debugbench_pnyx
收藏资源简介:
PNYX-DebugBench 数据集基于 Rtian/DebugBench 数据集,包含原始数据集的所有字段,并新增了 initialization_code、test_code 和 stderr 字段。数据集按语言配置(如 python3)划分,每个语言配置下又按难度级别(easy、medium、hard)进行分割。该数据集适用于文本生成和问答任务,主要用于代码调试和测试。数据集中的示例经过测试,部分原始示例因无法转换为所需格式或存在错误而被排除。数据集的使用示例展示了如何通过 hf_evaluate 进行测试。
The PNYX-DebugBench dataset is built upon the Rtian/DebugBench dataset, preserving all fields from the original dataset while introducing three new fields: initialization_code, test_code, and stderr. The dataset is partitioned by language configurations (e.g., python3), and within each language configuration, it is further split by difficulty levels: easy, medium, and hard. This dataset supports text generation and question answering tasks, and is primarily intended for code debugging and testing applications. All instances within the dataset have been validated; some original instances were excluded as they could not be converted to the required format or contained errors. Usage examples of the dataset demonstrate how to conduct evaluations using hf_evaluate.



