Custom masking about mae HOT 11 OPEN

Hi,

The PatchShuffle class is doing two things in sequence:

1. create the mask, in which the cnn output here is only help to specify the dimensions.
2. use the mask to mask out the input.

You can of course implement these two things separately with two classes or functions. I implemented in this way only for convinent. And it is different with the official implementation since when I wrote the code, the official one was not yet released.

And it is also very straightforward to understand which patch comes from which region of the image. Say your input is 224x224 image, and patch size is 14, then you will get a 16x16 grid of patches from the conv and each patch on this grid is from a 14x14 region from the original image without overlapping.

from mae.

Hi, thank you for sharing the code.
why did not you use sine-cosine positional embedding as it is mentioned in the paper?

from mae.

I don't find where they mention of using sin-cos positional embedding in the paper. Actually, the original ViT paper clearly mentioned that a "learned" positional encoding is added after patchfication. Also for images, it is not necessary to use the sin-cos positional encoding since there is no extrapolation beyond the trained length. Could you point out where you read it?

from mae.

Sure, in the paper https://arxiv.org/abs/2111.06377, on page 11, first paragraph.

from mae.

ah, I see. Thanks for the reference. I didn't pay much attention to this detail. But, as I said, I don't think it will make a large difference to the result. Feel free to experiment with that.

from mae.

Also, I just checked their official code and they don't even follow this detail. The code uses the ViT model from timm which follows the details in the ViT paper with learned positional encoding.

from mae.

https://github.com/facebookresearch/mae/blob/main/models_mae.py
You can see that they utilized the frozen positional embedding using the sine-cosine approach.

from mae.

Ah, thanks for the correction. I had looked at a wrong file. Then I don't know why they don't like to follow the ViT architecture precisely.

from mae.

Based on what I studied, we do not have specific rules for choosing the positional embedding. However, I want to try the sine-cosine approach, and see the result. If I test it, I will inform you.
I want to be sure that this implementation considers other details mentioned in the paper. I checked it, however, I want to be sure.

from mae.

Oh, I don't think I followed all the details from the paper precisely. As in the readme, the purpose of this code is only to verify the idea of mae, not a replicate. For example, I think I didn't implement the normalization for reconstruction loss. There could be more details that I missed.

from mae.