Software Engineering Courses (SECourses)•13mo ago

<@205854764540362752> question about flux dev fine tune and reg. images. A thought occured to me tha

@Furkan Gözükara SECourses question about flux dev fine tune and reg. images. A thought occured to me that if you were training something which is completely unknown to flux dev, say i.e. a particular alien race which lives on planet X, and your fine tune is a particular character of that race. Is this a situation were regularisation images would make sense? Training imgs would be the character, and reg. images just images of many different kind of aliens in this race. tokens ohwx planetxalien

�🍭🎀 𝒜𝓋𝒶 𝐹𝓇𝒾𝑔𝑔 🎀🍭yes, when using depth/canny you can't use segment in the prompt, you will have t...

Timson•12/21/24, 12:26 PM

the question was: do I have to train separate model based on FLUX.1 Fill checkpoint for that? Or I have to train on the base flux dev then extract lora and apply it to flux fill?

uisato•12/21/24, 2:59 PM

https://www.youtube.com/watch?v=pyNTS4oYGFU

YouTubeuisato

Spectral Analysis - [Short Concept Video for @thevoidz_ ]

For more experiments, head over to: https://linktr.ee/uisato

#animation #ai #cinematic

TTimson <@205854764540362752> can you use FLUX.1 Fill and FLUX.1 Depth with lora / dream...

Furkan Gözükara SECourses•12/21/24, 10:17 PM

i think not working that way yet. but i was able to use redux with fine tuned model

�🍭🎀 𝒜𝓋𝒶 𝐹𝓇𝒾𝑔𝑔 🎀🍭<@205854764540362752> question about flux dev fine tune and reg. images. A thoug...

Furkan Gözükara SECourses•12/21/24, 10:18 PM

reg images just not working so probably wouldnt make sense

Furkan Gözükara SECourses•12/21/24, 10:18 PM

just try to improve dataset

D4UNTLESS•12/22/24, 9:01 AM

How long does the massed compute "Initializing" take? It's been more than 10 minutes now

Timson•12/22/24, 9:32 AM

the bigger initialisation image - the longer it takes

10 minutes is sensible for larger desktop images

TTimson the bigger initialisation image - the longer it takes 🤷‍♂️ 10 minutes is sensib...

D4UNTLESS•12/22/24, 9:59 AM

Thanks! It took around 15 minutes. I have started the training

DD4UNTLESS How long does the massed compute "Initializing" take? It's been more than 10 min...

Furkan Gözükara SECourses•12/22/24, 10:07 AM

last time it took 13 for me

DD4UNTLESS Thanks! It took around 15 minutes. I have started the training 🤞

Furkan Gözükara SECourses•12/22/24, 10:08 AM

after init training takes me like 3 minutes to start including model download

D4UNTLESS•12/22/24, 11:06 AM

DD4UNTLESS Click to see attachment

D4UNTLESS•12/22/24, 11:07 AM

@Furkan Gözükara SECourses

DD4UNTLESS Click to see attachment

Furkan Gözükara SECourses•12/22/24, 11:15 AM

what config

Furkan Gözükara SECourses•12/22/24, 11:15 AM

did you generate samples?

Furkan Gözükara SECourses•12/22/24, 11:16 AM

are your all images 1024x1024?

D4UNTLESS•12/22/24, 11:21 AM

Batch 7 config. Dreambooth training. Images are in different resolutions. I turned on bucketing

DD4UNTLESS Batch 7 config. Dreambooth training. Images are in different resolutions. I turn...

Furkan Gözükara SECourses•12/22/24, 11:23 AM

that is why

Furkan Gözükara SECourses•12/22/24, 11:23 AM

reduce batch size to 6

Furkan Gözükara SECourses•12/22/24, 11:23 AM

batch size 7 barely fits when all 1024x1024 and no bucket

FFurkan Gözükara SECourses did you generate samples?

D4UNTLESS•12/22/24, 11:43 AM

You mentioned in the tutorial to not do sampling. I just followed your tutorial.

FFurkan Gözükara SECourses reduce batch size to 6

D4UNTLESS•12/22/24, 11:43 AM

Can I use batch 1 48gb on A6000? Will it take longer?

Timson•12/22/24, 12:07 PM

Btw, I still don't understand what's the point of batching if you have 100% CUDA cores utilisation on batch size 1?

DD4UNTLESS Can I use batch 1 48gb on A6000? Will it take longer?

Furkan Gözükara SECourses•12/22/24, 2:15 PM

yes you can use

Furkan Gözükara SECourses•12/22/24, 2:15 PM

it will take a little bit longer

TTimson Btw, I still don't understand what's the point of batching if you have 100% CUDA...

Furkan Gözükara SECourses•12/22/24, 2:15 PM

it still slightly improves speed

Furkan Gözükara SECourses•12/22/24, 2:16 PM

but for quality it reduces a little bit

Furkan Gözükara SECourses•12/22/24, 2:16 PM

you can calculate

Furkan Gözükara SECourses•12/22/24, 2:16 PM

batch size 1 = 6.4 second / it = 7*6.4 second for 7 samples

Furkan Gözükara SECourses•12/22/24, 2:16 PM

batch size 7 = 29.5 second / it

Furkan Gözükara SECourses•12/22/24, 2:16 PM

also i recommend now L40 with our coupon now only 49 cents

D4UNTLESS•12/22/24, 3:35 PM

I'm using cropping + padding to bring all my images to 1024x1024
So I'm not gonna use bucketing. Should I go with batch size 6?

DD4UNTLESS I'm using cropping + padding to bring all my images to 1024x1024 So I'm not gonn...

Furkan Gözükara SECourses•12/22/24, 4:16 PM

you shouldnt get

Furkan Gözükara SECourses•12/22/24, 4:16 PM

i tested many times

Furkan Gözükara SECourses•12/22/24, 4:16 PM

batch size 7 should work no bucket @ 1024 px

DD4UNTLESS Click to see attachment

D4UNTLESS•12/22/24, 4:36 PM

So this error was because of different resolutions? If I use all images with 1024x1024 now (no bucketing) I will not get errors at batch size 7? Just need to get confirmation from you

DD4UNTLESS So this error was because of different resolutions? If I use all images with 102...

Furkan Gözükara SECourses•12/22/24, 4:39 PM

yes

Furkan Gözükara SECourses•12/22/24, 4:39 PM

it was using more vram

FFurkan Gözükara SECourses yes

D4UNTLESS•12/22/24, 6:18 PM

ok thanks! I'll try again after a few hours

johnasdf.•12/22/24, 6:54 PM

Has anyone gotten this message when fine-tuning a Flux model? ValueError: mutable default <class 'timm.models.maxxvit.MaxxVitConvCfg'> for field conv_cfg is not allowed: use default_factory

Jjohnasdf.Has anyone gotten this message when fine-tuning a Flux model? ValueError: mutab...

Furkan Gözükara SECourses•12/22/24, 6:58 PM

which model ?

johnasdf.•12/22/24, 6:59 PM

Initially I fine-tuned with AGFlux_v21, a de-distilled model. Now I am trying to take a second pass on it.

johnasdf.•12/22/24, 7:02 PM

ChatGPT thinks it's either timm, Torch, or Kohya

johnasdf.•12/22/24, 7:04 PM

I noticed there are two venvs created when I ran the kohya setup v17... venv, and venv3. Is that what was intended to happen?

Jjohnasdf.Initially I fine-tuned with AGFlux_v21, a de-distilled model. Now I am trying to...

Furkan Gözükara SECourses•12/22/24, 7:06 PM

report to kohya