Qodo/deep_code_bench

Deep Code Bench数据集包含关于代码的问答对，旨在评估深度学习模型对代码的理解能力。每个示例包括问题、预期答案、支持答案的事实列表以及有关难度、范围等额外信息的元数据。

The Deep Code Bench dataset contains question-answer pairs concerning code, designed to evaluate the code understanding capability of deep learning models. Each example includes a question, an expected answer, a list of factual statements supporting the answer, and metadata with additional information about difficulty, scope, etc.