Introduction

This project offer a python script that help you setup diffusers training environment and also do training quickly, support the following train methods:

[1] [GPU]dreambooth-lora
[2] [GPU]dreambooth
[3] [GPU]text-to-image
[4] [TPU]dreambooth
[5] [TPU]text-to-image

Note:

• Tested on Ubuntu and TPU v3-8, will not work on Windows
• The tpu training only work on Google TPU v3

---

Installation

Make sure you have python3 installed. Check that by:

python3 --version

Make sure you have questionary and distro to make this script function properly:

pip3 install questionary
pip3 install distro

Clone this repo:

git clone https://github.com/Eric-liucn/diffusers_train_tools.git

Run the script

cd diffusers_train_tools
python3 main.py

One line command:

python3 --version && pip3 install questionary distro && git clone https://github.com/Eric-liucn/diffusers_train_tools.git && cd diffusers_train_tools && python3 main.py

Status

This script is still work in progress, don't use it now...