Dataset · sentence-transformers
Natural Questions question-answer pairs for embeddings
sentence-transformers/natural-questions
English question-answer pairs in Sentence Transformers format for training search and semantic similarity models.
What's inside
An adapted version of the Natural Questions dataset consisting of string pairs in question and answer columns. Contains from 100,000 to 1,000,000 English examples. Data is prepared for direct use with the Sentence Transformers library.
What you can do with it
- Training embedding models (bi-encoders) for semantic search
- Building and testing search components in RAG systems
- Tasks
- Embeddings, Sentence similarity
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 3.6K
- Likes
- 52
Models trained on it
- sparse-encoder-testing/splade-bert-tiny-nqEmbeddings 34.9K
- jaredpalmer/kev-27bText classification 3.4K
- mradermacher/kev-27b-GGUFMultiple choice 1.6K
- techdotus/Ethoric-10MText generation 1.3K
- mradermacher/Llama-3.2-3B-Instruct-natural-questions-i1-GGUF 665
- mradermacher/rush-small-qwen3-reranker-0.6b-v1-GGUF 593
- sparse-encoder-testing/splade-bert-tiny-nq-openvinoEmbeddings 379
- sparse-encoder-testing/splade-bert-tiny-nq-onnxEmbeddings 328
How to load it
from datasets import load_dataset
ds = load_dataset("sentence-transformers/natural-questions")