must be something with the sampler type, i got it too, but it works correct with the right one, try
must be something with the sampler type, i got it too, but it works correct with the right one, try normal and others
Ji always read the captions and edit them manually if i see something is wrong
DI'm doing some testing training flux on kohya but I'm having a bleeding problem, I'm training pictures of myself as a kid and as and adult and a friend, so the captions are "myname man" "myname boy" "myfriendname man" and the concepts bleed, this problem doesn't happen training sdxl models, there is no bleeding. maybe the problem is because I'm not training the text encoder?
DI hade a similar problem with my sd3 testing
Fye not training text encoder will cause
Fmore bleeding
Fsince you only train unet
Fclip l training not supported yet
Dohh thanks, so may be is that
II've found it was the vae, are for shell not for dev
DAnother question: the CLIP models, such as clip_l, are stored in separate files from the base model. If I use a concept that doesn’t exist in CLIP, such as ohwx or another non-existent word, would it still work during inference if it's not part of the model because the model was not trained? Additionally, if I train clip_l, would it be saved as a separate file? Currently, the text encoders are not in the same safetensors as the base model
Fit will work but lesser degree
Fsince you will train unet of the model
For if you have another question explain in more details
Fstarted second 4 gpu machine
YYeah, it a lot faster to edit prompts than write them
DThe doubt arises because the text encoder is in a different file, and if I train it, would Koha save it as a separate file? In that case, should I also use my custom clip_l for inference? Ideally, the text encoders should be included in the same SafeTensor file, as happens in SDXL
Fif you trained you have to use it during inference for best quality
Dthanks, I hope in the future the text encoders can be embebed in the model file
Fkohya will support very likely when supporting clip l training
Dthanks!! I hope so!! and I hope that 24gb vram will be enough to train flux + clip_l
Fcurrently only unet trained
Fwith kohya lora
F24 gb barely sufficient
Fupdated post with experiments
DI will check it
F
Iwith 24GB only Loras?
FYe this is only lora but I think fine tuning has better optimizations we will see. Also I asked kohya since there is half model training but still uses 18 gb gpu
Fmore than 50 epoch completed time to start doing some testing
Dso nothing for 16 GB?
Fthere is one config supposed to work but not working i messaged him
Fi think fine tuning may work for 16
Ffor lora i think there is bug
F32 rank full precision lora is 613 mb
Dwdym fine tuning on 16 GB but not lora?!
Fye 
Fbecause of optimiziations
Dhow can fine tuning work on 16 GB when the model is already 12B * 2 bytes?
Fthere is half model training for lora but it is peaking like this
Fa lot of optimization with speed loss
FFthis above is lora training half model
Fit peaks 18 gb
JDo you have a config I can download for Flux training on Kohya, just to get setup quickly and I can make the changes for my computer specific information?
Dspeed loss?
Fyes slower training
Dah you mean loss of performance. was thinking of loss functions
Fi am preparing very best one
Xflux vs grok?