GitHub - pyladiesams/personalization-with-text-to-image-diffusion-models-feb2024: Get familiar with different fine-tuning techniques for text-to-image models, and learn how to teach a diffusion model a concept of your choosing (original) (raw)
🎨 Fine-tuning text-to-image diffusion models for personalization and subject-driven generation
Presentation: Personalization of Diffusion Models with 🧨Diffusers
📚 Workshop description
During the workshop you will get familiar with different fine-tuning techniques for text-to-image models, and learn how to easily teach a diffusion model a concept of your choosing (special style, a pet, faces, etc) with as little as 3 images depicting your concept.
🛠️ Requirements
Python >= 3.10, acquaintance with Diffusion models, Text-to-Image models.
NOTE 💡 While we will briefly go over diffusion models and specifically Stable Diffusion, we will not get into detail, and assume some familiarity with the diffusion process and architecture of stable diffusion models.
TIP 💌 If you're not familiar with diffusion models but interested in doing this workshop, check this (free & open-sourced) introductory diffusion class 🤓
▶️ Usage
- Clone the repository
- Start Jupyter lab and navigate to the workshop folder or Use Google collab and import Jupyter notebooks there.
- Open the first workshop notebook
- [Option1] Install requirements with
pip install -r requirements.txt
- [Option1] Install requirements with
- [Option2] Run the
Setupcells in the notebook
- [Option2] Run the
🎬 Video record
Re-watch this YouTube stream
🤝 Credits
This workshop was set up by @pyladiesams and @linoytsaban