weight 1

This LoRA was trained only on full-body images, with a limited number of training samples, so it performs poorly when generating half-body portraits.