Software Engineering Courses (SECourses)•15mo ago

you are making unsubstantiated claims that you know more about how Flux was created than the general

you are making unsubstantiated claims that you know more about how Flux was created than the general public. anyone can read papers. you have to substantiate your claim.

Ddxqb you are making unsubstantiated claims that you know more about how Flux was crea...

crystalwizard•10/6/24, 9:40 PM

i owe you nothing. i direct you, if you want a place to start your research other than arXiv, to all the posts on this discord ALONE where people talk about how many times they've tried to train flux for something and how it doesn't work. even Furkan himself has expressed issues with that. and that is all i'm giving you for free. go do the research

crystalwizard•10/6/24, 9:40 PM

be just as snarky as you like now, i'm done with you

dxqbOP•10/6/24, 9:41 PM

blocked

crystalwizard•10/6/24, 9:44 PM

blocking on discord is funny. you still see that the other person has posted something, but you can't see what. you can't tell if they're talking about you or not. blocking on discord is a good way to drive yourself crazy. and when you're the one that's tried to start an arguement, and then block the person you were trying to argue with because they wouldn't play your game, that says something about you to everyone watching

B0nd•10/6/24, 9:45 PM

Whats the best upscaler model that allows commercial use?

BB0nd Whats the best upscaler model that allows commercial use?

crystalwizard•10/6/24, 9:45 PM

topaz's models

Ccrystalwizard topaz's models

B0nd•10/6/24, 9:46 PM

can they be intergrated into comfyUI?

BB0nd can they be intergrated into comfyUI?

crystalwizard•10/6/24, 9:47 PM

no. they're propiratory and exclusive to topaz, and not free - but they are the best and you can use the output commercially

Ccrystalwizard no. they're propiratory and exclusive to topaz, and not free - but they are the ...

B0nd•10/6/24, 9:47 PM

yeah its not the output commercially im looking to do. Its for a the backend of a SaaS product.

daroth•10/6/24, 9:53 PM

cmon everyone, be nice to each other

MMaxivi @Dr. Furkan Gözükara have you tried reg images with dreambooth?

dsienra•10/6/24, 9:59 PM

@Dr. Furkan Gözükara you can try flux finetuning with regularization and in the last epochs remove the regularization, may be the model get resemblance at the end but stays more flexible

Ddsienra @Dr. Furkan Gözükara you can try flux finetuning with regularization and in the ...

Furkan Gözükara SECourses•10/6/24, 10:01 PM

it is an idea

Furkan Gözükara SECourses•10/6/24, 10:01 PM

may work

Furkan Gözükara SECourses•10/6/24, 10:02 PM

i am testing each gpu config to get max speed

Furkan Gözükara SECourses•10/6/24, 10:02 PM

man this is taking time

dsienra•10/6/24, 11:41 PM

flux-dev-de-distill is giving this error @kohya_tech is already informed and he said that he will fix it as soon a possible.

Furkan Gözükara SECourses•10/7/24, 12:13 AM

wow batch size 7 way faster than before

Furkan Gözükara SECourses•10/7/24, 12:13 AM

kohya did huge improvement

Furkan Gözükara SECourses•10/7/24, 12:14 AM

training takes under 2 hours haha

Furkan Gözükara SECourses•10/7/24, 12:14 AM

3000 steps

Furkan Gözükara SECourses•10/7/24, 12:19 AM

Furkan Gözükara SECourses•10/7/24, 12:20 AM

ah but my current speed test of batch size 7 not accurate

Furkan Gözükara SECourses•10/7/24, 12:20 AM

i need to give bigger

Furkan Gözükara SECourses•10/7/24, 12:20 AM

ye made the images count 256 testing its speed again

Furkan Gözükara SECourses•10/7/24, 12:22 AM

ye for 256 images not that fast

Furkan Gözükara SECourses•10/7/24, 12:22 AM

so i guess best would be testing 14 images

dsienra•10/7/24, 12:24 AM

@crystalwizard You keep claiming that flux-dev can't be trained I was doing many tests with flux-dev-de-distill, I think that it can work. Its behavior on inference is completely different than regular flux-dev it works perfect without distilled cfg, what they did to model probably works, its behavior is similar to sdxl, if you prompt random you get random noise but if you prompt properly you get a perfect image, on regular flux-dev with distilled cfg you always get a coherent image, did you test it?

Furkan Gözükara SECourses•10/7/24, 12:25 AM

14 images 200 epochs taking 3 hours

Furkan Gözükara SECourses•10/7/24, 12:25 AM

batch size 7

Furkan Gözükara SECourses•10/7/24, 12:25 AM

so total 1$

FFurkan Gözükara SECourses so total 1$

dsienra•10/7/24, 12:27 AM

great price!!

Ddsienra @crystalwizard You keep claiming that flux-dev can't be trained I was doing many...

crystalwizard•10/7/24, 12:33 AM

yeah. i spent a solid month testing flux

Ddsienra great price!!

Furkan Gözükara SECourses•10/7/24, 12:36 AM

yep

Furkan Gözükara SECourses•10/7/24, 12:36 AM

new configs

Furkan Gözükara SECourses•10/7/24, 12:36 AM

Ccrystalwizard yeah. i spent a solid month testing flux

dsienra•10/7/24, 12:38 AM

Try this one https://huggingface.co/nyanko7/flux-dev-de-distill its behavior is completely different

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

dsienra•10/7/24, 12:39 AM

they did something to it, may be it surprises you

Ddsienra Try this one https://huggingface.co/nyanko7/flux-dev-de-distill its behavior is ...

crystalwizard•10/7/24, 12:40 AM

i know what they are doing, and it's not what they think they are doing.

crystalwizard•10/7/24, 12:42 AM

plus - it's 100% impossible to put information back that's been removed if you don't know what it is - and they don't have the original weights.

dsienra•10/7/24, 12:42 AM

@Dr. Furkan Gözükara Kohya updated nyanko7/flux-dev-de-distill is training now.

Ccrystalwizard plus - it's 100% impossible to put information back that's been removed if you d...

dsienra•10/7/24, 12:46 AM

I agree you cant get that was deleted back, but it doesn't mean that you can't put new information on its place

Ddsienra I agree you cant get that was deleted back, but it doesn't mean that you can't ...

crystalwizard•10/7/24, 12:46 AM

but that doesn't undo what was done, does it?

dsienra•10/7/24, 12:47 AM

I agree

crystalwizard•10/7/24, 12:47 AM

if you have a bag of chocolate chips, i take them out, you have an empty bag. if i then put potato chips in it, do you have a bag of choclate chips again?

crystalwizard•10/7/24, 12:47 AM

and even if i put more chocolate chips in it, do you ahve the same chips you had?

dsienra•10/7/24, 12:48 AM

We want a functional model I dont care if it is identical to the original if it works

dsienra•10/7/24, 12:48 AM

flux pro is private and we will never have it

crystalwizard•10/7/24, 12:49 AM

you'd be better off just using SDXL - what was done to flux was not only the distill process, but other things as well. those were done to try to fix an issue that the neural network has. that's resulted in flux being rigid, untrainable, and giving very strange results outside of a narrow window.

crystalwizard•10/7/24, 12:49 AM

and the licenses do not allow you to modify flux the way that some people are trying to do

Furkan Gözükara SECourses•10/7/24, 12:50 AM

https://www.patreon.com/posts/112099700

Patreon

Kohya FLUX Fine Tuning (Full Checkpoints) Training Full Tutorial Fo...

Get more from SECourses: FLUX, Tutorials, Guides, Resources, Training, Scripts on Patreon

you are making unsubstantiated claims that you know more about how Flux was created than the general

Similar Threads

Similar Threads

Similar Threads