Trained LoRA to capture J.K.'s art style on SDXL. Built with 150 curated images over 100 epochs.
Works best at 1.0 strength, but can use 0.6 - 0.8 for subtle influence. Trigger: "jdotkdot5"
Training:
- 150 images × 100 epochs
- Dim/Conv: 32, Alpha: 16
- AdamW8bit + Huber loss
- Cosine w/restarts scheduler
- LR: 7e-5 + Unet LR: 5e-5
- Dropouts: 0.08
- Base: NoobAI Epsilon 1.0
Enjoy! 🎨
Trained LoRA to capture J.K.'s art style on SDXL. Built with 150 curated images over 100 epochs.
Works best at 1.0 strength, but can use 0.6 - 0.8 for subtle influence. Trigger: "jdotkdot5"
Training:
- 150 images × 100 epochs
- Dim/Conv: 32, Alpha: 16
- AdamW8bit + Huber loss
- Cosine w/restarts scheduler
- LR: 7e-5 + Unet LR: 5e-5
- Dropouts: 0.08
- Base: NoobAI Epsilon 1.0
Enjoy! 🎨