Okay gotchu!! Wont caption, and will try your config thank you! 🙂
Okay gotchu!! Wont caption, and will try your config thank you! 
Forange hair white beard
![Dazzastrous [4090]](https://cdn.discordapp.com/avatars/871389071686135838/c4a5f4d81d9eeb18d856295d692e2c07.webp?size=40)
Dwhat about mustache
Fi didnt set it :d
Fman deciding quality really hard between different timestep samplers
Dpink mohican
Fhaha added mustache of french i guess
F
Dvery nice
Dhipster look
F
DNow, you've become Doc Phineas from Pawn Stars
DUse that guy as inspiration for fine tuning testing lol.
Fstarting single layer training experiments
Flora
Fsingle block training speed
Freally fast
AWhen you have Quality 1 and 2 in your Flux_Fine_Tune_v2.zip file, does '1' indicate the better quality?
TWow amazing, did you use your classical 15 image dataset or the new one with different expressions
Fclassical 15 images
Fi do experiment on it
Fyes quality 1 better
AI'm about to run dreambooth for flux - I presume that we still are using regularization images, as for SDXL?
Fdoing 8x single layer training
Fno dont use reg images
Fit doesnt work
Aok
AI'll leave the reg folder location blank in kohya.
AI'm using your flux_fine_tune_v2 config.
Fyep good one
Fyes leave blank
Nlet us know how it goes
JI got this warning while doing install Step 2 for Kohya. The bat files you've provided automatically install xformers correct? I just ignore this warning?
L@Dr. Furkan Gözükara do you know anything about Fluxgym, have you tried it or is it in your plans? they say that it works with kohya sd-scripts... the most powerful training script for Stable Diffusion, Flux, etc.
I just saw that information that's why my query
I just saw that information that's why my query
Dhas anyone tried reg images with flux yet?
DI thought Kohya did not have that function, is it released now?
DDoes not work. Doc has tested and posted his experiments.
DIs this the ComfyUI one or the standalone website one?
AMy flux fine tuning is chugging away on a 24gb card. I notice that the s/its value gets a little longer with each epoch. It was at 9 s/it and has crept to 20 s/it by epoch 20. My vram sits at 24 gb with 0.6 gb in regular ram.
KThat flexibility!
KThat's dreambooth finetune?
KDoesn't work
Syou think 15 scar images would be enough to generate lora to draw realistic big cut scar on my face?
Fyep latest one will fix the issues dont worryu
Fi dont plan kohya gui working fine
Fi tried doesnt work
Fdid like 6-7 different trainings
Fpublished yesterday
Ftesting check points right now
Fyes
Fmore usually better