SD-Trainer, also known as lora-scripts, is a popular open-source training toolkit for Stable Diffusion that wraps the well-known kohya-ss sd-scripts in a friendly graphical interface with presets and one-click environment setup. It covers LoRA and DreamBooth training, and its newer Train WebUI aims to be a complete Stable Diffusion training studio with tagging and dataset tools built in. The project has about 6,200 stars on GitHub and is still maintained, with updates in 2026.
Fine-tuning a diffusion model normally means wrestling with command-line flags, caption files and environment setup before training even starts. SD-Trainer removes that friction: install scripts create the virtual environment, the GUI exposes the important kohya-ss parameters as forms with sensible presets, and one click starts training. The Train WebUI adds the surrounding workflow โ a WD 1.4 tagger for automatic captioning, a tag editor for cleaning datasets, and TensorBoard integration for watching loss curves โ so dataset preparation, training and monitoring all live in one place. It runs on Windows and Linux with Python 3.10 and works with the Stable Diffusion model family that kohya’s scripts support, making it one of the most approachable ways for artists to train a LoRA of their own style or character.
SD-Trainer is free and open source under the AGPL-3.0 license. There is no paid version; your only cost is the GPU time to train your models.
SD-Trainer is for Stable Diffusion artists, character and style creators, and hobbyist fine-tuners who want reliable LoRA training without living in the command line, as well as trainers who appreciate having captioning, tagging and monitoring in one studio.