# Multilingual-e5 sparse vector support

**URL:** <https://discuss.elastic.co/t/multilingual-e5-sparse-vector-support/368497>\
**Category:** Elasticsearch\
**Created:** [October 9, 2024, 7:10am UTC](https://discuss.elastic.co/t/multilingual-e5-sparse-vector-support/368497 "2024-10-09T07:10:21Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![czek0](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/czek0/32/136480_2.png) [@czek0](https://discuss.elastic.co/u/czek0)\
**Post date:** [October 9, 2024, 7:10am UTC](https://discuss.elastic.co/t/multilingual-e5-sparse-vector-support/368497/1 "2024-10-09T07:10:21Z")

</div>

Hello  
Upon investigatoing into your examples for the miltilingual-e5-base, I see you are using the **dense vector** type as the passage\_embedding. Is it possible to use sparse vector types as the passage\_embedding? And how can we extend it to support different dimensions in the sparse\_vector. Unless there is a default dimension size for the sparse vector type.

see here:|[Google Colab](https://colab.research.google.com/github/elastic/elasticsearch-labs/blob/main/notebooks/search/04-multilingual.ipynb#scrollTo=TF_wxIAhD07a)