Software Engineering Courses (SECourses)•15mo ago

@Dr. Furkan Gözükara So I've been checking out the Batch size 7 config. And It seems that your Learn

@Dr. Furkan Gözükara So I've been checking out the Batch size 7 config. And It seems that your Learning Rate is 0.00001 which is 2.5x higher than when batch size is 1. What made you choose that Learning rate?

Noel Camus•10/17/24, 12:20 AM

I used the "flux-dev-de-distill" model for Dreambooth training, and the resulting file is very large. Is this because I haven't extracted the LoRA from it yet?

AIGambino•10/17/24, 12:48 AM

I assume most everyone here has watched the Behind the scene from Corridor Digital on their rock paper scissors Anime. If not here it is.. https://youtu.be/_9LX9HSQkWo?feature=shared&t=675
Does anyone know about custom models in SDXL? I am curious.. When training their model, does he mean he trained a model on him and his costume, then used that model as the base to train the style into that same model?

Dazzastrous [4090]•10/17/24, 1:15 AM

i want 190gb

VVinch @Dr. Furkan Gözükara So I've been checking out the Batch size 7 config. And It s...

Furkan Gözükara SECourses•10/17/24, 8:55 AM

fully testing

NNoel Camus I used the "flux-dev-de-distill" model for Dreambooth training, and the resultin...

Furkan Gözükara SECourses•10/17/24, 8:55 AM

yes full checkpoints

Furkan Gözükara SECourses•10/17/24, 8:55 AM

but sizes are different

Furkan Gözükara SECourses•10/17/24, 8:55 AM

weird

AAIGambino I assume most everyone here has watched the Behind the scene from Corridor Digit...

Furkan Gözükara SECourses•10/17/24, 8:56 AM

yes i explained what they did here https://youtu.be/kmT-z2lqEPQ

YouTubeSECourses

Video To Anime - Generate An EPIC Animation From Your Phone Recordi...

RunPod: https://bit.ly/RunPodIO. Discord : https://bit.ly/SECoursesDiscord. Turn your real video into an animation with just 1 click by using Stable Diffusion for free. If I have been of assistance to you and you would like to show your support for my work, please consider becoming a patron on

https://www.patreon.com/SECourses

Playlist of Sta...

Video To Anime - Generate An EPIC Animation From Your Phone Recordi...

ar.1•10/17/24, 10:46 AM

can I do multi GPU for FLUX dev dreambooth/finetune ?

ar.1•10/17/24, 11:18 AM

@Dr. Furkan Gözükara how long did it take to train that 10k images model you did ? and what hardware ?

Aar.1 can I do multi GPU for FLUX dev dreambooth/finetune ?

Furkan Gözükara SECourses•10/17/24, 12:58 PM

yes 80 gb gpus

Aar.1 @Dr. Furkan Gözükara how long did it take to train that 10k images model you did...

Furkan Gözükara SECourses•10/17/24, 12:58 PM

i have shared all info on patreon timings too

Furkan Gözükara SECourses•10/17/24, 12:58 PM

please read

Furkan Gözükara SECourses•10/17/24, 1:19 PM

time record fine tuning tutorial

Furkan Gözükara SECourses•10/17/24, 1:19 PM

daily life stuff taking huge time :d

Furkan Gözükara SECourses•10/17/24, 2:59 PM

wow windows flux training speed amazing

Furkan Gözükara SECourses•10/17/24, 2:59 PM

almost alike linux

Maxivi•10/17/24, 3:00 PM

@Dr. Furkan Gözükara https://pypi.org/project/torch/

PyPI

torch

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Maxivi•10/17/24, 3:00 PM

torch 2.5 offically out

Furkan Gözükara SECourses•10/17/24, 3:00 PM

added triton

Furkan Gözükara SECourses•10/17/24, 3:01 PM

nice

Furkan Gözükara SECourses•10/17/24, 3:01 PM

now we gonna wait xformers

Furkan Gözükara SECourses•10/17/24, 3:01 PM

i am recording tutoriakl

Furkan Gözükara SECourses•10/17/24, 3:01 PM

triton added : 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

FFurkan Gözükara SECourses triton added : https://www.patreon.com/posts/112099700

druhl•10/17/24, 3:05 PM

This only impacts training on windows, correct?

druhl•10/17/24, 3:05 PM

Will it also impact inference/ generations on windows?

Ddruhl This only impacts training on windows, correct?

Furkan Gözükara SECourses•10/17/24, 3:27 PM

it depends on scripts

druhl•10/17/24, 3:39 PM

How much data does a HF account allow to store?

druhl•10/17/24, 3:41 PM

Think I'll be needing a lot of space tonight.

Ddruhl How much data does a HF account allow to store?

Furkan Gözükara SECourses•10/17/24, 3:42 PM

i think unlimited :d

Furkan Gözükara SECourses•10/17/24, 3:42 PM

they are amazing

Furkan Gözükara SECourses•10/17/24, 3:43 PM

man this fine tuning video being long as usual :d

FFurkan Gözükara SECourses man this fine tuning video being long as usual :d

druhl•10/17/24, 3:45 PM

I thought the loras video covered almost all of the settings?

Ddruhl I thought the loras video covered almost all of the settings?

Furkan Gözükara SECourses•10/17/24, 3:45 PM

yes true did

Furkan Gözükara SECourses•10/17/24, 3:45 PM

but i add some extra new stuff like auto crop

Furkan Gözükara SECourses•10/17/24, 3:45 PM

showing some swarmui settings

druhl•10/17/24, 3:45 PM

Ohh cool.

druhl•10/17/24, 3:47 PM

Can I have SwarmUI find models in my comfy unet folder? lol

druhl•10/17/24, 3:47 PM

I'm running sooo low on space

druhl•10/17/24, 3:47 PM

I did wanna get rid of duplicates of the same model

Furkan Gözükara SECourses•10/17/24, 4:17 PM

huge news

Furkan Gözükara SECourses•10/17/24, 4:17 PM

https://www.reddit.com/r/StableDiffusion/comments/1g5uose/hallo2_highresolution_audiodriven_portrait_image/

From the StableDiffusion community on Reddit: Hallo2 High-Resolutio...

Explore this post and more from the StableDiffusion community

Ddruhl Can I have SwarmUI find models in my comfy unet folder? lol

Furkan Gözükara SECourses•10/17/24, 4:17 PM

sure you can give different folder paths

dsienra•10/17/24, 5:41 PM

@Dr. Furkan Gözükara Update on https://huggingface.co/nyanko7/flux-dev-de-distill Training - "T5 Attention Mask and T5-XXL both disabled" lr 0.0001 When starts to overtrain the subjects start to bleed to each other, up to 80 epochs no bleeding and very good resemblance. when is so overtrained on 200 epochs all subjects get mixed together, For inference works perfect with flux-de-distill and also on regular flux-dev and hyper-flux, on regular flux-dev and hyper-flux the resemblance diminish very little may be improves with lower lr, Now I'm going to train with a much lower lr to avoid overtraining and get finer detail learning, I'll use lr for unet 0.00003 and TE 0.00005 (at inference flux-dev-de-distill cfg 3.5, for flux-dev and hyper-flux cfg 1 and distilled cfg 3.5)

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