# Rasa 3 NLU Pipeline Custom Component

**URL:** <https://forum.rasa.com/t/rasa-3-nlu-pipeline-custom-component/51946>\
**Category:** Rasa Open Source\
**Created:** [March 14, 2022, 1:13pm UTC](https://forum.rasa.com/t/rasa-3-nlu-pipeline-custom-component/51946 "2022-03-14T13:13:56Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![daniel-rand-rasa](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/daniel-rand-rasa/32/19820_2.png) [@daniel-rand-rasa](https://forum.rasa.com/u/daniel-rand-rasa)\
**Post date:** [March 14, 2022, 1:13pm UTC](https://forum.rasa.com/t/rasa-3-nlu-pipeline-custom-component/51946/1 "2022-03-14T13:13:56Z")

</div>

would like some help building a custom component into my NLU pipeline. Specifically, I am trying to pre-process user messages before extracting entities, more specifically I am attempting to apply spell check and case-correction in order to improve the correctness of my entity extractor. However, while I am able to train my model successfully, when I run my application I get the message:

rasa.core.agent - Could not load model due to Error deserializing graph schema. Can’t find class for graph component type
