<h2 id="a-lora-model-trained-using-kohya.-trigger-word:-pekomama-a0d5azsnb">A LoRA model trained using Kohya. Trigger word: pekomama</h2>Tends to be sticky at 1.0 strength, I recommend you use 0.5 strength to get other outfits.Technical Details:<ul><li>Resolution: 1024px</li><li>Network: LoRA (32/16 dim/alpha)</li><li>Steps: 2525</li><li>Optimizer: Adafactor</li><li>Learning Rates: 1e-4 (UNet), 5e-5 (Text Encoder)</li><li>Features: Bucketing, caption shuffle, flip augmentation</li><li>SNR Gamma: 5, Noise Offset: 0.1</li><li>Schedule: Cosine with 3 restart cycles</li></ul>Use "pekomama" in prompts.

Files with SHA256: 0c3ab37bfca0b05bba0361941f1e596152cd092c41f1cd087218fe3d52ef4160