# Visualize attention maps of ViT

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fvisualize-attention-maps-of-vit%2F160252%3E%5C
**Category:** Models\
**Created:** [June 22, 2025, 10:58am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fvisualize-attention-maps-of-vit%2F160252 "2025-06-22T10:58:03Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![asifdegr8](https://avatars.discourse-cdn.com/v4/letter/a/6de8d8/32.png) [@asifdegr8](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fasifdegr8)%5C
**Post date:** [June 22, 2025, 10:58am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fvisualize-attention-maps-of-vit%2F160252%2F1 "2025-06-22T10:58:03Z")

</div>

I have create ViT model using timm for 5 class classification as follows.  
model = timm.create\_model(‘vit\_base\_patch16\_224’, pretrained=True, num\_classes=5)

I want to visualize attention maps for it. How to do this?

---

<div class="post-metadata">

**Author:** ![John6666](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/john6666/32/27664_2.png) [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [June 22, 2025, 12:02pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fvisualize-attention-maps-of-vit%2F160252%2F2 "2025-06-22T12:02:16Z")

</div>

Hmm… Like this?

> **[Visualize attention map for vision transformer · huggingface...](https://github.com/huggingface/pytorch-image-models/discussions/1232)**
>
> Hi, I want to extract attention map from pretrained vision transformer for specific image. How I can do that?

> **[Visualizing Attention Maps in Pre-trained Vision Transformers (Pytorch)](https://alessiodevoto.github.io/vit-attention/)**
>
> Goal: Visualizing the attention maps for the CLS token in a pretrained Vision Transformer from the timm library.

> **[app.py · timm/timmAttentionViz at main](https://huggingface.co/spaces/timm/timmAttentionViz/blob/main/app.py)**
>
> We’re on a journey to advance and democratize artificial intelligence through open source and open science.