# Token classification probability and scoring

**URL:** <https://discuss.huggingface.co/t/token-classification-probability-and-scoring/2172>\
**Category:** 🤗Transformers\
**Created:** [November 23, 2020, 10:43am UTC](https://discuss.huggingface.co/t/token-classification-probability-and-scoring/2172 "2020-11-23T10:43:15Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Zack](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/zack/32/4845_2.png) [@Zack](https://discuss.huggingface.co/u/Zack)\
**Post date:** [November 23, 2020, 10:43am UTC](https://discuss.huggingface.co/t/token-classification-probability-and-scoring/2172/1 "2020-11-23T10:43:16Z")

</div>

I am building an NER model, I must have for each tagged entity a probability to indicate the confidence prediction for each label,  
I tried to just use the model output for the last classification layer, but it seems for me it’s not a probability and all values are so close (see below).

 ![image](https://us1.discourse-cdn.com/hellohellohello/original/2X/e/e12cce4ad2e54b5ba69c72a108e9c210c4ab7a17.png)

Any additional information would be helpful!  
Thank u  
Zakarya