AISE-TUDelft/HumanEvalNext

HumanEvalNext是一个基于HumanEval的改进代码生成基准，通过EvalPro框架对质量进行了严格的提升，并经过了同行评审。它包含了任务ID、提示、入口点、标准解决方案和测试等字段，并提供了训练数据集，共有82个示例。HumanEvalNext is an improved code generation benchmark based on HumanEval, enhanced by the EvalPro framework through a rigorous quality improvement process and peer review. It includes fields such as task ID, prompt, entry point, canonical solution, and test, and provides a training dataset with a total of 82 examples.