README.md

metadata

language:
  - en
license: cc-by-sa-4.0
tags:
  - python
  - stackoverflow
  - question-answering
  - instruct
size_categories:
  - 1K<n<10K

StackOverflow Python QA Dataset

Description

This dataset contains high-quality Python questions and answers scraped from StackOverflow. It is designed for instruct-tuning Large Language Models (LLMs) or for question-answering tasks.

• Source: StackOverflow
• Selection Criteria:
  • Tag: python
  • Score: > 10
  • Must have an accepted answer
• Content: Pairs of instructions (Question) and outputs (Accepted Answer).

Dataset Structure

The dataset consists of the following fields:

• question_id: The unique identifier for the question on StackOverflow.
• score: The vote score of the question.
• tags: A list of tags associated with the question.
• creation_date: The timestamp of when the question was asked.
• instruction: The full question title and body, cleaned of HTML.
• output: The full body of the accepted answer, cleaned of HTML.

Sample Entry

{
  "question_id": 12345,
  "score": 50,
  "tags": ["python", "list-comprehension"],
  "instruction": "How do I create a list... \n\n I have a loop that...",
  "output": "You can use a list comprehension:\n\n [x for x in iterable]"
}

---

Created by Atif

• GitHub: @meeAtif
• Linkedin: @meAtif
• HuggingFace: @meeAtif
• Email: iatif@proton.me
• DM For: Collaborations, Feedback, or Updates

---

---