Home Creators Posts Import Register Favorites Logout

Content

Here you will find 5 optimized training presets for STYLE LORA inside Kohya SS. Whether you're training locally or using Runpod.What's inside?

3 Local Training Presets: Designed for those who love to train on their own systems.

Normal Style Training: For your everyday needs.

Fast Style Training: For those in a hurry

LowVRAM: Optimized for local systems with limited VRAM.

2 Runpod Presets: Perfect for those using Runpod's GPU renting platform. These presets automatically locate the location of the stable diffusion XL model for you.

Normal Style Training: The classic training preset.

Fast Style Training: Train your models as fast as possible

Why These Presets?

For LORA training, every little optimization can make a huge difference. These presets are the culmination of countless hours of experimentation and refinement. By using them, you're ensuring that your LORA STYLE training is both efficient and effective, regardless of where or what you're training.

And as always, supporting me on Patreon allows me to keep creating helpful resources like this for the AI art community. Thank you for your support - now go train some awesome LORAs!

👉Download Style Presets👈

password: PRESETSTYLE

Files

Comments

Zen (edited)

Comment edits

2026-09-08 20:00 There are no SDXL-RUNPOD here. There are only for RUNPOD-STYLE, is it the same one?
2023-09-20 22:07

Mage

So I can't extract this at all. Keeps giving me an error and giving me an empty folder :(

GorillAi

Is there anyway to Run SDXL outside the root of Driver (C:) ? I don’t have enough space in my driver. Can I run it from another Driver installed in my PC?

Just call me Jay

Can i ask you a question? My LORAs generate lot of 'confetti" portraits. Like if Picasso decided to to paint my photos. What may be causing it?

Marcelo Bennett

Hi! is there any settings (SDXL character) that I can use to train faster without reducing quality? i have 24gb of vram. Thanks!!

Aitrepreneur

you can increase the batch size for faster training

Marcelo Bennett

Thanks man, I will try. Apart from that, I still have the preset that you shared for Loras based on 1.5 a while ago. based on your SDXL video .. does the same apply to 1.5? I mean, instance prompt / class prompt and using an instance prompt form a character that sd1.5 already knows. I can try it myself for sure but perhaps you already have the answer. Thanks again! I've learned so much from you!!

Aitrepreneur

yes the principle is the same, again not sure about the training parameters of course, but try them out

Enrique AG

How many repeat numbers should be in the folder name for the SDXL style?

Enrique AG

regardless of # of images?

Remgrandt

What is the image size I should aim for with my training images? I have some that over 4,000x4,000, is that too big?

Aitrepreneur

that's a really good resolution, that's also what I usually use, but no need for anything bigger though

DAVID OSTLER

I'm mid way through your LoRA tutorial, the auto caption described " a woman smiling" Should I include things like smile, blue eyes, or emotions in the caption?

Aitrepreneur

emotions yes but nothing that is part of her face, unless she wears different colors lenses on each photos, if she always have blue eyes, leave it be

BadgerWi Johnson

I have a problem the .json file that I created in Kohay is not being seen in stable diffusion when I go to look at and load my loras. Loras I got from you or others I grabbed from CIVITAI show up just not he one I created in Kohay.

Aitrepreneur

what json file? A lora is a safetensors format

BadgerWi Johnson

I thought I was following the video step by step. I must have missed something to get the wrong output. Sorry to bug you, I'll go back and run through the Kohya video again.

BadgerWi Johnson

I can't figure out where I'm going wrong. I set the saved trained model as safetensors. I go to my output file and I have json. Well it back to the old fashioned way of training. 😀 .

Aitrepreneur

you can dm me some screenshots of the training windows, parameters etc

Some Guy

I resubbed because i deleted these and lost them now when i down load these versions i cant open them

Some Guy

nm 7zip would not work. had to use winrar gross

Pedro Mattos

I followed your step by step and trained my first style loras, they look really good but for some reason both of them (I've trained two) glitch a lot on any values above 0.7 - 0.8 (1.0 is just not usable at all) and I see that with the style models you've released I can use 1.0 just fine. What did I do wrong?

Aitrepreneur

well it's actually not an issue, they could be a bit overtrained so chose a model that is less trained but if you get good results by modifying the value that's also ok, values for Loras are made to be experimented and modified depending on the model you're using

Aitrepreneur

was the training successful? what is the size of the file? did you put the file into the lora folder?

Aitrepreneur

6gb??????? huh....that's wayyyy too big for a lora....meaning you probably made a mistake and instead of clicking on lora tab you clicked on dreambooth training instead :)

Aitrepreneur

Sould be yes but don't forget to put them in the stable diffusion folder instead of lora

Aitrepreneur

the biggest differences are the batch size and training resolution

Xavier Larrodé

Hello, do you have a version for SD1.5 ? Thanks

Aitrepreneur

you can use them as is, it's the same parameters except the model change (untick sdxl model checkbox) and change the reolution training from 1024,1024 to 512,512

Serhii Huranchyk

Can you please help me. I am new in SD training. Constantly getting this error. " returned non-zero exit status 1." What should i change ?

Serhii Huranchyk

raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)

Aitrepreneur

you need to send me the full error log, this is not the right error code

Serhii Huranchyk

Sorry, I am a kind of stupid in this field with white letters and numbers on a black background. Just in case I have 2080ti (i changed on fp16, but it doesn't work with any option) and 128gb ram - don't know if this information is needed at all. Please tell me whether this is a complete error or whether the entire command window needs to be copied. Error looks like this: Traceback (most recent call last): File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\sd-scripts\sdxl_train.py", line 792, in train(args) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\sd-scripts\sdxl_train.py", line 198, in train accelerator = train_util.prepare_accelerator(args) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\sd-scripts\library\train_util.py", line 4031, in prepare_accelerator accelerator = Accelerator( File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\venv\lib\site-packages\accelerate\accelerator.py", line 450, in __init__ raise ValueError(err.format(mode="bf16", requirement="PyTorch >= 1.10 and a supported device.")) ValueError: bf16 mixed precision requires PyTorch >= 1.10 and a supported device. Traceback (most recent call last): File "C:\Users\Serhio\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "C:\Users\Serhio\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 86, in _run_code exec(code, run_globals) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\venv\Scripts\accelerate.exe\__main__.py", line 7, in File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 47, in main args.func(args) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 1017, in launch_command simple_launcher(args) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 637, in simple_launcher raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd) subprocess.CalledProcessError: Command '['D:\\AI\\RUN STABLE DIFFUSION\\Kohyatraining\\kohya_ss\\venv\\Scripts\\python.exe', 'D:\\AI\\RUN STABLE DIFFUSION\\Kohyatraining\\kohya_ss/sd-scripts/sdxl_train.py', '--bucket_no_upscale', '--bucket_reso_steps=64', '--cache_latents', '--cache_latents_to_disk', '--caption_extension=.txt', '--enable_bucket', '--min_bucket_reso=256', '--max_bucket_reso=2048', '--gradient_checkpointing', '--learning_rate=0.0012', '--learning_rate_te1=1e-05', '--learning_rate_te2=1e-05', '--logging_dir=D:/Training/Pol_Kurucz_Style/log', '--lr_scheduler=constant', '--lr_scheduler_num_cycles=15', '--max_data_loader_n_workers=0', '--resolution=1024,1024', '--max_train_steps=8160', '--mixed_precision=bf16', '--optimizer_args', 'scale_parameter=False', 'relative_step=False', 'warmup_init=False', '--optimizer_type=Adafactor', '--output_dir=D:/Training/Pol_Kurucz_Style/model', '--output_name=MILES_ALDRIDGE_STYLE', '--pretrained_model_name_or_path=stabilityai/stable-diffusion-xl-base-1.0', '--save_every_n_epochs=1', '--save_model_as=safetensors', '--save_precision=fp16', '--train_batch_size=5', '--train_data_dir=D:/Training/Pol_Kurucz_Style/img', '--xformers']' returned non-zero exit status 1.

Aitrepreneur

the error says that you chose bf16 for mixed precision and your GPU doesn't support it, change it to fp16 instead

Serhii Huranchyk

Sorry for bothering again. Can you please tell what to fix this time ? Traceback (most recent call last): File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\sd-scripts\sdxl_train.py", line 792, in train(args) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\sd-scripts\sdxl_train.py", line 262, in train train_dataset_group.cache_latents(vae, args.vae_batch_size, args.cache_latents_to_disk, accelerator.is_main_process) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\sd-scripts\library\train_util.py", line 1949, in cache_latents dataset.cache_latents(vae, vae_batch_size, cache_to_disk, is_main_process) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\sd-scripts\library\train_util.py", line 964, in cache_latents cache_batch_latents(vae, cache_to_disk, batch, subset.flip_aug, subset.random_crop) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\sd-scripts\library\train_util.py", line 2297, in cache_batch_latents raise RuntimeError(f"NaN detected in latents: {info.absolute_path}") RuntimeError: NaN detected in latents: D:\Training\Pol_Kurucz_Style\img\20_miles aldridge style\IMG_9634.JPG Traceback (most recent call last): File "C:\Users\Serhio\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "C:\Users\Serhio\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 86, in _run_code exec(code, run_globals) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\venv\Scripts\accelerate.exe\__main__.py", line 7, in File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 47, in main args.func(args) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 1017, in launch_command simple_launcher(args) File "D:\AI\RUN STABLE DIFFUSION\Kohyatraining\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 637, in simple_launcher raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd) subprocess.CalledProcessError: Command '['D:\\AI\\RUN STABLE DIFFUSION\\Kohyatraining\\kohya_ss\\venv\\Scripts\\python.exe', 'D:\\AI\\RUN STABLE DIFFUSION\\Kohyatraining\\kohya_ss/sd-scripts/sdxl_train.py', '--bucket_no_upscale', '--bucket_reso_steps=64', '--cache_latents', '--cache_latents_to_disk', '--caption_extension=.txt', '--enable_bucket', '--min_bucket_reso=256', '--max_bucket_reso=2048', '--gradient_checkpointing', '--learning_rate=0.0012', '--learning_rate_te1=1e-05', '--learning_rate_te2=1e-05', '--logging_dir=D:/Training/Pol_Kurucz_Style/log', '--lr_scheduler=constant', '--lr_scheduler_num_cycles=15', '--max_data_loader_n_workers=0', '--resolution=1024,1024', '--max_train_steps=8160', '--mixed_precision=fp16', '--optimizer_args', 'scale_parameter=False', 'relative_step=False', 'warmup_init=False', '--optimizer_type=Adafactor', '--output_dir=D:/Training/Pol_Kurucz_Style/model', '--output_name=MILES_ALDRIDGE_STYLE', '--pretrained_model_name_or_path=D:/AI/RUN STABLE DIFFUSION/stable-diffusion-webui/models/Stable-diffusion/sd_xl_base_1.0_0.9vae.safetensors', '--save_every_n_epochs=1', '--save_model_as=safetensors', '--save_precision=fp16', '--train_batch_size=5', '--train_data_dir=D:/Training/Pol_Kurucz_Style/img', '--xformers']' returned non-zero exit status 1.

Aitrepreneur

ok wait are you using one of my training presets? Because I see a lot of parameters missing here?

Aitrepreneur

Can you send me a full screen of the training parameters tab?

Aitrepreneur

yeah I don't even get what version you are using since they changed it so much and you didn't show me the most important part of what I was looking for, dm me on discord or patreon and I will take a look via teamviewer

mAurelius

Hey legend. I'm trying to follow your guide to train a woman. I get the following with no training taking place. Using the standard sdxl preset. Thanks 23:04:44-103537 INFO Start training LoRA Standard ... 23:04:44-104584 INFO Validating model file or folder path E:/stable-diffusion-webui/models/Stable-diffusion/sd_xl_base_1.0_0.9vae.safetensors existence... 23:04:44-105538 INFO ...valid 23:04:44-106539 INFO Validating output_dir path E:/kohya_ss/training Sets/Katharine McPhee/model existence... 23:04:44-106539 INFO ...valid 23:04:44-107539 INFO Validating train_data_dir path E:/kohya_ss/training Sets/Katharine McPhee\img existence... 23:04:44-108539 INFO ...valid 23:04:44-108539 INFO Validating reg_data_dir path E:/kohya_ss/training Sets/Katharine McPhee/reg existence... 23:04:44-109537 INFO ...valid 23:04:44-109537 INFO Validating logging_dir path E:/kohya_ss/training Sets/Katharine McPhee/log existence... 23:04:44-110534 INFO ...valid 23:04:44-110534 INFO log_tracker_config not specified, skipping validation 23:04:44-111534 INFO resume not specified, skipping validation 23:04:44-111534 INFO vae not specified, skipping validation 23:04:44-111534 INFO lora_network_weights not specified, skipping validation 23:04:44-112602 INFO dataset_config not specified, skipping validation 23:04:44-113105 INFO Folder 20_Katharine McPhee Woman: 20 repeats found 23:04:44-113621 INFO Folder 20_Katharine McPhee Woman: 61 images found 23:04:44-114139 INFO Folder 20_Katharine McPhee Woman: 61 * 20 = 1220 steps 23:04:44-114139 WARNING Regularisation images are used... Will double the number of steps required... 23:04:44-115141 INFO Regulatization factor: 2 23:04:44-116142 INFO Total steps: 1220 23:04:44-117141 INFO Train batch size: 1 23:04:44-117141 INFO Gradient accumulation steps: 1 23:04:44-118141 INFO Epoch: 10 23:04:44-118141 INFO max_train_steps (1220 / 1 / 1 * 10 * 2) = 24400 23:04:44-118141 INFO stop_text_encoder_training = 0 23:04:44-119141 INFO lr_warmup_steps = 0 23:04:44-121141 INFO Saving training config to E:/kohya_ss/training Sets/Katharine McPhee/model\JMHS-KatharineMcPhee_20240418-230444.json... 23:04:44-122141 INFO Executing command: "" launch --dynamo_backend no --dynamo_mode default --mixed_precision bf16 --num_processes 1 --num_machines 1 --num_cpu_threads_per_process 2 "E:/kohya_ss/sd-scripts/sdxl_train_network.py" --config_file "./outputs/tmpfilelora.toml" with shell=True 23:04:44-124145 INFO Command executed. '""' is not recognized as an internal or external command, operable program or batch file. 23:04:44-266525 INFO Training has ended.

Aitrepreneur

dm me the full error log and some screenshots of the parameters + GPU

Aitrepreneur

Are you sure you're doing things right? Look at this: https://imgur.com/a/OAdhgsO First upload the preset file inside the kohya folder, right click on the file, select "copy path" then inside kohya, paste the path, DO NOT FORGET to put a / before the path, then click the load button

Ricardo Campos

it works! my fault! Thanks for all your work