Open Yangr116 opened 3 years ago
Hello, thanks for your best work!
I would like to know why you use the function "_init_weights" in all class, including 'MLP, Attention'? Maybe, one init_weights function is enough in the 'class PyramidVisionTransformerV2(nn.Module)'.
Is it will get a better initialization?
Hello, thanks for your best work!
I would like to know why you use the function "_init_weights" in all class, including 'MLP, Attention'? Maybe, one init_weights function is enough in the 'class PyramidVisionTransformerV2(nn.Module)'.
Is it will get a better initialization?