Recommends the following parameters.
Basic Model: FLUX.1 - dev-fp8
LoRA Strength: 1.0
Sample Prompt:
""Vibrant neon hues illuminate a midnight gas station bustle as a confident young Asian girl, dressed in a crisp white 'GSX1100S' t-shirt and gleaming silver Suzuki Katana motorcycle, strikes a bold pose astride her vehicle. Dark hair whipping slightly in the wind, she gazes off-camera with a sense of liberation. Slim-jeans and sleek sneakers accentuate her stance under the gas station's vibrant neon lights as she holds a red helmet at her side.",suzuki katana"
============================================================
LoRA Training Parameters
Dataset: 23 Photos (1024x768)
Use Base Model: FLUX.1 - dev-fp8 (Yellow Duck)
Network Module: LoRA
Trigger words: suzuki katana
Image Processing Parameters
Repeat: 10
Epoch: 6
Save Every N Epochs: 1
Total Repetitions Per Image: 60; Total Steps: 1380
Seed: -
Clip Skip: -
Text Encoder learning rate: 0.00001
Unet learning rate: 0.0001
LR Scheduler: constant
Optimizer: AdamW8bit
Network Dim: 64
Network Alpha: 32
Gradient Accumulation Steps: -
Shuffle caption: -
Keep n tokens: -
Noise offset: 0.03
Multires noise discount: 0.1
Multires noise iterations: 10
conv_dim: -
conv_alpha: -
Batch Size: -
---------------------------------------------------------------------------------------------------