Annoy-PyEdu-Rs-Raw
收藏资源简介:
该数据集是一个经过处理的PythonEdu-Rs数据集,源自HuggingFaceTB团队。数据以JSON Lines格式存储,每个样本代表一个Python编程任务,包含问题描述、输入输出要求、参考代码、入口函数名、输入输出对列表、原始来源、推理类型和元信息等字段。它专门设计用于支持代码生成、编程教育和代码理解等任务,特别是针对Python函数的输入输出推理。需注意部分样本的输入输出对可能为空,且问题描述可能因转换而信息不足。数据集采用odc-by许可证。
This dataset is a processed version of the PythonEdu-Rs dataset, with raw data sourced from the HuggingFaceTB team. The data is stored in JSON Lines format, and the primary file is named '0_368500_filtered_v2_ds25.sced.jsonl'. Each sample represents a Python programming task, containing the following fields: `problem_description`: the task's problem description, `io_requirements`: input-output requirements and constraints, `refcode`: reference code that may include package imports, helper functions, and the main entry function, `funcname`: the name of the entry function, `ios`: a list of input-output pairs, each containing input parameters and return values, with some pairs possibly empty, `source`: the original code source, `category`: the assigned inference type, and `meta`: sample metadata. This dataset is applicable to tasks such as code generation, programming education, or code comprehension, particularly for input-output reasoning of Python functions. It should be noted that the list of input-output pairs for some samples is empty, as the input-output size exceeded the constraint limits during code execution; furthermore, conversions based on Large Language Models (LLMs) may be imperfect, resulting in insufficient problem description information for certain samples. This dataset is released under the ODC-By license.
数据集概述
- 数据集名称:Annoy-PyEdu-Rs-Raw
- 来源:改编自 HuggingFaceTB 团队的原始 PythonEdu-Rs 数据集
- 许可协议:odc-by
数据格式
每个样本以 JSON 行格式存储,位于 0_368500_filtered_v2_ds25.sced.jsonl 文件中,包含以下字段:
| 字段 | 说明 |
|---|---|
problem_description |
函数的问题描述 |
io_requirements |
输入/输出要求和约束 |
refcode |
参考代码,包括导入的包(可选)、辅助函数(可选)和主入口函数 |
funcname |
入口函数名称 |
ios |
输入输出示例列表,每个元素包含 input(输入参数)和 output(返回值) |
source |
原始代码文件的来源 |
category |
分配给该样本的推理类型 |
meta |
该样本的元信息 |
注意事项
- 部分
ios字段为空,原因在于执行代码后输入/输出规模过大,超出预设约束,因此未存储或用于后续处理。 - 由于基于 LLM 的转换不完美,部分问题描述可能缺少足够信息来描述代码,开发者将此留作未来工作,以进一步优化数据并发布改进版本。
相关资源链接
- 论文:https://huggingface.co/papers/xxxx.xxxxx
- 项目页面:https://specx.github.io/
- 发布资源集合:https://huggingface.co/collections/sdadafdaf4546/specx-67a978e28fd926b56a4f55a2
- 代码仓库:https://github.com/ShieldAssistant/Annoy




