| --- |
| Modalities: Text |
| language: |
| - ar |
| license: cc-by-4.0 |
| task_categories: |
| - question-answering |
| - multiple-choice |
| task_ids: |
| - multiple-choice-qa |
| modalities: |
| - text |
| formats: |
| - parquet |
| tags: |
| - arabic |
| - non-stem |
| - question-answering |
| - mcqa |
| - multiple-choice-questions |
| - educational |
| - education |
| - humanities |
| - social-science |
| - history |
| - geography |
| - economics |
| - polity |
| - general-knowledge |
| - reasoning |
| - exam-preparation |
| - competitive-exams |
| - nlp |
| - llm |
| - instruction-tuning |
| - sft |
| - rlhf |
| - self-supervised-learning |
| - knowledge-retrieval |
| - educational-ai |
| - arabic-language |
| - multilingual |
| pretty_name: STEM QnA Sample Dataset |
| size_categories: |
| - 1K<n<10K |
| --- |
| **Dataset Description:** |
|
|
| **This dataset is a large-scale collection of Arabic Non-STEM Question Answering (QA) data, containing 70,518 question-answer pairs, designed to support the development and training of advanced NLP systems and AI models for language understanding, reasoning, knowledge retrieval, and educational learning in Arabic.** |
|
|
| It is well-suited for **Supervised Fine-Tuning (SFT) and Reinforcement Learning with Human Feedback (RLHF) workflows**, helping improve model performance in question answering, reasoning, and multilingual general knowledge tasks. |
|
|
| **Dataset Specification** |
|
|
| -Total: 70,518 |
| -Modality: Arabic text (MCQ-based question-answer pairs) |
| -Type: Educational / (Non-STEM) |
| -Data Source: Curated academic and general knowledge resources |
| -Data Nature: Real-world and curated data |
| -Content: Questions with options and correct answers |
| |
| **Key Use Cases** |
|
|
| -Question Answering (QA) (MCQ-based) in Arabic |
| -Automated tutoring and educational assistants |
| -Knowledge retrieval systems |
| -Model benchmarking and evaluation |
| |
|
|
| **Value of This Dataset** |
|
|
| -Supports learning of general knowledge domains in Arabic |
| -Improves reasoning and comprehension capabilities of AI models |
| -Enables development of multilingual QA systems |
| -Enhances accuracy of LLMs in non-STEM domains |
| -Provides broad coverage across diverse subjects |
| |
| **Basic JSON Schema** |
| ```json |
| { |
| "answer_type": "string", |
| "q_string": "string", |
| "q_option": ["string"], |
| "q_answer": "string", |
| "lang_code": "string", |
| "category": "string" |
| } |
| ``` |
| **Full Dataset Overview** |
|
|
| 6.5M+ Questions / 1.8B+ Tokens |
|
|
| This scale provides extensive domain coverage, rich contextual learning, and significantly improves language understanding, reasoning, and model performance. |
|
|
| **Data Creation** |
|
|
| Procured through formal agreements and generated in the ordinary course of business. |
|
|
| **Considerations** |
|
|
| This dataset is provided for research and educational purposes only. It contains only sample data. For access to the full dataset and enterprise licensing options, please visit our website [InfoBay.AI](https://infobay.ai/) or contact us directly. |
|
|
| -Ph: (91) 8303174762 |
| -Email: datareq@infobay.ai |
