Open yangzhip opened 2 years ago
@yangzhip you need to set image_embed_dim = 512
when instantiating DiffusionPrior
@yangzhip you need to set
image_embed_dim = 512
when instantiatingDiffusionPrior
Yes, the dimension I have changed in the config file is 512, but there is still an error. The character of the text in the code after the clip is 768. How to change it here? Thank you!
@yangzhip could you share your script?
RuntimeError: Sizes of tensors must match except in dimension 1. Expected size 768 but got size 512 for tensor number 3 in the list.
my clip feature is 512 how can i train this prior