Due to limited image
Therefore it is necessary to press the weight of Unet and TE down.

Main tag : milk bath
Secondary tag : nude, navel, nipples, pussy,
LoRA str : 1.0-1.2
Prompt str : 1.0-1.2
Denoising str : 0.25 to 0.40 [if use upscale by 2+ reduce step]

Image Train : 159

Training Powershell info :

• DIM/Alpha : 16/16

• Train batch size : 4

• Training step : 2150x4x1 = 8600 step

• Learning rate Unet : 1e-4

• Text Encoder Ratio : 0.5

• resolution : 512

• mix/save_precision = BF16

• optimizer_type = AdamW

• GPU : RTX 4060 16GB 160W

• Torch 2.2 + cu12.1 + Xformers

• Vram used : 11.1GB

• Speed : 1.70it/s

• Loss Train : 0.124

• Loss Test : ???

• Local minimum : 5 time

• Global Peak minimum : 0.016

• Global Peak Maximum : 0.4627

• UNet weight average strength: 0.01267

• Text Encoder (1) weight average strength: 0.0101132