You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the proposed paper, the method compares with other methods by the segmentation result in model: Deit S/tiny AR S/tiny,
while the release code in imagenet-seg-eval is compared in model: vit-base.
Is this setting not matching to the paper?
The text was updated successfully, but these errors were encountered:
The released code in imagenet-seg-eval is DeiT-S.
The original code(https://github.com/hila-chefer/Transformer-Explainability) contains vit-base, but the eval code we released is DeiT-S.
You can find model code in imagenet_seg_eval_ICE.py line 159.
In the proposed paper, the method compares with other methods by the segmentation result in model: Deit S/tiny AR S/tiny,
while the release code in imagenet-seg-eval is compared in model: vit-base.
Is this setting not matching to the paper?
The text was updated successfully, but these errors were encountered: