You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
i have written the transunet code in tensorflow using Resnet-50 and without it i.e., using CNN layers and the parameters Image_size=(224,224,3), embed_dim=512, MLP size= 3072, num_head=12, num_transformer_layer=12. The number of parameters i got is 409,616,001. so i would like to know where i need to change so that i get 105 million parameters. pliz help me
The text was updated successfully, but these errors were encountered:
i have written the transunet code in tensorflow using Resnet-50 and without it i.e., using CNN layers and the parameters Image_size=(224,224,3), embed_dim=512, MLP size= 3072, num_head=12, num_transformer_layer=12. The number of parameters i got is 409,616,001. so i would like to know where i need to change so that i get 105 million parameters. pliz help me
The text was updated successfully, but these errors were encountered: