# SequenceClassification Head is finetuned for sentiment task?

**URL:** <https://discuss.huggingface.co/t/sequenceclassification-head-is-finetuned-for-sentiment-task/12628>\
**Category:** Models\
**Created:** [December 8, 2021, 8:33am UTC](https://discuss.huggingface.co/t/sequenceclassification-head-is-finetuned-for-sentiment-task/12628 "2021-12-08T08:33:26Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![rishtrip](https://avatars.discourse-cdn.com/v4/letter/r/13edae/32.png) [@rishtrip](https://discuss.huggingface.co/u/rishtrip)\
**Post date:** [December 8, 2021, 8:33am UTC](https://discuss.huggingface.co/t/sequenceclassification-head-is-finetuned-for-sentiment-task/12628/1 "2021-12-08T08:33:26Z")

</div>

Does HuggingFace finetune the RobertaForSeqClassification head for sentiment prediction? In other words, is the SeqClassification head (i.e. additional layers over BertModel) trained by HuggingFace?  
I am asking this to know why the SeqClassification head does so good for sentiment task?
