Flow2Code

该数据集名为Flow2Code，包含了三种类型的流程图与代码段的配对，这三种类型分别是代码流程图、UML流程图和伪代码流程图，覆盖了15种编程语言。该数据集由四个关键数据集（HumanEval-X、MBXP、MCEval和ClassEval）构建而成，涵盖了多样化的编程语言和任务复杂度。规模上，它包含了5,622个代码段，与之对应的流程图有16,866张。该数据集的任务是进行基于流程图的代码生成评估。

The dataset named Flow2Code encompasses three types of paired flowcharts and code snippets, specifically code flowcharts, UML flowcharts, and pseudocode flowcharts, covering 15 programming languages. It is constructed from four core datasets: HumanEval-X, MBXP, MCEval, and ClassEval, which cover diverse programming languages and task complexities. In terms of scale, it contains 5,622 code snippets with 16,866 corresponding flowcharts. The core task of this dataset is flowchart-based code generation evaluation.