Software Engineering Courses (SECourses)•3y ago

I learned something about using ControlNet with SDXL. Not every option in ControlNet has an SDXL mo

I learned something about using ControlNet with SDXL. Not every option in ControlNet has an SDXL model. I suspect that is why some say that ControlNet is only partially compatible with SDXL.

DDigital [Starburst]I learned something about using ControlNet with SDXL. Not every option in Contr...

Furkan Gözükara SECourses•11/26/23, 11:44 PM

ah i see

Furkan Gözükara SECourses•11/26/23, 11:44 PM

accurate

Furkan Gözükara SECourses•11/26/23, 11:44 PM

you need to have XL in the name of the selected model

Furkan Gözükara SECourses•11/26/23, 11:44 PM

class images datasets updated : https://civitai.com/articles/2285/massive-4k-resolution-woman-and-man-class-ground-truth-stable-diffusion-regularization-images-dataset

Massive 4K Resolution Woman & Man Class Ground Truth Stable Diffusi...

Download link > https://www.patreon.com/posts/87700469 The best ever released Stable Diffusion classification / regularization images dataset ju...

Zono50•11/27/23, 12:32 AM

What's the best way to upscale human photos to get the best detail?

FFurkan Gözükara SECourses you cna stop training and download checkpoints

DR.Siva•11/27/23, 2:09 AM

Just need a clarification, if we stop training midway, can we continue again?
Also how can we prevent installing from the start for each notebook we create on kaggle or any platform?
In case we can preserve the setup, we can simply go and change, say, the training images, and simply hit train.. Last time on runpod, it was consuming space, so I was loosing money.
I

DR.Siva•11/27/23, 3:13 AM

For those who get Automatic1111 stuck in cmd , here is the solution : https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/2452

TLDR : Just go to properties of CMD and uncheck "Quick Edit Mode"

GitHub

SD command window is just stucked, need to hit Enter to continue · ...

No error when this happen In Windows, i noticed some recently version have this problem, when i set hundred to thousand batch count, it just stucked random at batch xx (sometime at batch 2, sometim...

ZZono50 What's the best way to upscale human photos to get the best detail?

Furkan Gözükara SECourses•11/27/23, 9:38 AM

i would try models from here : https://openmodeldb.info/

OpenModelDB

OpenModelDB is a community driven database of AI Upscaling models. We aim to provide a better way to find and compare models than existing sources.

DDR.Siva Just need a clarification, if we stop training midway, can we continue again? Al...

Furkan Gözükara SECourses•11/27/23, 9:38 AM

you can continue by starting a new training from last checkpoint as a base model

DDR.Siva For those who get Automatic1111 stuck in cmd , here is the solution : https://gi...

Furkan Gözükara SECourses•11/27/23, 9:39 AM

i always mention this in tutorials hit enter if you see cmd stuck

Furkan Gözükara SECourses•11/27/23, 9:40 AM

i got huge results with llava

Furkan Gözükara SECourses•11/27/23, 9:40 AM

when training a style

DR.Siva•11/27/23, 9:43 AM

just did a massive x/y/z plot with various models :

Some lora`s are bad independent of which epoch it is (higher)
Even starting Epochs are good sometimes .
Each models(from civitai) produce different face with different Lora/epoch.

FFurkan Gözükara SECourses you can continue by starting a new training from last checkpoint as a base model

DR.Siva•11/27/23, 9:47 AM

where should we use the checkpoint ? ok , dont want to go into that , may be it will be useful for someone.. , dont want to waste your time.
I and many would love if you can do a detailed explanation on Automatic1111 , because i feel there is loads of potential / tinkering to be done on it.. and also on absolute necessary extension , for eg adetailer ,

DDR.Siva where should we use the checkpoint ? ok , dont want to go into that , may be it ...

Furkan Gözükara SECourses•11/27/23, 9:57 AM

you know we give sdxl 1.0 base

Furkan Gözükara SECourses•11/27/23, 9:57 AM

in the kohya gui

Furkan Gözükara SECourses•11/27/23, 9:57 AM

you give your last checkpoint there

DR.Siva•11/27/23, 10:18 AM

oh ok. so we can just upload as dataset and link it in the code.

DDR.Siva oh ok. so we can just upload as dataset and link it in the code.

Furkan Gözükara SECourses•11/27/23, 10:27 AM

you can give checkpoint path

DR.Siva•11/27/23, 10:27 AM

yes.. understood !!

I am very keen on trying dreambooth..

DR.Siva•11/27/23, 10:28 AM

one question , how can we save the current state of setup ? in kaggle ?

Furkan Gözükara SECourses•11/27/23, 10:28 AM

so many things to make tutorial of

DDR.Siva one question , how can we save the current state of setup ? in kaggle ?

Furkan Gözükara SECourses•11/27/23, 10:28 AM

you save checkpoint and use it

Furkan Gözükara SECourses•11/27/23, 10:28 AM

not state

DR.Siva•11/27/23, 10:29 AM

No i meant , the notebook setup

Furkan Gözükara SECourses•11/27/23, 10:29 AM

yes on there

Furkan Gözükara SECourses•11/27/23, 10:29 AM

give checkpoint file path

Furkan Gözükara SECourses•11/27/23, 10:29 AM

/kaggle/working/mycheckpoint.safetensors

Furkan Gözükara SECourses•11/27/23, 10:29 AM

or in temp that one

DR.Siva•11/27/23, 10:29 AM

till kohya setup , it takes around 30-40 min , if something bad happens , the session is terminated or jupyter is off , all setup is lost

DR.Siva•11/27/23, 10:31 AM

i meant " !bash gui.sh --headless" till this line in code

DR.Siva•11/27/23, 10:31 AM

isnt this when we start using hte gui of kohya

DR.Siva•11/27/23, 10:32 AM

is there a way to save till this state , the notebook ?

DR.Siva•11/27/23, 10:32 AM

so that we dont want to wait everytime 30-40 min to install from first ?

DR.Siva•11/27/23, 10:32 AM

also can i turn on gpu before training ?

DR.Siva•11/27/23, 10:33 AM

or is it needed while installing hte requiremnts to properly config ?

DR.Siva•11/27/23, 10:33 AM

started the first cell now

DR.Siva•11/27/23, 10:33 AM

dont want ot screenshot your notebook codes

DR.Siva•11/27/23, 10:34 AM

for first step itself it will take 15 min i guess.. i thought , if at all we can save the state of notebook at a specific point , it will easy for us to open it again and do necessary changes and start from where we left

DR.Siva•11/27/23, 10:45 AM

ended

DR.Siva•11/27/23, 10:45 AM

roughly 12 min

DR.Siva•11/27/23, 10:48 AM

Running on local URL: http://145645667546745

To create a public link, set share=Trueshare=True in launch()launch().

DR.Siva•11/27/23, 10:48 AM

solid 15 min till we reach gui

DR.Siva•11/27/23, 11:03 AM

again caching latent at 31:35 min , this too will fail i think , may be i will reduce few images from training ?

DR.Siva•11/27/23, 11:32 AM

only 5 min to finish caching

DR.Siva•11/27/23, 11:32 AM

83%|████████████████████████████████▌ | 1516/1818 [30:08<06:00, 1.19s/it]
[E ProcessGroupNCCL.cpp:828] [Rank 1] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=3, OpType=ALLREDUCE, Timeout(ms)=1800000) ran for 1802663 milliseconds before timing out.

DR.Siva•11/27/23, 11:32 AM

DR.Siva•11/27/23, 11:40 AM

reduced to 13 images - 521 steps , lets see

DR.Siva•11/27/23, 11:41 AM

23 min now ..

I learned something about using ControlNet with SDXL. Not every option in ControlNet has an SDXL mo

Similar Threads