Model Training Summary:

• Dataset: Focused on images of trimmed pubic areas on females. (NOT EXPLICIT - HARDCORE) standing mostly

• Captioning: Generated using Koyha_SS with GIT captioning.

• Training Platform: Conducted on Civit’s onsite trainer, targeting 2,475 steps.

Training Configuration:

• Engine: Kohya

• Learning Rates:

  • UNet LR: 0.0005

  • Text Encoder LR: 0

• Clip Skip: 1

• LoRA Configuration:

  • Type: LoRA

  • Network Dimensions: 2

  • Alpha: 16

• Training Details:

  • Repeats: 30

  • Resolution: 1024

  • Batch Size: 4

  • Max Epochs: 6

  • LR Scheduler: Cosine with Restarts (3 cycles)

  • Min SNR Gamma: 5

  • Noise Offset: 0.1

  • Bucket Enabled: True

  • Optimizer: AdamW8Bit

• Augmentation: No flip augmentation; captions were not shuffled.

Notes:

• Model is designed to handle images of trimmed pubic areas on females.

• Training emphasizes a high number of steps and epochs to ensure model accuracy and refinement.