Software Engineering Courses (SECourses)•15mo ago

I Installed the "Kohya_GUI_Flux_Installer_v40" script which works perfectly for flux. However I also

I Installed the "Kohya_GUI_Flux_Installer_v40" script which works perfectly for flux.
However I also want to train an sdxl model.
When trying to do so I get the error "ModuleNotFoundError: No module named 'xformers'".

Is pip install xformers the correct way?

Noel Camus•10/15/24, 2:36 PM

dreambooth fine-tuning should i choose float? i remember lora is float.

Ddeepblue_1337 I Installed the "Kohya_GUI_Flux_Installer_v40" script which works perfectly for ...

Furkan Gözükara SECourses•10/15/24, 2:38 PM

currently xformers not supported on torch 2.5

Furkan Gözükara SECourses•10/15/24, 2:38 PM

so make a seperate install and dont run torch 2.5

NNoel Camus dreambooth fine-tuning should i choose float? i remember lora is float.

Furkan Gözükara SECourses•10/15/24, 2:39 PM

float is best

Furkan Gözükara SECourses•10/15/24, 2:39 PM

but fp16 also works decent

Zet•10/15/24, 2:40 PM

No matter what Rank config I use, I run out of VRAM

Lora Training/16GB Vram, using Rank 5
Idle 0.5/16GB used
Are there any updates in branch sd3-flux.1 that are not accounted for in the .json files? Settings wise? I must be missing something

Noel Camus•10/15/24, 2:42 PM

thanks

ZZet No matter what Rank config I use, I run out of VRAM 😦 Lora Training/16GB Vram, ...

Furkan Gözükara SECourses•10/15/24, 2:42 PM

error is easy

Furkan Gözükara SECourses•10/15/24, 2:42 PM

you are loading into dreambooth

Furkan Gözükara SECourses•10/15/24, 2:42 PM

or at least once loaded

Furkan Gözükara SECourses•10/15/24, 2:42 PM

get a fresh config from zip

Furkan Gözükara SECourses•10/15/24, 2:43 PM

load into lora

Zet•10/15/24, 2:43 PM

I don't think so, it's flux_network script

Zet•10/15/24, 2:43 PM

Zet•10/15/24, 2:43 PM

ZZet Click to see attachment

Furkan Gözükara SECourses•10/15/24, 2:44 PM

if you loaded once it becomes corrupted

Zet•10/15/24, 2:44 PM

ok, let me start fresh

Noel Camus•10/15/24, 2:47 PM

im trying to train a gaming setup lora (use dreambooth extract) i have 500+ images should i decrease images?

NNoel Camus im trying to train a gaming setup lora (use dreambooth extract) i have 500+ ima...

Furkan Gözükara SECourses•10/15/24, 2:47 PM

well it depends on purpose

Furkan Gözükara SECourses•10/15/24, 2:48 PM

if you read latest work 256 images worked very well

Furkan Gözükara SECourses•10/15/24, 2:48 PM

for style training more better usually

FFurkan Gözükara SECourses if you loaded once it becomes corrupted

Zet•10/15/24, 2:53 PM

Same result with fresh .json... and it's way over

ZZet Same result with fresh .json... and it's way over

Furkan Gözükara SECourses•10/15/24, 2:54 PM

please message me your json file

Furkan Gözükara SECourses•10/15/24, 2:54 PM

from private message

Furkan Gözükara SECourses•10/15/24, 2:54 PM

also show your vram before starting training

Furkan Gözükara SECourses•10/15/24, 2:58 PM

@Zet your config is dreambooth corrupted config

Furkan Gözükara SECourses•10/15/24, 2:58 PM

why you dont follow what i say

deepblue_1337OP•10/15/24, 3:29 PM

Fresh install of Kohya_GUI_Flux_Installer_v40.zip and Kohya_FLUX_DreamBooth_v7.zip both yield:

ERROR: Cannot install -r requirements.txt (line 1), -r requirements.txt (line 10), diffusers[torch]==0.25.0 and huggingface-hub==0.24.5 because these package versions have conflicting dependencies.

Ddeepblue_1337 Fresh install of Kohya_GUI_Flux_Installer_v40.zip and Kohya_FLUX_DreamBooth_v7.z...

Zet•10/15/24, 3:46 PM

change requirements.txt huggingface-hub==0.24.5 to huggingface-hub==0.25.2

FFurkan Gözükara SECourses why you dont follow what i say

Zet•10/15/24, 3:49 PM

I am, or at least I am trying my very best

Ddeepblue_1337 Fresh install of Kohya_GUI_Flux_Installer_v40.zip and Kohya_FLUX_DreamBooth_v7.z...

Furkan Gözükara SECourses•10/15/24, 3:51 PM

this looks like bmaltasi error

Ddeepblue_1337 Fresh install of Kohya_GUI_Flux_Installer_v40.zip and Kohya_FLUX_DreamBooth_v7.z...

Furkan Gözükara SECourses•10/15/24, 3:51 PM

we need to report this

FFurkan Gözükara SECourses we need to report this

Zet•10/15/24, 3:56 PM

Yes, some of the newer packages require huggingface-hub>=0.25. And careful, 0.26 removes cached_download

Zet•10/15/24, 3:56 PM

that will need attention as well

Furkan Gözükara SECourses•10/15/24, 3:58 PM

i am making a fresh install to test

dsienra•10/15/24, 4:05 PM

@Dr. Furkan Gözükara New test Flux-De-Distilled with this settings the lora at 160 epochs looks severely undertrained, may be Flux-De-Distilled requieres higher LR

dsienra•10/15/24, 4:06 PM

I test it on the de-distilled model and on regular Flux-Dev

dsienra•10/15/24, 4:08 PM

I'm going to resume the training with LR 0.0001

dsienra•10/15/24, 4:08 PM

and see if it converges

Ddsienra @Dr. Furkan Gözükara New test Flux-De-Distilled with this settings the lora at 1...

Furkan Gözükara SECourses•10/15/24, 4:15 PM

totally could be

Furkan Gözükara SECourses•10/15/24, 4:15 PM

i will research it later hopefully

Furkan Gözükara SECourses•10/15/24, 4:15 PM

i repoırted erroır to kohya : https://github.com/bmaltais/kohya_ss/issues/2901

GitHub

SD3 branch ERROR: Cannot install -r requirements.txt (line 1), -r r...

SD3 flux branch is broken when doing fresh install @kohya-ss @bmaltais Collecting wcwidth>=0.2.5 Using cached wcwidth-0.2.13-py2.py3-none-any.whl (34 kB) INFO: pip is looking at multiple version...

FFurkan Gözükara SECourses totally could be

dsienra•10/15/24, 4:18 PM

I gonna resume from 160 epochs at 0.001 for 10 epochs and test if get signs of converging, I will report

Furkan Gözükara SECourses•10/15/24, 4:53 PM

guys dont use 24gb massed compute instances - or get at least 2x GPU for bigger RAM

Furkan Gözükara SECourses•10/15/24, 4:53 PM

Vigilence•10/15/24, 5:43 PM

How to know what the vram is when selecting a gpu? Its not shown for most gpus.

VVigilence How to know what the vram is when selecting a gpu? Its not shown for most gpus.

Furkan Gözükara SECourses•10/15/24, 5:57 PM

best way is searching google

Ddsienra I gonna resume from 160 epochs at 0.001 for 10 epochs and test if get signs of c...

dsienra•10/15/24, 6:00 PM

@Dr. Furkan Gözükara now with lr at 0.0001 and 10 more epochs can see some resemblance with lora weight at 1.5, so is learning

Elwii•10/15/24, 7:19 PM

General question on dreambooth finetuning: When switching from 15 img to 150 img, do you still train as many steps as with 15 img or is it 10x more?

EElwii General question on dreambooth finetuning: When switching from 15 img to 150 img...

Furkan Gözükara SECourses•10/15/24, 7:37 PM

sadly no formula

I Installed the "Kohya_GUI_Flux_Installer_v40" script which works perfectly for flux. However I also

Similar Threads

Similar Threads

Similar Threads