This model was trained in Shivam Shrirao colab with 85 images at 8500 steps.

Works good, but it's powerful styling other models by merging