# Fine tuning camemBERT on custom dataset for intent classification

**URL:** <https://forum.rasa.com/t/fine-tuning-camembert-on-custom-dataset-for-intent-classification/44566>\
**Category:** Rasa Open Source\
**Created:** [June 13, 2021, 9:39pm UTC](https://forum.rasa.com/t/fine-tuning-camembert-on-custom-dataset-for-intent-classification/44566 "2021-06-13T21:39:02Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![anojl](https://avatars.discourse-cdn.com/v4/letter/a/3bc359/32.png) [@anojl](https://forum.rasa.com/u/anojl)\
**Post date:** [June 13, 2021, 9:39pm UTC](https://forum.rasa.com/t/fine-tuning-camembert-on-custom-dataset-for-intent-classification/44566/1 "2021-06-13T21:39:02Z")

</div>

Hello community! I am trying to use the model CamemBERT for french language in my chatbot project and i have a custom dataset with intents and specific entities. so i’m using the model name camemBert from hg\_tranformers and i downloaded a pretrained model for it. My question is how can I fine tune this model in order to perform well on my custom dataset ? do I need to re-trained this model on my dataset or i have to add an extra layer for intent classification ? I am new in using this type of language model. Thank you

---

<div class="post-metadata">

**Author:** ![nik202](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/nik202/32/16098_2.png) [@nik202](https://forum.rasa.com/u/nik202)\
**Post date:** [June 14, 2021, 1:00am UTC](https://forum.rasa.com/t/fine-tuning-camembert-on-custom-dataset-for-intent-classification/44566/2 "2021-06-14T01:00:31Z")

</div>

You need to check this thread [RASA and camemBERT - #16 by khatba](http://forum.rasa.com/t/rasa-and-camembert/31882/16) hope it will help?

---

<div class="post-metadata">

**Author:** ![anojl](https://avatars.discourse-cdn.com/v4/letter/a/3bc359/32.png) [@anojl](https://forum.rasa.com/u/anojl)\
**Post date:** [June 15, 2021, 4:49pm UTC](https://forum.rasa.com/t/fine-tuning-camembert-on-custom-dataset-for-intent-classification/44566/3 "2021-06-15T16:49:20Z")

</div>

Thann you for your answer! Yes i checked that thread but they only discuss how to use the pretrained model directly in the pipeline, my issue is i don’t want to use the pretrained model untill i fine tune it on my custom dataset . So how do I re-train this model on my dataset before i apply it in my pipeline ?