Type | |
Stats | 467 381 106 |
Reviews | (70) |
Published | Feb 10, 2025 |
Base Model | |
Training | Steps: 1,750 Epochs: 10 |
Usage Tips | Clip Skip: 2 Strength: 1 |
Trigger Words | kikirara vivi |
Hash | AutoV2 A1FB8DE196 |
[Illustrious] Kikirara Vivi
Disclaimer
Do not use my LoRA to produce AI image and tagging her fanart hashtag in Twitter/X or if you have an enough contribution in that image then fine.
Description
Pink hair character again but with purple hair too.
all images is generated with WAI-IL (832x1216 + Hi-Res Fix 2x)
trigger word
debut costume
kikirara vivi, pink hair, purple hair, multicolored hair, very long hair, twintails, black bow, hair intakes, ahoge, blue eyes, white shirt, black bra, bra over clothes, purple sleeves, purple pants, boots, navel, black shorts, black belt
any costume
kikirara vivi, pink hair, purple hair, multicolored hair, very long hair, twintails, black bow, hair intakes, ahoge, blue eyes
Limitations
hairstyle is not flexible
Training Details
LoRA rank
LoRA standard
dataset
70 images
parameters
resolution = 1024
batch size = 2
network dim,alpha = 16,12
mix/save precision = bf16/bf16
optmizer = Lion (weight_decay=0.1)
UNet LR = 4e-5
TE LR = 1e-05
scheduler = cosine_with_min_lr (min_lr_ratio 0.2)
l2 loss
steps
epochs = 10
total steps = 1750
repeat = 5
tools
kohya-ss GUI v24.3.0 FORKED by vjumpkung
torch 2.5.0 cu124
Runpod RTX 4090
avg weight
UNet average weights : 0.009911941787735983
TE1 average weights : 0.007731242881466945
TE2 average weights : 0.006260246460442431
*This LoRA is for studying LoRA training with new technique so do not use for damaging the vtuber (also support her too).
*If you see my model reposted on other image generation service please report to me.