Software Engineering Courses (SECourses)•17mo ago

my pod proxy address gave me this, dr

Mattheus ChediakOP•8/20/24, 11:04 AM

do i have to run it again ?

Mattheus ChediakOP•8/20/24, 11:05 AM

i saw some issues on github, and i seems very common

MMattheus Chediak my pod proxy address gave me this, dr

Furkan Gözükara SECourses•8/20/24, 11:31 AM

yep

Ddembel sow Hey everyone! I'm brand new to AI art and trying to wrap my head around all thes...

Furkan Gözükara SECourses•8/20/24, 11:31 AM

you can see this master kohya tutorial

Furkan Gözükara SECourses•8/20/24, 11:31 AM

https://youtu.be/sBFGitIvD2A

YouTubeSECourses

Become A Master Of SDXL Training With Kohya SS LoRAs - Combine Powe...

In this tutorial, you will learn how to install Automatic1111 Web UI for SDXL. How to use LoRAs with Automatic1111 SD Web UI. How to install Kohya SS GUI scripts to do Stable Diffusion training. How to train LoRAs on SDXL model with least amount of VRAM using settings. All of the details, tips and tricks of Kohya trainings. How to do x/y/z plot ...

Cclarkcheekykent since we cant train the text encoder right now, it makes no sense to caption the...

Furkan Gözükara SECourses•8/20/24, 11:32 AM

it still make sense according to the situation

Furkan Gözükara SECourses•8/20/24, 11:32 AM

because when it is training unet it uses those caption vectors

Furkan Gözükara SECourses•8/20/24, 11:32 AM

we cache them

VVAMP https://huggingface.co/InstantX/FLUX.1-dev-Controlnet-Union

Furkan Gözükara SECourses•8/20/24, 11:33 AM

they are working really hard

Furkan Gözükara SECourses•8/20/24, 11:40 AM

rented 12 gb gpu to test 12 gb flux lora training

Furkan Gözükara SECourses•8/20/24, 11:53 AM

testing flux training on rtx 3060

FFurkan Gözükara SECourses testing flux training on rtx 3060 😄

J2_Nedo•8/20/24, 11:58 AM

...i hope you are also thinking of us lonely 4090 users

JJ2_Nedo ...i hope you are also thinking of us lonely 4090 users 😉

Furkan Gözükara SECourses•8/20/24, 12:02 PM

i rent another machine for 24 gb gpus :d

Furkan Gözükara SECourses•8/20/24, 12:03 PM

so on 1 machine 48 gb on 1 machine 24 gb running

Furkan Gözükara SECourses•8/20/24, 12:03 PM

please read here

https://www.patreon.com/posts/kohya-flux-lora-110293257

Patreon

Kohya FLUX LoRA and Fine Tuning Training Full Tutorial For Local Wi...

Get more from SECourses: Tutorials, Guides, Resources, Training, FLUX, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözü on Patreon

Furkan Gözükara SECourses•8/20/24, 12:03 PM

by the way 12 gb still fails i think kohya not properly testing

MMaxivi @Dr. Furkan Gözükara my results with kohya of walking dead star andrew lincoln

sergio_gonzo•8/20/24, 12:13 PM

how do you do for get that clean expresions?

Ssergio_gonzo how do you do for get that clean expresions?

Furkan Gözükara SECourses•8/20/24, 12:27 PM

i will try with my training lets see capability of flux

Furkan Gözükara SECourses•8/20/24, 12:27 PM

uploading newest trainings for testing

virginia•8/20/24, 12:32 PM

@Dr. Furkan Gözükara ifound kohya does not work at the moment, right?

virginia•8/20/24, 12:36 PM

because json cannot be loaded from the local disk

Vvirginia @Dr. Furkan Gözükara ifound kohya does not work at the moment, right?

Furkan Gözükara SECourses•8/20/24, 12:40 PM

it works i am treaining with kohya

�

𝔸𝖨ΞӾ•8/20/24, 12:45 PM

I've finally was able to get Kohya to work thanks to your instructions and with your configs, so thank you @Dr. Furkan Gözükara

�

𝔸𝖨ΞӾ•8/20/24, 12:47 PM

by the way are you running all of your trainings concurrently through one single instance of Kohya? I know you rent the multiple GPUs but can you control them a certain way with their ID like you did with SwarmUI ?

�𝔸𝖨ΞӾ by the way are you running all of your trainings concurrently through one single...

Furkan Gözükara SECourses•8/20/24, 12:59 PM

yes via command line arguments

Furkan Gözükara SECourses•8/20/24, 12:59 PM

export CUDA_VISIBLE_DEVICES=3

Furkan Gözükara SECourses•8/20/24, 12:59 PM

and also in accelerator settings

Furkan Gözükara SECourses•8/20/24, 12:59 PM

--gpu_ids=1

Mattheus ChediakOP•8/20/24, 1:04 PM

https://betterze.github.io/TurboEdit/

TurboEdit

TurboEdit: Instant text-based image editing

FFurkan Gözükara SECourses yes via command line arguments

�

𝔸𝖨ΞӾ•8/20/24, 1:15 PM

That's amazing, thank you Dr.! I'll be testing it out myself soon to see how I can train multiple Loras at once. Would be awesome if you could demonstrate this process in your next Kohya Flux tutorial video

�𝔸𝖨ΞӾ That's amazing, thank you Dr.! I'll be testing it out myself soon to see how I c...

Furkan Gözükara SECourses•8/20/24, 1:48 PM

yes i will show

Furkan Gözükara SECourses•8/20/24, 1:50 PM

Starting new 8x A6000 GPU on Massed Compute

More tests on the way for FLUX ultimate LoRA workflow - for 48 GB, 24 GB and 12 GB GPUs
I should test fine tuning too