Hello!
Get an error when trying to resume training on full pretrain, code cant find global.step
When training validation of image_embed_gs is always a row of blank gray images

And i never get very good results from image 2 maps condition
also i see zero embbed as an option, would this model do well on an exclusive image2maps translation?
Hello!
Get an error when trying to resume training on full pretrain, code cant find global.step
When training validation of image_embed_gs is always a row of blank gray images
And i never get very good results from image 2 maps condition
also i see zero embbed as an option, would this model do well on an exclusive image2maps translation?