caption_dropout_rate: 0.05
enable_bucket: true
epoch: 50
fp8_base: true
gradient_accumulation_steps: 1
gradient_checkpointing: true
learning_rate: 1.0
lr_scheduler: "cosine"
max_train_steps: 7350
optimizer_type: "Prodigy"
optimizer_args = ["decouple=True", "weight_decay=0.01", "d_coef=0.8",
"use_bias_correction=True", "safeguard_warmup=False", "betas=0.9,0.99",]
resolution: "1024,1024"
原本并不打算创建LoRA,因为Pony模型可以直接通过提示词生成,但为了测试图像预处理脚本而制作了它。
构图过于固定,难以理解其存在的意义。