Software Engineering Courses (SECourses)•15mo ago

My training on dev2pro with the same parameters (same concept) would also be completed soon.

Ddruhl My training on de_distill just completed (dreambooth on a single person). The ab...

dsienra•10/18/24, 5:42 PM

great!!!

Ddruhl My training on dev2pro with the same parameters (same concept) would also be com...

dsienra•10/18/24, 5:43 PM

dev2pro was made to train loras and then use them on regular flux I think, is not really de-distilled

dsienra•10/18/24, 5:45 PM

I never try it, but I can try and see if it helps with concept bleeding.

druhlOP•10/18/24, 5:55 PM

It is not truely de-distilled true.

druhlOP•10/18/24, 5:55 PM

I don't like what it does to hands so far lol

Ddruhl It is not truely de-distilled true.

druhlOP•10/18/24, 5:56 PM

With thresholding, it is able to generate nicely at CFG 3.5.

druhlOP•10/18/24, 5:57 PM

It's in a strange place b/w flux dev and de-distill.

FFurkan Gözükara SECourses original vs supir https://imgsli.com/MzA4NzQx/0/2

Neuromasters•10/18/24, 6:02 PM

I still use jegernaught and which models in Super now give the best result for realism for CG graphics? 3D, 2D graphics and drawing for example anime

Ddruhl My training on de_distill just completed (dreambooth on a single person). The ab...

Furkan Gözükara SECourses•10/18/24, 6:44 PM

awesome

NNeuromasters I still use jegernaught and which models in Super now give the best result for r...

Furkan Gözükara SECourses•10/18/24, 6:45 PM

i didnt test for it actually

Furkan Gözükara SECourses•10/18/24, 6:45 PM

now i need to record intro part of the tutorial

Furkan Gözükara SECourses•10/18/24, 7:28 PM

time to edit

Furkan Gözükara SECourses•10/18/24, 7:28 PM

ar.1•10/18/24, 8:13 PM

anyone know if its possible to increase the vocab of Flux Dev ? Like to train the clip models to add new vocab ?

druhlOP•10/18/24, 8:14 PM

Yes, you can train the clip or T5 through captions

Ddruhl Yes, you can train the clip or T5 through captions

ar.1•10/18/24, 8:28 PM

both at the same time ?

Aar.1 anyone know if its possible to increase the vocab of Flux Dev ? Like to train th...

Furkan Gözükara SECourses•10/18/24, 10:02 PM

no way to increase as far as i know. everything still will get tokenized

FFurkan Gözükara SECourses no way to increase as far as i know. everything still will get tokenized

ar.1•10/18/24, 10:02 PM

I think i figured it, I increased it but havent tested it yet

Furkan Gözükara SECourses•10/18/24, 10:03 PM

i turned off fast boot and hibernate

Furkan Gözükara SECourses•10/18/24, 10:04 PM

still getting random BSOD

Furkan Gözükara SECourses•10/18/24, 10:04 PM

lets see if this will fix

wardensc2•10/19/24, 10:04 AM

Hi @Dr. Furkan Gözükara what is the best way to check whether our lora or checkpoint is overfitted

wardensc2•10/19/24, 10:05 AM

whether is not follow the prompt style or body proportion is not normal or something else ?

Wwardensc2 Hi @Dr. Furkan Gözükara what is the best way to check whether our lora or checkp...

Furkan Gözükara SECourses•10/19/24, 10:08 AM

analyzing grid

Furkan Gözükara SECourses•10/19/24, 10:08 AM

in the upcoming video i have explained

Furkan Gözükara SECourses•10/19/24, 10:08 AM

editing right now

Furkan Gözükara SECourses•10/19/24, 10:08 AM

This is why you should get minimum 64 GB RAM memory

FFurkan Gözükara SECourses in the upcoming video i have explained

wardensc2•10/19/24, 10:14 AM

Great Im looking for it

Furkan Gözükara SECourses•10/19/24, 10:24 AM

kohya is working on fixing bleed issue

Furkan Gözükara SECourses•10/19/24, 10:24 AM

https://x.com/kohya_tech/status/1847581169933832521

Kohya Tech (@kohya_tech) on X

Differential Output Preservation(仮)、FLUX.1で学習してみた。prior_loss_weightは10。一応効果ありそうだけど、diffがだいぶ学習不足なだけかもしれない……。

Twitter

•

10/19/24, 10:10 AM

FFurkan Gözükara SECourses https://x.com/kohya_tech/status/1847581169933832521

Maxivi•10/19/24, 10:33 AM

no we have to wait for bmaltais to merge to gui

MMaxivi no we have to wait for bmaltais to merge to gui

Furkan Gözükara SECourses•10/19/24, 10:37 AM

yep

Furkan Gözükara SECourses•10/19/24, 10:38 AM

the videos waiting me to edit :/

Furkan Gözükara SECourses•10/19/24, 10:38 AM

FFurkan Gözükara SECourses yep

Maxivi•10/19/24, 11:00 AM

kohya updated

Maxivi•10/19/24, 11:00 AM

now we wait for bmalt

snowmonkey•10/19/24, 11:51 AM

I'm looking for ComfyUI workflow with Flux Schnell. Anyone got ?

Ssnowmonkey I'm looking for ComfyUI workflow with Flux Schnell. Anyone got ?

druhlOP•10/19/24, 2:37 PM

https://comfyanonymous.github.io/ComfyUI_examples/flux/

ComfyUI_examples

Flux Examples

Examples of ComfyUI workflows

Ddruhl https://comfyanonymous.github.io/ComfyUI_examples/flux/

snowmonkey•10/19/24, 4:33 PM

Thanks!

snowmonkey•10/19/24, 5:06 PM

Does anyone know a good tutorial to use runpod-worker-comfy ? I just want to deploy an API to generate images https://github.com/blib-la/runpod-worker-comfy

GitHub

GitHub - blib-la/runpod-worker-comfy: ComfyUI as a serverless API o...

ComfyUI as a serverless API on RunPod. Contribute to blib-la/runpod-worker-comfy development by creating an account on GitHub.

dsienra•10/19/24, 5:53 PM

Update https://huggingface.co/nyanko7/flux-dev-de-distill Lora training, removing the class did not help with class bleeding, the result was almost the same, and maybe worse because difficulted the prompting on inference. I'm doing another test with just one class and some regularization images, idk if kohya new code was implemented yet, hope this will help too.

nyanko7/flux-dev-de-distill · Hugging Face

Pelolisu•10/19/24, 6:37 PM

New technique I'm introducing for identity preservation.

PPelolisu New technique I'm introducing for identity preservation.

Furkan Gözükara SECourses•10/19/24, 6:43 PM

wow this is good

PPelolisu New technique I'm introducing for identity preservation.

art-ai_usa-girl•10/19/24, 7:35 PM

more information please

Aart-ai_usa-girl more information please

Pelolisu•10/19/24, 7:52 PM

this feature is part of new webui in development, qwen-flux-wan-such This feature is part of a new webui under development, when finished, Dr. will make a tutorial on how to use it. For now I'm in the final stages and presenting some previews here. This feature only uses the image-to-image diffusion pipeline and a controlnet to refine.

FFurkan Gözükara SECourses in the upcoming video i have explained