plstcharles-saifh/pyine-v1-augments

该数据集包含1,091,003条由PyINE框架生成的LLM提示结果。每一行都是一个单独的提示/响应对，通过查询LLM来注释或增强Python代码解决方案（例如添加文档提示、注入错误、生成误导性注释）。数据集结构包括不同的提示类型（如代码存根、代码摘要、提示/文档等）和多个列（如记录ID、提示文本、LLM响应文本、创建元数据等）。数据集来源于PyINE-v1跟踪数据集，并使用了OpenAI的gpt-5-mini模型。

This dataset contains 1,091,003 LLM prompting results generated by the PyINE framework. Each row is a single prompt/response pair produced by querying an LLM to annotate or augment Python code solutions (e.g. adding documentation hints, injecting bugs, generating misleading comments). The dataset structure includes different prompt types (e.g., code stubbing, code summary, hints/docs, etc.) and multiple columns (e.g., record ID, prompt text, LLM response text, creation metadata, etc.). The dataset is derived from the PyINE-v1 trace dataset and uses OpenAIs gpt-5-mini model.