v-prediction model
dataset: ~26k sfw & nsfw images
base model
custom node pack
lsnet with kaloscope to query artist tags from an image
modified trainer based on aozora