# How Can I use cashed models from HuggingFace?

**URL:** <https://discuss.huggingface.co/t/how-can-i-use-cashed-models-from-huggingface/61679>\
**Category:** 🤗Transformers\
**Created:** [November 9, 2023, 4:45pm UTC](https://discuss.huggingface.co/t/how-can-i-use-cashed-models-from-huggingface/61679 "2023-11-09T16:45:17Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![khabibulloevms](https://avatars.discourse-cdn.com/v4/letter/k/a6a055/32.png) [@khabibulloevms](https://discuss.huggingface.co/u/khabibulloevms)\
**Post date:** [November 9, 2023, 4:45pm UTC](https://discuss.huggingface.co/t/how-can-i-use-cashed-models-from-huggingface/61679/1 "2023-11-09T16:45:17Z")

</div>

I want to use models from: [https://huggingface.co/ARTeLab/mbart-summarization-mlsum](https://stackoverflow.com/) in offline mode, meaning that after downloading them from Hugging Face, they will be saved locally and I will be able to use them offline. However, I don’t know how to do this. If anyone has already figured this out, please advise me. I use these lines to download models:

```auto
from transformers import MBartTokenizer, MBartForConditionalGeneration
tokenizer = MBartTokenizer.from_pretrained("ARTeLab/mbart-summarization-mlsum")
model = MBartForConditionalGeneration.from_pretrained("ARTeLab/mbart-summarization-mlsum")

```

The problem is that when I run this line, I download several files from the repository at once, and I don’t know which one is then used for tokenization:

 ![image](https://us1.discourse-cdn.com/hellohellohello/original/3X/9/6/96dc036932fc219bb9296a777a8bea569a224f97.png)

If they are all used, how can I know how to use them?