---
license: cc-by-sa-3.0
task_categories:
- text-retrieval
language:
- en
pretty_name: Wikpedia Paragraphs MPNet Embeddings
---

Embeddings of the [english Wikipedia](https://huggingface.co/datasets/wikipedia) [paragraphs](https://huggingface.co/datasets/olmer/wiki_paragraphs) using [all-mpnet-base-v2](https://huggingface.co/sentence-transformers/all-mpnet-base-v2) sentence transformers encoder.  
The dataset contains 43 911 155 paragraphs from 6 458 670 Wikipedia articles.  
The size of each paragraph varies from 20 to 2000 characters.  
For each paragraph there is an embedding of size 768.  
Embeddings are stored in numpy files, 1 000 000 embeddings per file.  
For each embedding file, there is an ids file that contains the list of ids of the corresponding paragraphs.  
__Be careful, dataset size is 151Gb__.