when you delete their folder they are
when you delete their folder they are uninstalled
95 Replies
Thank you Furkan, can you please refer me to your tutorial where you teach how to install kohya_ss?
here : https://youtu.be/TpuDOsuKIBo
SECourses
YouTube
Generate Studio Quality Realistic Photos By Kohya LoRA Stable Diffu...
#Kohya SS web GUI DreamBooth #LoRA training full tutorial. You don't need technical knowledge to follow this tutorial. In this tutorial I have explained how to generate professional photo studio quality portrait / self images for free with Stable Diffusion training.
Our Discord server ⤵️
https://bit.ly/SECoursesDiscord
If I have been of assis...
step by step shown
Hello Furkan, thank you for your reply. In this tutorial of yours “Generate Studio Quality Realistic Photos By Kohya LoRA Stable Diffusion Training - Full Tutorial”, after upgrading the Torch to version 1.13.1, you say (time 42:41) “don’t forget to replace cuDNN dll files again.” Can you please explain I do I do this?
Also, in the same tutorial, immediately afterwards you explain how to upgrade to xFormers version 0.0.18 (time 42:45). Since I don’t currently have xFormers installed on my computer at all, will such upgrade install these on my computer or should I not make this upgrade?
Thank you for your continued assistance.
1 : 5:04 How to install newer cuDNN dll files to increase training speed
2 : yes upgrade and install same procedure
but i prefer not to use xformers if you dont get out of vram error
Hello Furkan, in the process of upgrading my Torch version I get the following Fatal error after pasting the command from the Gist. What do I do wrong?
i would do fresh install on c drive
here a great tutorial with python : https://youtu.be/B5U7LJOvH6g
SECourses
YouTube
How To Install Python, Setup Virtual Environment VENV, Set Default ...
How to install Python, have multiple Python installations, set system wide default Python version. How to compose venv for any Python installation, change Python default path, and install SD web UI properly. Discord: https://bit.ly/SECoursesDiscord.
If I have been of assistance to you and you would like to show your support for my work, please...
Thank you Furkan, I’ll follow your recommended tutorial to re-install Python, even though I’m terrified to do such tasks since, as I told you I’m an artist, not a computer geek.
Is it possible that the problem is related to the fact that every time I activate my Stable Diffusion by double clicking “webui-user.bat”, I get the “Entry Point Not Found” pop-up window (enclosed), and only after I click “OK” I get the Stable Diffusion URL?
first time seeing this error
if working though
you can ignore
Thank you Furkan. When you say “I would do fresh install on c drive”, what exactly I should install on C drive? My stable-diffusion-webui, as well as my kohya_ss are installed on D drive. I would appreciate if you can specify what exactly I should install, and on what drive? Have a nice weekend.
i think both python and stable diffusion
you can also delete cache folder
Thank you
Hello Furkan,
How are you?
I’ve carefully followed both your tutorials:
1. “How to Install Python, Setup Virtual Environment VENV, Set Default Python System Path & Install Git”
2. “Generate Studio Quality Realistic Photos By Kohya LoRA Stable Diffusion Training - Full Tutorial”
As per your advice, I've installed everything fresh (in a new folder carrying your name).
Everything went OK with all the installations in the first tutorial (automatic 1111), but in your second tutorial, when I reached the final step of launching the Kohya ss GUI on Windows (time 7:33 in your tutorial) I got on my cmd prompt a few errors (please see the enclosed screenshot), and the Kohya ss GUI page was NOT opened.
Please let me know how to rectify this, since I have meticulously followed each and every step of both your tutorials.
Thank you,
Nissan
easy
something else using your port
gui.ps1 --listen 127.0.0.1 --server_port 7860 --inbrowser --share
change port like 7861
add this to run prompt : --server_port 7861
Dear Furkan, Thanks a lot. Now it works OK.
How can I find what is using my port?
I wish to clean it so that I can follow your guidance without changing or modifying anything.
no idea 😄
Hello Furkan,
This morning I wanted to launch my new Kohya_ss in order to continue following your tutorial “Generate Studio Quality Realistic Photos By Kohya LoRA Stable Diffusion Training - Full Tutorial”.
I first doubleclicked “upgrade.bat” in order to update the program. After this update was made I doubleclicked “gui.bat” in order to get the URL and open the Kohya_ss page.
However, immediately after doubleclicking “gui.bat”, the black prompt window was opened for a second or two and then was closed, without giving me the required URL.
Please help me to understand why yesterday it was OK and not today, and what should I do in order to regularly normally obtain the URL for the Kohya_ss page.
Thanks so much for your assistance.
Dear Furkan,
Probably because I rearranged yesterday night my folders, I have started again the installation process in the “setup.bat” file inside the cloned folder. Eventually I received a URL but it says that there are modules installed outside the virtual environment and this can cause issues (please see the enclosed screenshot).
It also gives warning about a wrong version of the huggingface package.
Dear Furkan, what shoud I do in order to make it right?
if not works make a fresh install
it installed that warning message as you see
Hello Furkan,
Thank you for your reply.
1. I don’t understand what you mean when you say “it installed that warning message as you see”. What is installed according to the message? There are modules that are installed outside the venv. I have very carefully followed your tutorials for a complete fresh install, step by step, so I don’t understand why the modules are not installed exactly where they should be.
Also, is there a way to change the huggingface version? This is a much easier and shorter way than a fresh install, which I’ve just completed.
2. When I LORA train an object, say the sculptures of Giacometti, do I need to use regularisation images? If positive, which ones and how I can generate/download them?
The only regularisation images that I know are for man and woman.
Thank you for your continued support, Nissan
1 : if working dont touch
2 : yes you need new reg images. try to find a word that can represent most sculptures of Giacometti
Furkan,
Do you have a tutorial about how to create object regularization images?
Do you have any object regularization files on your patreon?
Is there a site where I can download object regularization files?
Thank you,
unfortunately i dont have. it depends on which object
Thank you
Let’s assume that I found the word that represents the objects that I want to train. What should I do now with that word in order to create the regularisation images?
just use that class word
to generate class images
nothing else
Thank you Furkan
Hello Furkan, how are you?
I’ve just received this error message upon starting a new Lora training.
What should I do in order for the training to continue?
Thank you,
Nissan
where is the above part of error
After showing this error it stopped the training.
ok but error message displayed above
Sorry, I don't understand. Please can you tell me what I should do in order to complete the training?
Furkan, would you like me to send you the whole cmd message?
yes
Hi Furkan, enclosed are 7 screen shots that together make the whole cmd.
you have out of ram error
how much ram you have? what is your gpu?
My GPU is NVIDIA Geforce RTX 2080 Ti. How do I find how much ram I have?
Also, all my Training Parameters are the same as you explain in your tutorial "Generate Studio Quality Realistic Photos By Kohya LoRA Stable Diffusion Training."
this is where you got error
i see your error
do a fresh install
install with torch 2 please
Hi Furkan, thank you for your help.
1. As per your advice I reinstalled kohya_ss a month ago, on 9.6.2023, and since then I made quite a few successful LORA trainings. Can you please explain why something is all of a sudden wrong now which necessitates a new reinstall? Please note that I’m not a computer savvy and such installations are frightening and upsetting me. Is there a way to bypass the reinstallation process?
2. Can you please refer me to your installation tutorial where you explain how to install torch 2? I did not do it in the previous reinstallation a month ago, so this might be the reason for the current problem?
it is much easier now
SECourses
YouTube
Generate Studio Quality Realistic Photos By Kohya LoRA Stable Diffu...
#Kohya SS web GUI DreamBooth #LoRA training full tutorial. You don't need technical knowledge to follow this tutorial. In this tutorial I have explained how to generate professional photo studio quality portrait / self images for free with Stable Diffusion training.
Our Discord server ⤵️
https://bit.ly/SECoursesDiscord
If I have been of assis...
GitHub
Stable-Diffusion/Tutorials/Generate-Studio-Quality-Realistic-Photos...
Best Stable Diffusion and AI Tutorials, Guides, News, Tips and Tricks - FurkanGozukara/Stable-Diffusion
i dont know what could have broken your install :/
Thank you Furkan. Enjoy your weekend
Hello Furkan, how r u? I hope u r well.
As you suggested I reinstalled kohya_ss, carefully following your SDXL Training With Kohya LoRA of July 18, but unfortunately it ended immediately as was the case the previous time.
On the cmd window the following was written: “ValueError: bf16 mixed precision requires PyTorch >= 1.10 and a supported device.” So I changed to fb16 but again, the training stopped immediately with the following note on the cmd window:
“CalledProcessError: Command '['D:\AI_FurkanGozukara\LORA\New kohya_ss 21.7.2023\kohya_ss\venv\Scripts\python.exe', './sdxl_train_network.py', '--enable_bucket', '--min_bucket_reso=256', '--max_bucket_reso=2048',
'--pretrained_model_name_or_path=D:/AI_FurkanGozukara/stable-diffusion-webui/models/Stable-diffusion/sd_xl_base_0.9.safetensors', '--train_data_dir=D:\AI_FurkanGozukara\LORA\Kamila\img',
'--reg_data_dir=D:\AI_FurkanGozukara\LORA\Kamila\reg', '--resolution=1024,1024', '--output_dir=D:\AI_FurkanGozukara\LORA\Kamila\model', '--logging_dir=D:\AI_FurkanGozukara\LORA\Kamila\log', '--network_alpha=1',
'--save_model_as=safetensors', '--network_module=networks.lora', '--text_encoder_lr=0.0004', '--unet_lr=0.0004', '--network_dim=256', '--output_name=kamila20_ohwxLORAmodel', '--lr_scheduler_num_cycles=10', '--no_half_vae',
'--learning_rate=0.0004', '--lr_scheduler=constant', '--train_batch_size=1', '--max_train_steps=8000', '--save_every_n_epochs=1', '--mixed_precision=fp16', '--save_precision=fp16', '--cache_latents', '--cache_latents_to_disk',
'--optimizer_type=Adafactor', '--optimizer_args', 'scale_parameter=False', 'relative_step=False', 'warmup_init=False', '--max_data_loader_n_workers=0', '--bucket_reso_steps=64', '--xformers', '--bucket_no_upscale']' returned non-zero
exit status 3221225477.”
I’m despaired and call again for your support. Please help me.
Thank you, Nissan
hello
wht was your gpu
RTX2080 TI 11GB Nvidia GeForce
you cant train SDXL yet
not enough vram
i am looking for a solution and workflow
Thank you Furkan, waiting to hear from you.
Hello Furkan,
I want to remind you that on 1st July, after my previous kohya_ss reinstallation, I successfully trained quite a few persons, with my NVIDIA Geforce RTX 2080 Ti.
Unfortunately, since 12 July, when I sent you my error shown in the cmd, my work is stopped and I cannot generate LORA trainings.
Please Furkan, show me the way and guide me how to have again a successful LORA training system on my computer.
I’m waiting for your solution and workflow.
Thank you again Furkan for your continued support.
hello. were you my patreon supporter?
Yes I was and I am your patreon supporter.
i can connect your computer and help you
but it has to be now because i will be out of city for 7 days
private messaged you
Good, what should i do?
please open anydesk
but asap because i am out of time
or i will message you again when i find time because i really have to go out now
sorry about this
i am bringing my laptop a long with me
No problem Furkan, as soon as you have time pls send me a whatsup message to my phone +972-5453-67890. Thank you again.
... and then I will open my computer to work with you.
Will do 👍
Furkan, I see that you are busy so please just tell me, at your earliest convenience please, what tutorial you have to install kohya_ss with SDXL on RunPod, and I will do it myself.
I need to have again the possibility to train persons with LORA because, as I told you, I don’t have this capability already for 2 weeks, and therefore I cannot work.
Waiting for your instructions, I thank you for your kind consideration.
i am so sorry
are you online right now?
this is auto installer
https://www.patreon.com/posts/for-runpod-kohya-84898806
i am gonna test it now for you
it will support sdxl
doing install on both templates right now
tested install with the above linked script @nissanle
both templates working perfect
please let me know what else you need
Thank you so much Furkan. I'll let you know of the results.
Hello Furkan,
Contrary to what you told me, the sdxl could NOT be installed on my computer.
It took me long hours and days to try and change all the possible parameters, to no avail. Retrying to run the training, I always got error messages. I enclose here only 3 examples of many more.
Finally despaired as I am, I reinstalled the old LORA and now I work with it and with NO sdxl.
You should understand that I’m not computer savvy and when you write to me “it will support sdxl” I believe every word you say and this is why I spent so much time trying to make it happen. I’d very much prefer and appreciate if you will tell me that you are not sure of the outcome.
Thank you and have a nice weekend.
hello. when and where i did tell you? so far my minimum VRAM settings for SDXL lora training is 12 GB VRAM
Hi Furkan,
On Tuesday 25/7 you sent me the enclosed message with the link to your “Automatic Kohya SS LoRA Installer” and your comment “it will support sdxl”.
Same day I replied to you and said that “I'll let you know of the results”, on which you replied with the OK emoji.
As per your request, I send you the screenshots of the above.
Dear Furkan, please don’t misunderstand me. I very much appreciate your expertise and your very kind support and I trust that whatever happened is only a misunderstanding, and maybe a lack of time upon your going on vacation. I only wish to let you know that as an artist and not a computer savvy, I fully follow your guidance and instructions and unfortunately, this time I wasted a whole week on this issue.
Having said that, I know that you are not only an expert but also kind and honest man and I look forward to working with you for a long time.
Wishing you a great weekend, Nissan
it is runpod installer script
on runpod you dont have vram limit
pick rtx 3090 and it will work
you know what is runpod right?
Do you have a tutorial explaining how to do this installation with runpod? Will I have to do this installation with runpod each and every time I want to train a person?
until you delete your pod you dont need reinstall
i have auto runpod installer and a runpod tutorial
hopefully will make updated runpod tutorial next
auto installer
https://www.patreon.com/posts/for-runpod-kohya-84898806
runpod kohya tutorial
SECourses
YouTube
How To Install And Use Kohya LoRA GUI / Web UI on RunPod IO With St...
How to install famous Kohya SS LoRA GUI on RunPod IO pods and do training on cloud seamlessly as in your PC. Then use Automatic1111 Web UI to generate images with your trained LoRA files. Everything is explained step by step and amazing resource GitHub file is provided with necessary commands. If you want to use Kohya's Stable Diffusion trainers...
Thank you.
I remember you said it somewhere but where is the setting for 12 vram LoRA training?
i am recording video right now for that
Hi Furkan,
I’m trying to increase an image resolution with ControlNet tile model, a procedure I successfully made many times in the past but now, after updating my Automatic1111 to the recently released 1.6 version it doesn’t work and I get the enclosed output. Needless to mention, I did not touch my 4x-UltraSharp which is inside models>ESRGAN, as it was prior to the 1.6 update.
What should I do in order to solve this?
Thank you
i think there is a bug i saw some mention of tile in discussions
Hello Furkan,
As per your recent advise, I follow your “Become A Master Of SDXL Training With Kohya SS LoRAs - Combine Power Of Automatic1111 & SDXL LoRAs” tutorial, step by step.
After I installed Kohya_ss, when I double click “gui.bat” (your tutorial time 14:41) a black window opens for some 2 seconds and is immediately closed, and I don’t get the URL.
Pls advise what should I do in order to correct it?
hello
so you have problem at installing kohya
please open a cmd and type python
and show me what you see
Pls see the enclosed
ok looking corfrect
can you record a quick video
how you clone kohya and install
Dear Furkan,
I meticulously follow your instructions in “Become A Master Of SDXL Training With Kohya SS LoRAs - Combine Power Of Automatic1111 & SDXL LoRAs” tutorial.
1. Image 1 - In my main AI folder “AI_FurkanGozukara” I created a new folder “KOHYA”.
2. Image 2 - In the KOHYA path I created a cmd window.
3. Image 3 – In the cmd window I pasted the Kohya GUI Repo command.
4. Image 4 – Cloning the Kohya GUI Repo.
5. Image 5 – kohya_ss Repo is cloned. Now when I double click ‘gui.bat’ a black window opens for 1-2 seconds and is closed immediately without giving me the URL.
6. Image 6 – I created a cmd window in the KOHYA\kohya_ss path and typed ‘gui.bat’. The image shows the Error which I trust you know and understand.
Dear Furkan, I repeated the kohya cloning process a few more times and always got the same error.
What is wrong and what should I do?
Thank you.
my best guess is your python installation is corrupted
i would uninstall all python
restart
and resintall as shown in this video
SECourses
YouTube
How To Install Python, Setup Virtual Environment VENV, Set Default ...
How to install Python, have multiple Python installations, set system wide default Python version. How to compose venv for any Python installation, change Python default path, and install SD web UI properly. Discord: https://bit.ly/SECoursesDiscord.
If I have been of assistance to you and you would like to show your support for my work, please...
Thank you Furkan
Dear Furkan, I did exactly what you advised me but continue unfortunately to get the same result: after I double click gui.bat the black window opens for one second and then is immediately closed, without giving me the desired URL.
Pls advise what should be done?
I found out that I have a few old versions of Microsoft Visual C++, on top of the 2015-2022 version. Pls see the attached.
Is it possible that this causes the problem?
What do you suggest I do with these versions?
did you install first
via setup.bat file?
Now I installed again the setup.bat file and I got the URL. Thank you. Happy weekend.
you are welcome