# Should I need to use pre\_train-tokenizer?

**URL:** <https://discuss.huggingface.co/t/should-i-need-to-use-pre-train-tokenizer/18861>\
**Category:** 🤗Transformers\
**Created:** [June 8, 2022, 1:32pm UTC](https://discuss.huggingface.co/t/should-i-need-to-use-pre-train-tokenizer/18861 "2022-06-08T13:32:16Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Imran1](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/imran1/32/18779_2.png) [@Imran1](https://discuss.huggingface.co/u/Imran1)\
**Post date:** [June 8, 2022, 1:32pm UTC](https://discuss.huggingface.co/t/should-i-need-to-use-pre-train-tokenizer/18861/1 "2022-06-08T13:32:16Z")

</div>

If I have legal text data and I want to summarize, so the new word which not present in pre-train\_tokenizer (embading), obviously the the cousin semilarty will be zero.

Example  
cousin(“judiciare”, “judgement”).