Software Engineering Courses (SECourses)•15mo ago

I'm training a lora tree people 2 same class 1 different class with regularization on flux-dev-de-di

I'm training a lora tree people 2 same class 1 different class with regularization on flux-dev-de-distill, lets see if regularization images works with flux-dev-de-distill

dsienraOP•10/12/24, 5:21 PM

the regularization are person class

dsienraOP•10/12/24, 5:22 PM

the bleeding between subjects was fixed with flux-dev-de-distill but idk yet if works with regularization

dsienraOP•10/12/24, 5:36 PM

@Dr. Furkan Gözükara flux-dev-de-distill what I saw was that at 30 epoch's the resemblances was reduced with regularization images, lets see if a full 160 epochs fixes that.

Ddsienra @Dr. Furkan Gözükara flux-dev-de-distill what I saw was that at 30 epoch's the r...

druhl•10/12/24, 5:47 PM

How many images are there in your dataset?

Ddsienra I'm training a lora tree people 2 same class 1 different class with regularizati...

druhl•10/12/24, 5:48 PM

What tool do you use to make this curve? I don't let logs save at all lol. Perhaps I should reconsider.

dsienraOP•10/12/24, 5:52 PM

just tensorboard on kohya, 10 or 11 pictures per subject, I have a folder with real people as regularization, 2000 images different ages, genders and races

dsienraOP•10/12/24, 5:53 PM

I downloaded the regularization dataset a time ago with captions

Ddsienra just tensorboard on kohya, 10 or 11 pictures per subject, I have a folder with r...

druhl•10/12/24, 5:53 PM

Awesome!

druhl•10/12/24, 5:56 PM

For me, that option says start even though the training is on

@dsienra

dsienraOP•10/12/24, 5:58 PM

just press it

dsienraOP•10/12/24, 5:58 PM

it will start tensorboard

dsienraOP•10/12/24, 5:58 PM

is independent of trainig

dsienraOP•10/12/24, 6:00 PM

dsienraOP•10/12/24, 6:01 PM

just select the current training in the bottom or enable all to compare the loss curve with past trainings

dsienraOP•10/12/24, 6:01 PM

is just a guide

dsienraOP•10/12/24, 6:07 PM

in theory it should go down until starts overtraining and in that case it should go up again, when the training just starts the loos goes up and down, that is normal on the first epochs, but overall the graphic should go down during the entire training

deepblue_1337•10/12/24, 7:54 PM

Hello,

I am trying to follow the Patreon post on finetuning.
https://www.patreon.com/posts/huge-news-for-as-113531833

Patreon

Huge news for Kohya GUI - Now you can fully Fine Tune / DreamBooth ...

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

deepblue_1337•10/12/24, 7:55 PM

it says same process as lora, only different config. Do I still use the LoRa tab in the webinterface or do I use dreambooth?

Thanks
Daniel

Ddeepblue_1337 it says same process as lora, only different config. Do I still use the LoRa tab...

druhl•10/12/24, 7:59 PM

Do dreambooth

deepblue_1337•10/12/24, 8:01 PM

I have a RTX 3060 and am getting ~45sec/it, which would take quite long for 9200 steps, is that normal?

deepblue_1337•10/12/24, 8:02 PM

nvitop is showing >90% GPU utilization

Ddsienra @Dr. Furkan Gözükara flux-dev-de-distill what I saw was that at 30 epoch's the r...

Furkan Gözükara SECourses•10/12/24, 9:56 PM

nice test

Ddeepblue_1337 it says same process as lora, only different config. Do I still use the LoRa tab...

Furkan Gözükara SECourses•10/12/24, 9:57 PM

dreambooth tab

Ddeepblue_1337 I have a RTX 3060 and am getting ~45sec/it, which would take quite long for 9200...

Furkan Gözükara SECourses•10/12/24, 9:57 PM

yes normal

Furkan Gözükara SECourses•10/12/24, 9:57 PM

rtx 3060 takes around 40 second / it in long run

FFurkan Gözükara SECourses nice test

dsienraOP•10/12/24, 10:04 PM

The training was finished, resemblance is almost perfect, but it still looks a little undertrained, with regularization images it needs much more steps or increase the learning rate to get resemblance. I will resume the training from 160 repetitions per image and I will report back, there was no bleeding between the trained subjects

Ddsienra The training was finished, resemblance is almost perfect, but it still looks a l...

Furkan Gözükara SECourses•10/12/24, 10:05 PM

wow awesome

Furkan Gözükara SECourses•10/12/24, 10:06 PM

i am waiting that model to become more mature

Furkan Gözükara SECourses•10/12/24, 10:06 PM

will fully test it and compare

FFurkan Gözükara SECourses will fully test it and compare

dsienraOP•10/12/24, 10:27 PM

Nice, If you have to train more than one subject it is your only choice, in my tests is superior on everything, the only problem is if you are going to do a full training the inference of the resulting model will be slower because you have to set the cfg scale to 3.5, to overcome it what can be done is to do the finetune, extract the lora and then use the resulting lora on regular flux-dev and be able to use cfg scale 1 in conjunction with distilled cfg

Ddsienra Nice, If you have to train more than one subject it is your only choice, in my t...

Furkan Gözükara SECourses•10/12/24, 10:30 PM

true

Furkan Gözükara SECourses•10/12/24, 10:30 PM

once you use cfg > 1 = it doubles the duration of generation

FFurkan Gözükara SECourses once you use cfg > 1 = it doubles the duration of generation

dsienraOP•10/12/24, 10:36 PM

the quality of an extracted lora from a finetune is comparable to a finetune?

Ddsienra the quality of an extracted lora from a finetune is comparable to a finetune?

Furkan Gözükara SECourses•10/12/24, 10:37 PM

almost

Furkan Gözükara SECourses•10/12/24, 10:37 PM

i made a comparison

Furkan Gözükara SECourses•10/12/24, 10:37 PM

but i doubt it would work on

Furkan Gözükara SECourses•10/12/24, 10:37 PM

dev model

FFurkan Gözükara SECourses dev model

dsienraOP•10/12/24, 10:39 PM

the loras I'm training on flux-dev-de-distill works perfect on regular flux-dev, why the extracted lora will not work?

dsienraOP•10/12/24, 10:41 PM

not in the other way around the loras trained on regular flux-dev lost resemblance on flux-dev-de-distill on inference

Ddsienra the loras I'm training on flux-dev-de-distill works perfect on regular flux-dev,...

Furkan Gözükara SECourses•10/12/24, 10:42 PM

wow if working then yes

Furkan Gözükara SECourses•10/12/24, 10:42 PM

should work

Ddsienra the loras I'm training on flux-dev-de-distill works perfect on regular flux-dev,...

Furkan Gözükara SECourses•10/12/24, 10:42 PM

have you compared grid side by side?

dsienraOP•10/12/24, 10:43 PM

no I just created some pictures to test

dsienraOP•10/12/24, 10:47 PM

I resumed the training from the last lora checkpoint

dsienraOP•10/12/24, 10:55 PM

@Dr. Furkan Gözükara on flux-dev-de-distill to get resemblance with regularization I used a much higher lr 0.004 on 160 epochs looks still a little undertrained, my lora rank and alpha is set like this, this can be a problem

Ddsienra @Dr. Furkan Gözükara on flux-dev-de-distill to get resemblance with regularizati...

Furkan Gözükara SECourses•10/12/24, 10:55 PM

yep

Furkan Gözükara SECourses•10/12/24, 10:55 PM

you basically changed the LR i researched

Furkan Gözükara SECourses•10/12/24, 10:55 PM

reduced it like 1/8

Furkan Gözükara SECourses•10/12/24, 10:55 PM

i am offline

FFurkan Gözükara SECourses yep

dsienraOP•10/12/24, 10:57 PM

ohhhh, now I understands

I'm training a lora tree people 2 same class 1 different class with regularization on flux-dev-de-di

Similar Threads