# Adapter Training - Merging Weights

**URL:** <https://discuss.huggingface.co/t/adapter-training-merging-weights/35817>\
**Category:** Intermediate\
**Created:** [April 6, 2023, 9:00pm UTC](https://discuss.huggingface.co/t/adapter-training-merging-weights/35817 "2023-04-06T21:00:33Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Praneet](https://avatars.discourse-cdn.com/v4/letter/p/e9a140/32.png) [@Praneet](https://discuss.huggingface.co/u/Praneet)\
**Post date:** [April 6, 2023, 9:00pm UTC](https://discuss.huggingface.co/t/adapter-training-merging-weights/35817/1 "2023-04-06T21:00:33Z")

</div>

Hey guys,

Would like to preface this by saying that I am new to all this.  
I fine-tuned a T5 model using the adapter transformer library. I was just wondering what the function  
model.merge\_adapter(“adapter name”) does

 ![merge](https://us1.discourse-cdn.com/hellohellohello/original/2X/5/5ab4436bf14a388bd2b1465a1871ebbdc5b9c0b7.png)

Does it somehow merge the weights of the adapter with the base model itself?  
Does this mean I wouldn’t have to activate the adapter manually every time with model.set\_active\_adapters()  
Also, I was wondering if there is a way to always have a particular adapter head activated instead of having to activate it anew every time at the beginning of the script.

Any help would be appreciated. Thanks a ton!