Kohya SS
Stable Diffusion training — LoRA, Dreambooth, fine-tuning
📋 Overview
Kohya SS is the standard tool for training Stable Diffusion models — providing LoRA, Dreambooth, and fine-tuning with a comprehensive GUI. Unlike Unsloth (which focuses on LLMs), Kohya SS is purpose-built for diffusion models. It provides a user-friendly interface for dataset preparation, training configuration, and model export.
✨ Key Features
- •10K+ GitHub stars — standard SD training tool
- •LoRA, Dreambooth, and fine-tuning
- •Comprehensive GUI
- •Dataset preparation tools
- •Model export
- •Apache 2.0 licensed
🎯 The Problem It Solves
Training Stable Diffusion models requires complex command-line configuration. Kohya SS provides a GUI that makes this accessible.
🔧 How It Works
Kohya SS provides a GUI for dataset preparation (captioning, bucketing), training configuration (learning rate, network dimensions), and model export. It supports LoRA, Dreambooth, and full fine-tuning for Stable Diffusion models.
🚀 Installation & Quick Start
Installation
See websiteQuick Start
- See documentation
✅ Pros
- •Best SD training GUI
- •Comprehensive features
- •Apache 2.0 licensed
- •Active community
- •Comprehensive documentation
- •Production-ready
❌ Cons
- •Requires significant GPU
- •UI complex
- •Training takes hours
💬 Practitioner Verdict
“Kohya SS is the standard tool for training Stable Diffusion models — the GUI is genuinely differentiated. The trade-off: requires significant GPU resources, the UI is complex, and training takes hours. For users who want to train custom Stable Diffusion models, Kohya SS is the default.”
Self-Hosted (Free)
Open source, MIT/Apache licensed. Run it yourself.
⭐ Star & Clone on GitHubFree forever. Your infrastructure, your data.
📊 Specifications
- Language
- Python
- License
- Apache-2.0
- Platform
- Linux, macOS, Windows
- Supported Models
- REST API, CLI
💰 Pricing Reality
100% free Apache 2.0. No paid tier.