Welcome to v2.0 of EveryDream trainer! Now with more diffusers and even more features!
Please join us on Discord! https://discord.gg/uheqxU6sXN
If you find this tool useful, please consider subscribing to the project on Patreon or a one-time donation at Ko-fi.
If you're coming from Dreambooth, please read this for an explanation of why EveryDream is not Dreambooth.
Windows 10/11, Linux (Ubuntu 20.04+ recommended), or use the linux Docker container Python 3.10.x Nvidia GPU with 11GB VRAM or more (note: 1080 Ti and 2080 Ti may require compiling xformers yourself) 16GB system RAM recommended minimum
Single GPU is currently supported 32GB of system RAM recommended for 50k+ training images, but may get away with sufficient swap file and 16GB Ampere or newer 24GB+ (3090/A5000/4090, etc) recommended for 10k+ images unless you want to wait a long time
...Or use any computer with a web browser and run on Vast/Runpod/Colab. See Cloud section below.
Covers install, setup of base models, startning training, basic tweaking, and looking at your logs
Behind the scenes look at how the trainer handles multiaspect and crop jitter
Make sure to check out the tools repo, it has a grab bag of scripts to help with your data curation prior to training. It has automatic bulk BLIP captioning for BLIP, script to web scrape based on Laion data files, script to rename generic pronouns to proper names or append artist tags to your captions, etc.
Download and setup base models
Training - How to start training
Basic Tweaking - Important args to understand to get started
Advanced Tweaking - More stuff to tweak once you are comfortable
Advanced Optimizer Tweaking - Even more stuff to tweak if you are very adventurous
Chaining training sessions - Modify training parameters by chaining training sessions together end to end
Data Balancing - Includes my small treatise on model preservation with ground truth data
Validation - Use a validation split on your data to see when you are overfitting and tune hyperparameters