This Lora model was trained on 450 beheading images,
each image was manually tagged with one of the these tags

decapitation,severed_head,knife,axe,sword,head_flying,guillotine,choping_block,weapon_trail,neck_cutting,neck_spray,seppukku,head_trophy,reverse_blowjob,neck_fucking,body_pile,eager

a lot of these tags dont work as intended

decapitation,severed_head,head_flying,neck_spray, and neck_cutting
are the most consistent

based on the Grapefruit_v4 hentai model, use it for the best results
https://civitai.com/models/2583/grapefruit-hentai-model

CFG scale at 10 seems to work best
use the model on about 0.7-0.9 weight

https://files.catbox.moe/f608kn.safetensors

Im not to familiar with creating hypernetworks
this model is definitly too broad and a more focused one would probably do better
and it was trained at 512,512 when I could do bigger
but the results are impressive and AI is certainly capable of generating Guro content