jacobgil / vit-explain

Explainability for Vision Transformers

Repository from Github https://github.comjacobgil/vit-explain

jacobgil/vit-explain Issues

ViT explanation for regression problem (no class label)?
Updated 10 months ago
normalize (sum to 1) attention score seems not right
Updated a year ago3
What if the visual transformer does not have a class token?
Updated a year ago1
result = torch.eye(attentions[0].size(-1)) problem
Updated a year ago7
TypeError, IndexError, ValueError occured
Closed a year ago
Cannot use block.attn.fused_attn = False in another ViT model
Closed a year ago
[Feature request] Activation maximization code
Updated a year ago
Rollout for different ViT models
Updated 2 years ago1
Would this work with Swin Transformer
Closed 2 years ago1
Visualization Results on DeiT-Base
Updated 2 years ago1
Error in tensor size mismatch
Updated 3 years ago1
How to use this for a 3D grayscale image?
Updated 3 years ago
cv2 is bgr whereas matplotlib is rgb
Updated 3 years ago
Shape mismatch for timm vit model
Updated 3 years ago
Evaluating Explanation
Updated 3 years ago
Normalization of Vit_explain code
Updated 3 years ago
Clipped_advs is not having value between 0 - 1
Closed 3 years ago
Gradients in Grad Rollout
Updated 3 years ago1
What is the meaning of 'flat' variable in vit_rollout.py?
Updated 4 years ago1
"unexpected keyword argument 'head_fusion'" and "required positional argument: 'category_index'"
Updated 4 years ago
What do we do for an image shape is (256, 128)?
Closed 4 years ago
visualize timesformer
Updated 4 years ago
How do we know the index category?
Updated 4 years ago
Code for Google's ViT and complete example
Closed 4 years ago2
No separate `head_fusion` strategies for Gradient Attention Rollout?
Closed 4 years ago1
Can it be helpful for understanding vistual transformers?
Closed 5 years ago1