Dataset · microsoft
WikiQA: Questions and sentences for QA
microsoft/wiki_qa
English question-sentence pair corpus for open-domain question answering research.
What's inside
The set contains question-sentence pairs collected and annotated for open-domain question answering. This is the public WikiQA corpus from Microsoft in English. Metadata indicates a size between 10K and 100K examples, with the format not detailed in the card.
What you can do with it
- Train an extractive QA model in English
- Test the answer sentence retriever/ranker
- Evaluate an open-domain QA system
- Tasks
- Question answering
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 6.4K
- Likes
- 75
Models trained on it
- mradermacher/Safe4Qwen-1.7B-i1-GGUF 969
- FlameF0X/Qwen2-0.2B-itText generation 409
- mradermacher/Safe4Qwen-1.7B-GGUF 215
- DreamBlooms/Dohnuts-0.1.0-0.8B-GGUF 128
- parsecai/ragzillaText classification 56
- nmngpt0/bge-qa-lilsmall01Embeddings 37
- yophis/DRM-T5-Base-wikiqaText classification 11
- yophis/DRM-DeBERTa-v3-Base-wikiqaText classification 6
How to load it
from datasets import load_dataset
ds = load_dataset("microsoft/wiki_qa")