HHHedo

Hi, I have the same question. It seems that the center is the sum of features instead of mean.

> Hi, There are normalize operation, Beacause Cosine similarity only cares about direction！ Thanks for your reply!

Hi bin, I solve the problem of same attention score by removing the dimension normalization, and the performance is comparable. However, I am still confused about the pre-trained models and...

> Hi, please make sure that the weights are indeed fully loaded into your model without mismatch; you can set `strict=True` in `torch.load()`. There are multiple `embedder.pth` files available, and...

> > Hi, please make sure that the weights are indeed fully loaded into your model without mismatch; you can set `strict=True` in `torch.load()`. There are multiple `embedder.pth` files available,...

> Of course, you can download it from https://drive.google.com/file/d/1h7SSzndEm486-gAbwtKA81hO4TA84Ita/view?usp=sharing Thanks for your great work! I wonder whether you can release other pre-trained models (e.g., the models of chen_MIA and XIE_miccai),...

You may try a smaller weight decay.