v 1.1 update - I retrained this model with 25 images of Olivia Dunne for 20,000 steps.
I am testing training on SDXL, the file size is way bigger than textual inversion, but the quality seems to be much better. Trained on 22 images of Olivia Dunne for 19,200 steps. Please let me know if I should continue with SDXL or keep on releasing 1.5 textual inversions :D