MAE-pytorch

MAE-pytorch copied to clipboard

MAE prediction visualization code

9

Thank you for your contribution. I wonder if you plan to release the mask prediction visualization code?

tikboaHIT

knn eval of MAE

13

I eval vit_base of 500/1600 pretraining on imagenet1000 using knn metric. By loading all the pretained parameter with vit GAP method (not need cls token), the knn 20-NN result is...

Dongshengjiang

Hi again, sorry for the slow response in issue #26. I have some more clarifications and visualizations here. I agree that the sine-cosine embeddings are not learnable. However it seems...

atonderski

typo error local-rank

1

Running your code I faced with this problem. I check your code and see that you have local_rank instead. So please correct this error. ``` usage: MAE pre-training script [--batch_size...

Jalilnkh

It seems too slow to load data, and makes the gpu use rate is low.

maichm

Your work is so excellent! However, when I try to fine-tuning on my downstream regression task, I found that when using Mixup method, the parameter of the function contains: `num_classes=args.nb_classes`....

SCU-zly

在dataset-folder中，import accimage cannot be parsedhow to solve it here def accimage_loader(path: str) -> Any: import accimage try: return accimage.Image(path) except IOError: # Potentially a decoding problem, fall back to PIL.Image...

zhanglaoban-kk

Error reported in code finetune, AttributeError: 'VisionTransformer' object has no attribute 'get_num_layers'.

1


Liuhp133

Misaka17032

Hi, thanks for the nice code. How to implement Layer-wise learning rate decay on ResNet instead of ViT?

jingzhengli