STEM & Non-STEM Q&A Datasets for LLM Training
Collection
9 items • Updated
Sno int64 | Question string | Options string | Answer string | Explanation string | Subject string |
|---|---|---|---|---|---|
1 |
English Non-STEM Question & Answer Dataset
Overview
The English Non-STEM Question & Answer Dataset is a comprehensive educational dataset containing over 2.5 million English question-answer pairs across humanities, commerce, economics, business studies, law, social sciences, and general education.
Designed for Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), AI tutoring systems, educational chatbots, and question-answering applications, the dataset provides high-quality educational content with detailed explanations for every question.
Each sample includes multiple-choice options, the correct answer, and an explanation, making it suitable for supervised fine-tuning, instruction tuning, reasoning tasks, and educational AI research.
The dataset is distributed in both JSON and PDF formats.
Dataset Highlights
| Property | Value |
|---|---|
| Language | English |
| Total Questions | 2.5M+ |
| Domains | Non-STEM |
| File Formats | JSON & PDF |
| Question Type | Multiple Choice |
| Detailed Explanations | ✅ |
| Average Question Length | ~210 Words |
| AI Ready | ✅ |
Subject Coverage
The dataset covers a diverse collection of Non-STEM disciplines.
Humanities
- History
- Geography
- Political Science
- Civics
- Sociology
- Philosophy
- Psychology
- Social Science
Commerce & Business
- Commerce
- Business Studies
- Economics
- Management
- Entrepreneurship
- Marketing
- Financial Literacy
- Business Communication
Languages
- English Language
- General English
- Grammar
- Vocabulary
- Reading Comprehension
- Communication Skills
Law & Governance
- Law
- Constitution
- Public Administration
- Governance
- Public Policy
General Studies
- General Knowledge
- Current Affairs
- Teaching
- Library & Information Science
- Home Science
- Environmental Studies
- Miscellaneous Educational Topics
Dataset Features
- Over 2.5 Million English Questions
- High-quality educational content
- Multiple-choice question format
- Detailed answer explanations
- Human-reviewed educational material
- Competitive examination content
- School and university curriculum
- Structured JSON annotations
- Human-readable PDF documents
- AI-ready dataset
- Reasoning-focused questions
Sample Record
{
"question": "Which article of the Indian Constitution guarantees equality before the law?",
"options": [
"Article 14",
"Article 19",
"Article 21",
"Article 32"
],
"answer": "Article 14",
"explanation": "Article 14 guarantees equality before the law and equal protection of the laws within the territory of India."
}
Dataset Structure
English_NonSTEM_QA/
│
├── PDFs/
│ ├── History.pdf
│ ├── Economics.pdf
│ ├── Geography.pdf
│ ├── Business_Studies.pdf
│ └── ...
│
├── JSONs/
│ ├── History.json
│ ├── Economics.json
│ ├── Geography.json
│ ├── Business_Studies.json
│ └── ...
AI Tasks Supported
- Educational Question Answering
- Open-Domain Question Answering
- Reading Comprehension
- Retrieval-Augmented Generation (RAG)
- Supervised Fine-Tuning (SFT)
- Instruction Tuning
- Knowledge Retrieval
- Document Understanding
- Educational Search
- Reasoning
- Information Extraction
Applications
This dataset is ideal for:
- Large Language Model (LLM) Training
- Educational AI Assistants
- AI Tutors
- Question Answering Systems
- Digital Learning Platforms
- Examination Preparation Platforms
- Academic Search Engines
- Knowledge Base Construction
- Enterprise AI
- Educational Research
Industries
- Artificial Intelligence
- Education Technology (EdTech)
- Schools
- Universities
- Research Organizations
- Digital Publishing
- Government Education
- Online Learning Platforms
- Enterprise Knowledge Management
- Professional Training
Advantages
- Large-scale English educational corpus
- Comprehensive Non-STEM subject coverage
- Rich explanatory content
- AI-ready structured format
- Suitable for reasoning-based models
- Compatible with Hugging Face, ModelScope, and RAG pipelines
- Human-readable PDF references
- Well-structured JSON annotations
- Covers academic and competitive examination content
Dataset Statistics
| Metric | Value |
|---|---|
| Language | English |
| Questions | 2.5M+ |
| Subjects | 20+ |
| Formats | PDF & JSON |
| Question Type | Multiple Choice |
| Answer Explanations | Included |
| Average Question Length | ~210 Words |
License
This dataset is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) License.
You are free to use, modify, distribute, and build upon the dataset with appropriate attribution.
Citation
@dataset{english_nonstem_qa,
title={English Non-STEM Question & Answer Dataset},
year={2026},
license={CC BY 4.0}
}
Conclusion
The English Non-STEM Question & Answer Dataset is a large-scale educational resource containing over 2.5 million English question-answer pairs across history, geography, economics, commerce, business studies, law, political science, general knowledge, language studies, and other humanities disciplines. With detailed explanations, structured JSON annotations, and PDF source documents, it provides an excellent foundation for training Large Language Models (LLMs), Retrieval-Augmented Generation (RAG) systems, educational assistants, and intelligent question-answering applications.
- Downloads last month
- 12