I have used 2000 regularisation images and trained on 96 images of the subject. Trained on the model visiongenrealism.
It may or may not work with other models.
Use "ivnktrmp woman" as the trigger word. Yes both one is the token and other is the class name.