# Why is BertNormalizer not exposed on the tokenizers library?

**URL:** <https://discuss.huggingface.co/t/why-is-bertnormalizer-not-exposed-on-the-tokenizers-library/23340>\
**Category:** 🤗Tokenizers\
**Created:** [September 19, 2022, 12:03pm UTC](https://discuss.huggingface.co/t/why-is-bertnormalizer-not-exposed-on-the-tokenizers-library/23340 "2022-09-19T12:03:25Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![ndamulelonemakh](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/ndamulelonemakh/32/25399_2.png) [@ndamulelonemakh](https://discuss.huggingface.co/u/ndamulelonemakh)\
**Post date:** [September 19, 2022, 12:03pm UTC](https://discuss.huggingface.co/t/why-is-bertnormalizer-not-exposed-on-the-tokenizers-library/23340/1 "2022-09-19T12:03:25Z")

</div>

Hi all 🙂

I am trying to use the BertNormalizer() to train a ByteLevelBPETokenizer(). But it seems this normaliser is not exposed looking at the source(see attached image). I am using tokenizers==0.12.1

 ![tokenizersbert](https://us1.discourse-cdn.com/hellohellohello/original/2X/e/e76574ecd726199e5f7d1032303a3434a65a84eb.jpeg)

Was this intentional or is it a feature that will be added going forward?