Software Engineering Courses (SECourses)•17mo ago

disabeling t5 att masking saves 2 seconds on it. Rank 5

Nnovsky disabeling t5 att masking saves 2 seconds on it. Rank 5

Furkan Gözükara SECourses•8/28/24, 1:27 PM

yep

Furkan Gözükara SECourses•8/28/24, 1:27 PM

also i am about to put

Furkan Gözükara SECourses•8/28/24, 1:27 PM

torch 2.5 installer

Furkan Gözükara SECourses•8/28/24, 1:27 PM

it will give huge boost

Furkan Gözükara SECourses•8/28/24, 1:27 PM

let me update

geddon•8/28/24, 1:44 PM

I documented all my configurations when I was using Kohya_ss. Can these be ported directly into OneTrainer for LoRA model training?

Ggeddon I documented all my configurations when I was using Kohya_ss. Can these be porte...

Furkan Gözükara SECourses•8/28/24, 1:47 PM

very likely

Furkan Gözükara SECourses•8/28/24, 1:47 PM

i did for sdxl

nicearoni•8/28/24, 1:55 PM

does this mean i missed updating something? thought text encoder was supposed to work now?

024-08-28 09:54:18 INFO text_encoder is not needed for training. deleting to save memory. train_network.py:1013
INFO unet dtype: torch.float8_e4m3fn, device: cuda:0 train_network.py:1030

Nnicearoni does this mean i missed updating something? thought text encoder was supposed to...

Furkan Gözükara SECourses•8/28/24, 1:56 PM

i didnt update text encoder configs yet

Furkan Gözükara SECourses•8/28/24, 1:56 PM

i am still researching

Furkan Gözükara SECourses•8/28/24, 1:56 PM

will announce once made update

Furkan Gözükara SECourses•8/28/24, 1:56 PM

but you can copy pate unet LR to text encoder

Furkan Gözükara SECourses•8/28/24, 1:56 PM

and it works

Furkan Gözükara SECourses•8/28/24, 1:57 PM

are you using latest speed boost?=

Furkan Gözükara SECourses•8/28/24, 1:57 PM

we need to verify it works :d

nicearoni•8/28/24, 2:01 PM

i am, but testing something at 512 so can't compare speed to my 1024 runs

Nnicearoni i am, but testing something at 512 so can't compare speed to my 1024 runs

Furkan Gözükara SECourses•8/28/24, 2:01 PM

FFurkan Gözükara SECourses it will give huge boost

novskyOP•8/28/24, 3:05 PM

rank 5 with updated torch (I guess updated, but I did update)...1024, without t5 att masking

Nnovsky rank 5 with updated torch (I guess updated, but I did update)...1024, without t5...

Furkan Gözükara SECourses•8/28/24, 3:08 PM

nicer

Furkan Gözükara SECourses•8/28/24, 3:08 PM

i hope it works perfect

Furkan Gözükara SECourses•8/28/24, 3:08 PM

i will be able to test only during night

novskyOP•8/28/24, 3:10 PM

I will not finish this training, so I can't tell you result. But maybe I will try to start training at night. So could be some results tomorrow...or not)

Nnovsky I will not finish this training, so I can't tell you result. But maybe I will tr...

Furkan Gözükara SECourses•8/28/24, 3:38 PM

i will do same :d

macba•8/28/24, 7:43 PM

Hello everyone, quick question: what does this script do?

https://huggingface.co/MonsterMMORPG/Generative-AI/resolve/main/kohua_gui_pass_update.sh

As far as I understand, it replaces kohya_ss's gui.shgui.shwhen you use Furkan's kohya installer script (at least on Linux). What's the advantage? Thanks…

daroth•8/28/24, 7:45 PM

okay, lets see the local flux kohya training

Mmacba Hello everyone, quick question: what does this script do? https://huggingface.c...

nicearoni•8/28/24, 7:46 PM

I think updates a few libraries to newer versions, then patches original gui to not roll them back? not sure though.

daroth•8/28/24, 7:46 PM

do we still need to use the kohya training folder structure? even with the folder name #_name_class?

Ddaroth do we still need to use the kohya training folder structure? even with the folde...

nicearoni•8/28/24, 7:47 PM

that's what kohya is still expecting as of now (space between name and class though, not _ )

FFurkan Gözükara SECourses nicer

sergio_gonzo•8/28/24, 9:06 PM

Dr Furkan i hope everyting is ok, wich is your estimate tha could be a video tutorial of FLUX Lora step by steps? wonderfull work

Mmacba Hello everyone, quick question: what does this script do? https://huggingface.c...

Furkan Gözükara SECourses•8/28/24, 10:57 PM

it prevents kohya to revert back requirements

Furkan Gözükara SECourses•8/28/24, 10:58 PM

so our custom torch installation remains

Ssergio_gonzo Dr Furkan i hope everyting is ok, wich is your estimate tha could be a video tut...

Furkan Gözükara SECourses•8/28/24, 10:58 PM

thank you. still editing video

Evivrus•8/29/24, 3:02 AM

Has anyone gotten a slider Lora training working for Flux yet? I have seen a few sliders on Civit but I don't think they are trained the same way SD sliders were trained.

Anyone seen anything? I tried modifying the ai-toolkit code to get it to work but it didn't work out ;P

Furkan Gözükara SECourses•8/29/24, 3:49 AM

nope i didnt see yet

daroth•8/29/24, 9:02 AM

when the trained person should be blonde but her last name is "Blue"