read description in versions
the second version of the model. she's a little over-trained, so use lowered weights. also, the previous version works better with strange styles
my_LoRA_rakka_v2.safetensors