OneTrainer

OneTrainer is a one-stop solution for all your stable diffusion training needs.

Features

Supported models: Stable Diffusion 1.5, 2.0, 2.1, SDXL and inpainting models
Model formats: diffusers and ckpt models
Training methods: Full fine-tuning, LoRA, embeddings
Masked Training: Let the training focus on just certain parts of the samples.
Automatic backups: Fully back up your training progress regularly during training. This includes all information to seamlessly continue training.
Image augmentation: Apply random transforms such as rotation, brightness, contrast or saturation to each image sample to quickly create a more diverse dataset.
Tensorboard: A simple tensorboard integration to track the training progress.
Multiple prompts per image: Train the model on multiple different prompts per image sample.
Noise Scheduler Rescaling: From the paper Common Diffusion Noise Schedules and Sample Steps are Flawed
EMA: Train you own EMA model. Optionally keep EMA weights in CPU memory to reduce VRAM usage.
Aspect Ratio Bucketing: Automatically train on multiple aspect ratios at a time. Just select the target resolutions, buckets are created automatically.
Dataset Tooling: Automatically caption your dataset using BLIP, BLIP2 and WD-1.4, or create masks for masked training using ClipSeg or Rembg.
Model Tooling: Convert between different model formats from a simple UI.
Sampling UI: Sample the model during training without switching to a different application.

A more detailed list can be found here.

Installation

Installing OneTrainer requires Python 3.10. You can download Python here https://www.python.org/downloads/windows/. Then follow these steps:

Automatic installation (Windows)

Clone the repository git clone https://github.com/Nerogar/OneTrainer.git
Run install.bat

Manual installation (Windows and other systems)

Clone the repository git clone https://github.com/Nerogar/OneTrainer.git
Navigate into the cloned directory cd OneTrainer
Set up a virtual environment python -m venv venv
Activate the new venv venv\scripts\activate
Install the requirements pip install -r requirements.txt

Updating

Automatic update

Run update.bat

Manual update

Pull changes git pull
Activate the venv venv\scripts\activate
Re-Install all requirements pip install -r requirements.txt --force-reinstall

Usage

To start the UI, run start-ui.bat. You can find a quick start guide here., and a more detailed overview of different topics here.

If you need more control, OneTrainer supports two modes of operation. Command line only, and a UI. All commands need to be run inside the active venv created during installation.

All functionality is split into different scrips located in the scripts directory. This currently includes:

train.py The central training script
train_ui.py A UI for training
caption_ui.py A UI for manual or automatic captioning and mask creation for masked training
convert_model_ui.py A UI for model conversions
convert_model.py A utility to convert between different model formats
sample.py A utility to sample any model
create_train_files.py A utility to create files needed when training only from the CLI
generate_captions.py A utility to automatically create captions for your dataset
generate_masks.py A utility to automatically create masks for your dataset

To learn more about the different parameters, execute <scipt-name> -h. For example python scripts\train.py -h

Contributing

Contributions are always welcome in any form. You can open issues, participate in discussions, or even open pull requests for new or improved functionality. You can find more information here

Before you start looking at the code, I recommend reading about the project structure here. For in depth discussions, you should consider joining the Discord server.

Related Projects

MGDS: A custom dataset implementation for Pytorch that is built around the idea of a node based graph.
StableTuner: Another training application for Stable Diffusion. OneTrainer takes a lot of inspiration from StableTuner and wouldn't exist without it.
Visions of Chaos: A collection of machine learning tools that also includes OneTrainer.