# Improve Performance on zero shot classification

**URL:** <https://discuss.huggingface.co/t/improve-performance-on-zero-shot-classification/1255>\
**Category:** Beginners\
**Created:** [September 24, 2020, 7:16pm UTC](https://discuss.huggingface.co/t/improve-performance-on-zero-shot-classification/1255 "2020-09-24T19:16:22Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![IronCrystal](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/ironcrystal/32/383_2.png) [@IronCrystal](https://discuss.huggingface.co/u/IronCrystal)\
**Post date:** [September 24, 2020, 7:16pm UTC](https://discuss.huggingface.co/t/improve-performance-on-zero-shot-classification/1255/1 "2020-09-24T19:16:22Z")

</div>

Hello,

I am just getting started with machine learning, but I am trying to use the zero shot learning that was just added recently.

This is how I am initializing the classifier:

`classifier = pipeline("zero-shot-classification", model='facebook/bart-large-mnli', device=0)`

I’ve noticed that this is using only a single CPU core and a single GPU (even on aws instances with multiple GPUs).

Is there a flag that I need to use to tell it to use multiple GPUs? Or is this classifier limited to single GPU and single core performance?