# Retraining pre-trained NER model with new data samples

**URL:** <https://discuss.huggingface.co/t/retraining-pre-trained-ner-model-with-new-data-samples/51831>\
**Category:** 🤗Transformers\
**Created:** [August 23, 2023, 1:10pm UTC](https://discuss.huggingface.co/t/retraining-pre-trained-ner-model-with-new-data-samples/51831 "2023-08-23T13:10:14Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Abdo1Kamr](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/abdo1kamr/32/18813_2.png) [@Abdo1Kamr](https://discuss.huggingface.co/u/Abdo1Kamr)\
**Post date:** [August 23, 2023, 1:10pm UTC](https://discuss.huggingface.co/t/retraining-pre-trained-ner-model-with-new-data-samples/51831/1 "2023-08-23T13:10:14Z")

</div>

Hi,  
I have a pre-trained [NER model](https://huggingface.co/marefa-nlp/marefa-ner) this model has 19 class labels, and I want to fine-tune this model by having data that has 15 new class labels.

There are some class labels overlapping from the old 19 class labels and the new 15 class labels.

So, my questions:

1. How can fine-tune this model with all class labels “19 + 15”?
2. How can fine-tune this model with just 15 class labels?

Thanks.

---

<div class="post-metadata">

**Author:** ![Fatimetou](https://avatars.discourse-cdn.com/v4/letter/f/eb9ed0/32.png) [@Fatimetou](https://discuss.huggingface.co/u/Fatimetou)\
**Post date:** [May 3, 2024, 2:50pm UTC](https://discuss.huggingface.co/t/retraining-pre-trained-ner-model-with-new-data-samples/51831/2 "2024-05-03T14:50:35Z")

</div>

Hello,

I wonder of you found any solution?

Thanks