Well if there are no prompts then what
Well if there are no prompts then what is the basis for the classification images?
12 Replies
so just to be clear - I do have instance/class prompts, I just don't use any text file/descriptions on either class or instance sets
So then you are training specific prompts not just a group of images.
ok so the setup is the standard one - instance prompt -
ohwx man class prompt man
then you point the settings to the locations of your instance/class set
then input the number of epochs/steps and start trainingYea ok..... it seems like we are just talking about two different things then.
sorry so what are you talking about?
I am talking about training dreambooth with a set of your own images
I am talking training with a large set of a variety of images of various themes..... not a single person.
so you are trying to train styles?
No I ma trying to add images and concepts to an existing checkpoint.
Interesting.. this is considered "finetuning"?
So i can add different fashions maybe to my base model of my selfies?
To be honest I an not sure how or if it can be done.... that is what I am asking. I would like to take a checkpoint and add a bunch of stuff to it. Dr. Furkan indicates it can done so I will probably just have to wait for the video. 😅
copy, im in the same boat then
I have only gotten that done in Embeddings
wanna test it out in Dreambooth
It seems that someone at some point must have added additional tagged images to the original stable diffusion checkpoint to make the various other checkpoints that are used as the basis for merges..... so it seems like it must be doable but maybe it is not realistic.