Download
1 variant available
LORA trained on 970 images of artist ぺんたごん (Pentagon).
Doesn't require trigger words, recommended weight is 1.
Training and dataset parameters:
Images are scrapped from gelbooru (upside down portraits excluded), tagged by WD tagger;
Trained using kohya-ss LORA training GUI, rank 32, resolution 512x768, no regularisation folder, other settings were default;
Trained on anything-v4.5-pruned-fp16 model;
6 epochs, 6 steps per image for 40 close-up portraits cropped out of some of images in dataset, 1 step per image for the rest 970 images.