Build a config
Once your config is set, use it to create a
You can get your run's status with:
Once the run starts (
RUNNING), stream logs with:
- If you want a higher model quality at the expense of longer fine-tuning time:
- Increase the number of images in your dataset.
- For a faster fine-tuning run but lower model quality, do the opposite.
- If your instance prompt uses words that Stable Diffusion might not be familiar with, set
- If you don't have a lot of data, keep the learning rate low. For Dreambooth,
Monitor your run with Weights & Biases
Blueprint fine-tuning can integrate with Weights & Biases to monitor your fine-tuning run:
wandb_api_key="YOUR_API_KEY"to enable the integration.
image_log_stepsto control how many steps you see images for
This will enable you to see your model generate images as it is being fine-tuned.
DreamboothConfig reference for a complete list of parameters.
What should my instance prompt be?
Dreambooth teaches Stable Diffusion to associate a word with a visual concept (aka the object in your fine-tuning dataset). To do so, it needs a word that doesn't already have any concepts associated with it. Make up a meaningful but unique string to describe the concept.
In the example, the dataset is 115 pictures of a dog named Ollie. So we use the word
olliedog as a unique descriptor. But your instance prompt could be any string that isn't a real word and wouldn't be found in the existing model's training data.
How do I regularize the model?
To make sure your model can generate images of both your specific object and a generic object — i.e. it can generate an image of your dog and a random dog — regularize the model with:
class_promptset to the generic object (e.g.
"a photo of a dog")
with_prior_preservation=Trueand prior preservation images in your dataset.
Your model will be automatically deployed
Once the fine-tuning run is complete, your model will be automatically deployed. You'll receive an email when the deployment is finished and the model is ready to invoke.
You can turn off this behavior by setting
FinetuningRun.create() and instead deploy your model manually.
Once your model is deployed, you can invoke it:
The model returns:
- The generated image (using Pillow)
- A URL to the generated image
For more on your newly deployed model, see the
These images were generated for the following prompts:
portrait of olliedog as an andy warhol painting(left)
side profile of olliedog as a van gogh painting(right)