bleeding happens for 2 reasons
bleeding happens for 2 reasons
Fu-net and using same tokens
Fyou need different class token if you want to minize bleeding
Fand of course text encoder training
Fso like
Fohwx man
Fbbuk woman
_Can you suggest a juggernautXL version that works with this code?
Fdid you try 9?
Fi assume it would work
Msenor Gözükara, I sag your post on CogVLM 2, and im now wondering if about captions. Are full sentences like in your example "the image featurees a man standing indoors, with a neutral expression on his face. [...]" better then buzzwords like "ohwx man, close-up front view, facing camera, wearing brown blazer over graphic t-shirt, slicked-back brown hair, groomed mustache, urban background with blurred street and greenery,"?
Dwith your setting for xl I can train many subjets of the same class with no bleeding at all and great results
_Yes, it did not work with both v8 and v9, let me know if there’s a working version. I even tried RealvisXL_v4 too and got the same error: AttributeError: 'NoneType' object has no attribute 'tokenize’
Don xl I trained many loras and finetunes many subjects of the same class and no bleeding, the model learns the diference between person1 man and person 2 man, in my sd3 tests all bleeds, is like is no learning that the combination of person1 man is diferent of person2 man, it seems to learning all mixed no mather what token or class Im using
DLets see with TE enabled, I will report in some hours when the new training is finished
Fwell best would be how users gonna prompt
Fwow nice
Fdid you enable SDXL checkbox?
Fit could be also due to your training config
Fthe config makes difference
_Yes, also how would I know what ip-adapter it’s using. And is there any way we could upload multiple images as input from a folder?
Fmultiple images are getting averaged but currently we dont have it
Fwhat you are trying to do exactly?
Fwhen i have time i will test if it actually fails to convert diffusers
_I tried doing it by converting the safetensors which didn’t work, now I’ve tried by adding the model id from hugging face into the list and it works for realvisXL.
I need to upload multiple images and create a virtual tryon image generation
I need to upload multiple images and create a virtual tryon image generation
Ffor virtual try on
FIDM vton is best
_Can we integrate that into this ?
Fyou can if you know programming :d
Fbut already have idm vton app why dont you check it?
Fi just completed a training for a client with CogVLM2 captions
Fand results looking stunning
Fstill comparing with raw prompt
Fi just did a new training for 40 images
Fafter 20k steps it started to became overtrained
Freg images used
Fraw captions still provides better likeliness
Fi feel like 200 epoch yielded best
F200 x 40 x 2 = 16000 steps
PWhat's the most recent koyha install/ guide?
There are so many Patreon posts that I don't use that I can no longer keep up with the ones I might use.
Is there some way to sort them that I am missing?
There are so many Patreon posts that I don't use that I can no longer keep up with the ones I might use.
Is there some way to sort them that I am missing?
Fwhere you want to install on windows?
Pyes windows 10
Dupdate, with 25 epochs 3 repetitions and the TE learning rate at the same learning rate as the unet, the bleeding I think is fixed, I'm no so sure yet because is undertrained, I will resume the training for 25 more ephocs but the results are promising
Fkohya gui has auto install bat file
Fitself
Fyou only need to have python 3.10
FPatreon
Get more from SECourses: Tutorials, Guides, Resources, Training, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözükara on Patreon
Fi plan to fine tune sd3 with huge dataset
Fnon-commercial use free
Findividual commercial 50$ and company 500$ . what you think