Software Engineering Courses (SECourses)•17mo ago

@Dr. Furkan Gözükara i think your kohya installer is broken. im now getting windows os errors, and

@Dr. Furkan Gözükara i think your kohya installer is broken.

im now getting windows os errors, and gui itself doesn't want to open.

MMaxivi @Dr. Furkan Gözükara i think your kohya installer is broken. im now getting win...

Furkan Gözükara SECourses•8/28/24, 2:40 PM

can you tell me error

Ppat._.fenis I am batch upscaling individual images that comprise video, have there been any ...

Furkan Gözükara SECourses•8/28/24, 2:40 PM

sadly not but i feel you

Furkan Gözükara SECourses•8/28/24, 2:41 PM

i wonder if pytorch 2.5 can speed it up

pat._.fenis•8/28/24, 2:43 PM

It shows I am only using 7.2gb of my GPU ram

pat._.fenis•8/28/24, 2:44 PM

also states gradio 4.29.0 is available and it is using 4.19 during startup selection

FFurkan Gözükara SECourses can you tell me error

MaxiviOP•8/28/24, 2:44 PM

well im not getting the win os error now because the gui.bat is not doing anything anymore

MaxiviOP•8/28/24, 2:45 PM

its just black screen

MaxiviOP•8/28/24, 2:45 PM

i was getting permission errors as well

MMaxivi well im not getting the win os error now because the gui.bat is not doing anythi...

Furkan Gözükara SECourses•8/28/24, 2:46 PM

are you on windows 11?

MaxiviOP•8/28/24, 2:46 PM

yessir

Furkan Gözükara SECourses•8/28/24, 2:47 PM

windows 11 sucks

Furkan Gözükara SECourses•8/28/24, 2:47 PM

run gui.bat inside kohya ss folder and tell me waht you see

Furkan Gözükara SECourses•8/28/24, 2:47 PM

restart computer and try again too

MaxiviOP•8/28/24, 2:47 PM

i know i had to upgrade though, windows 10 was giving so much issues, random pc shutdowns, slowdowns etc. plus windows 10 soon wont be supported anymore

FFurkan Gözükara SECourses restart computer and try again too

MaxiviOP•8/28/24, 2:48 PM

already tried i was getting just a blank terminal

MaxiviOP•8/28/24, 2:48 PM

but im doing system restore right now back to 4 days ago

Furkan Gözükara SECourses•8/28/24, 2:48 PM

did you get any error at install?

Furkan Gözükara SECourses•8/28/24, 2:48 PM

windows 11 has problematic permissions with python apps :/

MaxiviOP•8/28/24, 2:48 PM

yeah i was getting some windows os error

MMaxivi yeah i was getting some windows os error

Furkan Gözükara SECourses•8/28/24, 2:48 PM

you need to fix it

Furkan Gözükara SECourses•8/28/24, 2:48 PM

it is permission related error

MaxiviOP•8/28/24, 2:49 PM

https://tenor.com/view/sad-gif-19231393

Tenor

Furkan Gözükara SECourses•8/28/24, 2:50 PM

i still use windows 10

but i also get random shutdowns and bluescreens sometimes :/

pixelscience•8/28/24, 2:52 PM

Hey there, thank you for the plethora of information, I'm learning a ton from all the work you're putting into this.

I'm trying to understand how Koyha is calculating the total optimization steps when I run my LORA training...

I have 30 images set to 10 repeats, so that part is clear, but it somehow seems to be automatically configuring things to hit 1600 steps.

running training / 学習開始
  num train images * repeats / 学習画像の数×繰り返し回数: 300
  num reg images / 正則化画像の数: 0
  num batches per epoch / 1epochのバッチ数: 300
  num epochs / epoch数: 6
  batch size per device / バッチサイズ: 1
  gradient accumulation steps / 勾配を合計するステップ数 = 1
  total optimization steps / 学習ステップ数: 1600

running training / 学習開始
  num train images * repeats / 学習画像の数×繰り返し回数: 300
  num reg images / 正則化画像の数: 0
  num batches per epoch / 1epochのバッチ数: 300
  num epochs / epoch数: 6
  batch size per device / バッチサイズ: 1
  gradient accumulation steps / 勾配を合計するステップ数 = 1
  total optimization steps / 学習ステップ数: 1600

The Epoch setting in Koyha doesn't seem to actually change anything it looks like, so where does it come up with 6 epochs?

dsienra•8/28/24, 3:06 PM

I'm doing new tests with TE training multiple people but it seems that still bleeds to the class, may be for flux you have to remove the class completely to avoid bleeding the people you are training between each other, and the bleed to the entire class "man" for example, may be changing the captions from "ohwx man" to only "ohwx" will do the magic, flux already knows what a man is and is not necessary to add it to the captions, In sdxl this wasn't a problem for me. the model learned that "name1 man" was different from "name2 man" in flux I didn't find a solution yet, I will try removing the class in my next training.

Ppixelscience Hey there, thank you for the plethora of information, I'm learning a ton from al...

Furkan Gözükara SECourses•8/28/24, 3:06 PM

i explained in the video

Furkan Gözükara SECourses•8/28/24, 3:06 PM

you must be setting somwhere epoch

Furkan Gözükara SECourses•8/28/24, 3:06 PM

or max epoch

Furkan Gözükara SECourses•8/28/24, 3:07 PM

still ediitng video to publish

Ddsienra I'm doing new tests with TE training multiple people but it seems that still ble...

Furkan Gözükara SECourses•8/28/24, 3:07 PM

yes told this in video

Furkan Gözükara SECourses•8/28/24, 3:07 PM

dont use class

Furkan Gözükara SECourses•8/28/24, 3:07 PM

it will work better

Furkan Gözükara SECourses•8/28/24, 3:07 PM

i tested just ohwx man and since its internally already captioning

Furkan Gözükara SECourses•8/28/24, 3:07 PM

it works

Furkan Gözükara SECourses•8/28/24, 3:07 PM

still will likely to bleed but lesser

FFurkan Gözükara SECourses i explained in the video

pixelscience•8/28/24, 3:07 PM

ahh, ok. I'll look into it more, thank you.

Ppixelscience ahh, ok. I'll look into it more, thank you.

Furkan Gözükara SECourses•8/28/24, 3:07 PM

once video published please watch fully

pixelscience•8/28/24, 3:07 PM

will do!

FFurkan Gözükara SECourses once video published please watch fully 😄

dsienra•8/28/24, 3:08 PM

off course I'm waiting... thanks

dsienra•8/28/24, 3:32 PM

the speed on WSL is inconsistent yesterday speed

dsienra•8/28/24, 3:32 PM

today speed

dsienra•8/28/24, 3:33 PM

rtx 3090

dsienra•8/28/24, 3:34 PM

4.16 it/s second is amazing, today 6.3 it/s

dsienra•8/28/24, 3:34 PM

yesterday just fixed it restarting the pc

dsienra•8/28/24, 3:35 PM

same config

Ddsienra today speed

Furkan Gözükara SECourses•8/28/24, 3:37 PM

lol this is weird

Furkan Gözükara SECourses•8/28/24, 3:37 PM

by the way i just tested and

Furkan Gözükara SECourses•8/28/24, 3:37 PM

fast 24 gb is 5.5 second it on windows

Furkan Gözükara SECourses•8/28/24, 3:37 PM

with torch 2.5

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

without it is like 7.7 second it

running training / 学習開始 num train images * repeats / 学習画像の数×繰り返し回数: 300 num reg images / 正則化画像の数: 0 num batches per epoch / 1epochのバッチ数: 300 num epochs / epoch数: 6 batch size per device / バッチサイズ: 1 gradient accumulation steps / 勾配を合計するステップ数 = 1 total optimization steps / 学習ステップ数: 1600

@Dr. Furkan Gözükara i think your kohya installer is broken. im now getting windows os errors, and

Similar Threads

Similar Threads

Similar Threads