Hello,
I want to get the embeddings for each patch (dim = 512, not 768) of the image, how can I achieve this? should I just multiply them with the projection matrix or there are some normalization step that I have to do?
Thank you
Hello,
I want to get the embeddings for each patch (dim = 512, not 768) of the image, how can I achieve this? should I just multiply them with the projection matrix or there are some normalization step that I have to do?
Thank you