accept t5xxl style prompts
training performed on a large dataset with llm-generated captions on the base dpo sd1.5 model
may produce NSFW material
Description
midjourney
high quality
base model
experimental
dpo
distillate
base models
flux1.d
t5xxl
sd3.5
big models reduction
Details
Downloads
653
Platform
CivitAI
Platform Status
Available
Created
11/13/2024
Updated
9/27/2025
Deleted
-
Files
sd15HK_v075.safetensors
Mirrors
Huggingface (2 mirrors)
CivitAI (1 mirrors)