Annoy-PyEdu-Rs
收藏资源简介:
Annoy-PythonEdu-Rs是Annoy项目中的一个数据集子集,专注于Python教育领域的代码执行轨迹生成。该数据集采用完全基于大语言模型的方法,利用DeepSeek-V2.5合成高质量的响应,以克服传统方法中无法获取确定性逆函数和自动构建轨迹模板化的局限性。数据集旨在为代码执行推理提供自由形式的自然语言推理轨迹,增强模型的表达能力和泛化性能。该数据集是完整数据集的一个子集,以ODC-By v1.0许可证发布。
Annoy-PythonEdu-Rs is a subset of the Annoy project, focusing on generating code execution traces in the Python education domain. This dataset adopts a fully large language model-based approach, using DeepSeek-V2.5 to synthesize high-quality responses to overcome the limitations of traditional methods, such as the inability to obtain deterministic inverse functions and the automation of trace templating. The dataset aims to provide free-form natural language reasoning traces for code execution reasoning, enhancing the models expressiveness and generalization performance. It is a subset of the complete dataset and is released under the ODC-By v1.0 license.




