DreamBooth training example

Running locally with PyTorch

Installing the dependencies

Dog toy example

Training with prior-preservation loss

Training on a 16GB GPU:

Training on a 12GB GPU:

Training on a 8 GB GPU:

Fine-tune text encoder with the UNet.

Using DreamBooth for pipelines other than Stable Diffusion

Inference

Inference from a training checkpoint

Training with Low-Rank Adaptation of Large Language Models (LoRA)

Training

Inference

Training with Flax/JAX

Training without prior preservation loss

Training with prior preservation loss

Fine-tune text encoder with the UNet.

Training with xformers:

Set grads to none

Experimental results

IF

Tips and Tricks

Stage II additional validation images

IF stage I LoRA Dreambooth

IF stage II LoRA Dreambooth

IF Stage I Full Dreambooth

IF Stage II Full Dreambooth

Stable Diffusion XL