Code Generation MBPP WizardCoder 15B Accuracy 51.8 # 49

Methods


No methods listed for this paper. Add relevant methods here