Completed
How to join our Discord community for help and tips
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Become a Master of SDXL Training with Kohya SS LoRAs - Combine Power of Automatic and SDXL LoRAs
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Pre-requirements of this tutorial
- 3 1-Click installer for Automatic1111 Web UI
- 4 How to install Automatic1111 Web UI for SDXL and SD 1.5 models
- 5 How to checkout and verify your installed Python version
- 6 Which Automatic1111 Web UI command line arguments you need for SDXL
- 7 Where to find all command line arguments of Automatic1111 Web UI
- 8 Where to download SDXL model files and VAE file
- 9 Which folders you need to put model and VAE files
- 10 Detailed explanation of what is VAE Variational Autoencoder of Stable Diffusion
- 11 How to set your VAE and enable quick VAE selection options in Automatic1111
- 12 What does Automatic and None options mean in SD VAE selection
- 13 Why you shouldn't use embedded VAE of SD 1.0 model
- 14 Correct resolution of SDXL - how to use SDXL
- 15 How to install Kohya SS GUI script for SDXL training
- 16 What to do if your CMD is not progressing
- 17 When you need to use FP16 instead of BF16
- 18 How to install Kohya on RunPod or on a Unix system
- 19 How to start Kohya GUI after installation
- 20 What are Stable Diffusion LoRA and DreamBooth rare token, class token, and more training
- 21 How to select SDXL model for LoRA training in Kohya GUI
- 22 How to save and load your Kohya SS training configuration
- 23 How to use my own used configuration for this tutorial video training
- 24 How to prepare your training images for Kohya LoRA or DreamBooth SDXL training
- 25 What kind of training images you should use for training
- 26 What kind of regularization images you should use? The logic of using ground truth images
- 27 What is number of repeating in Kohya SS. Which number you need to pick
- 28 Where will be your LoRA checkpoints saved
- 29 How to verify your training images dataset properly composed
- 30 How to set your generated LoRA file names
- 31 Which training parameters you should use for SDXL LoRA training
- 32 Why select train batch size 1 and gradient accumulation steps 1
- 33 The logic of number of epochs
- 34 Detailed explanation of Kohya SS training. What each parameter and option do
- 35 Which learning rate for SDXL Kohya LoRA training
- 36 Why do I use Adafactor
- 37 The rest of training settings
- 38 Which Network Rank Dimension you need to select and why
- 39 How to fix if you get out of VRAM error - not enough memory
- 40 What is Network Alpha of Kohya LoRA
- 41 Don't forget Gradient Checkpointing
- 42 How to continue training with Kohya LoRA training
- 43 What does print training command do
- 44 How to calculate number of steps for each Epoch
- 45 How to calculate how many regularization images you need
- 46 When you should increase batch size when doing Stable Diffusion training?
- 47 How number of total steps max training steps are calculated in Kohya training
- 48 How you can generate your own regularization / classification images
- 49 How to manually edit generated Kohya training command and execute it
- 50 How to start training in Kohya
- 51 How to do training on your second GPU with Kohya SS
- 52 How much VRAM is SDXL LoRA training using with Network Rank Dimension 32
- 53 SDXL LoRA training speed of RTX 3060
- 54 How to fix image file is truncated error
- 55 How to reach and contact me if you get an error
- 56 VRAM usage and speed testing of different Network Rank
- 57 How to use absolute min VRAM to make it work
- 58 When is first checkpoint generated and where they are saved
- 59 How to continue training from saved state
- 60 Auto saved configuration files
- 61 How to use LoRAs with Automatic1111 Web UI
- 62 How to assign previews to your LoRA files / checkpoints
- 63 How to do x/y/z LoRA checkpoint comparison to find best LoRA model
- 64 How to understand if your LoRA model is overtrained / cooked or not
- 65 Testing our LoRAs stylization capability
- 66 How to generate studio shot quality images that you can use on LinkedIn, Twitter, Instagram and such
- 67 How to find best generated images with using an AI tool
- 68 How to utilize ChatGPT to find very good prompts
- 69 How to utilize high-res fix and LoRA inpainting to get amazing quality distant shot images
- 70 How to fix hands and face
- 71 How to use same training command I used
- 72 When you need to reduce weight / emphasis of the rare token
- 73 How to join our Discord community for help and tips
