An Easy tool for adding additional embedding in Onetrainer diffusion trainer just point it into OT's .json preset and txt file with your concept keyword
This script is to mass adding additional embedding at once with concept.txt (or any name.txt) without hassle. We strive to make the OT's additional embedding feature easy to use
- Automaticly count the CLIP token, thanks to instant_clip_tokenizer module
- Load the onetrainer preset to adding additional embedding concepts into preset
- Clone this repository
https://github.com/Ketengan-Diffusion/EZOnetrainerPT.git - Navigate to cloned directory
cd EZOnetrainerPT - Install the required module by using
pip install -r requirements.txt - Run the script
python Pivotal-Embedding
Assuming we want use example.txt (You can see the example content in this repository) as the embedding concepts, so the usage will be
python "H:\Pivotal-Embedding.py" -s "H:\EZOnetrainerPT\example.txt" -lp "J:\OneTrainer\training_presets\#sdxl 1.0.json" -d FormattedEmbedding.json (*This command launched on windows)
The output should be on root folder of the script with name FormattedEmbedding.json
Next you can copy the FormattedEmbedding.json into Onetrainer's training_presets folder and load that preset. The loaded preset should be loaded with additional embedding configuration