Trained with 22 images, it's a second version, I fixed some things that needed
cm4c1dv7m089mks185t211lrn.safetensors