All Your Tech
patreon
UPDATED: SDXL Local LORA Training Guide: Unlimited AI Images of Yourself
🕑 Added 2024-05-14 21:22:07 +0000 UTC✨ Config Files: https://www.patreon.com/posts/koya-ss-tutorial-95710740
✨ Kohya SS Auto Installer: https://www.patreon.com/posts/1-click-for-ss-103541233
✨ regularization images: https://www.patreon.com/posts/koya-ss-tutorial-95710740
💻My Stable Diffusion PC: https://kit.co/AllYourTech/stable-diffusion-build
This guide contains everything you need to train your own LORA or Low Rank Adaptation model for Stable Diffusion XL (SDXL) using your home PC. This guide will allow you to train SDXL to generate images of yourself, or anyone else for that matter. i updated this to cover the changes that Kohya has made over the past few months.
Kohya SS: https://github.com/bmaltais/kohya_ss
Parameters: scale_parameter=False relative_step=False warmup_init=False
--- If you get an error : "No data found. Please verify arguments"
--- Click the "Prepare training data" button on the Dataset Preparation tab.
---------------------------------------------------------------
--- if you get an error: Creating venv...
Python was not found; run without arguments to install from the Microsoft Store, or disable this shortcut from Settings Manage App Execution Aliases.
--- it is because you need to check the box "add python to system path" when you install python
---------------------------------------------------------------------------
--- The following folders do not match the required pattern number_text:
ERROR Please follow the folder structure documentation found at docs\image_folder_structure.md
--- If you click on the TOOLS tabs and fill in the info there (location dataset images, reg images if you have any and then enter a folder in the DESTINATION TRAINING DIRECTORY box and click PREPARE TRAINING DATA it'll create the right structure and move all the files into there for you. Then click COPY INFO TO FOLDERS TAB to make sure you have the right info in the right place.
Comments
Brian Lovett
Have you seen any further progress? You might need to try the low vram settings file that I shared here and see if that helps with the time.
sophie moon
Hi :) First, I'd like to thank you for the excellent guide—it's been incredibly insightful so far! I have recently started training my LORA model and noticed that the process is taking an unusually long time. Based on my current progress, it seems like it might take around 24 hours to complete. I'm unable to attach an image, so here are the details: Steps: 0% Epoch: 1/20 Steps: 1% (51/6800) [37/6800, 39.51s/it, avr_loss=0.156]. I am training 17 images for the model and have added 34 Regularization images. Here are the specs of my computer: GPU: NVIDIA GeForce RTX 4060 Laptop GPU VRAM: 8GB RAM: 32GB Could there be a specific setting I'm missing, or is this duration expected? Your guidance would be greatly appreciated. Thank you! Best regards, Eli
Brian Lovett
Go to task manager and see if your gpu is being utilized. Training takes a long time, even with a GPU
Suaid Ishak
I'm using RTX3060 with 12GB VRAM. 64GB memory. Why is my training is so slow. How do i know that it is utilizing my GPU?
