I would call this LoRA a miracle, it was difficult to find good images to train since there are few and of low quality both in resolution and artistically. It usually generates errors in the eyes and hands, I suggest using your best prompts to solve this problem, "expressive eyes" usually works for me. To make good hands I still don't have a good prompt.

LoRA training with Base Model "Anything V5". Generate images in weight 0.7