plana (blue archive)[lokr]

概述

基于animagineXL v3和kohaku delta训练的Blue Archive计划人物lokr模型，也可用于生成Arona。

• 触发词为plana(blue archive)，采用详细标注方式训练，因此在希望完整还原人物特征时，需尽可能添加相关特征的提示词；

• 由于训练集中混有一定数量的Arona图片，因此也可使用arona (blue archive)生成Arona，但效果不如专门针对plana的模型；

• animagineXL v3模型（kohaku delta也可，但效果稍弱）本身即可通过提示词直接生成plana和arona，此lokr旨在通过训练进一步精确还原这两个人物的特征；

• 初步测试未发现对画风有明显影响，可考虑与画风/艺术家提示词或画风lora联合使用；

• 未测试这两个版本的lokr在非对应基础模型上的迁移性。

训练参数

遵循Fair AI Public License 1.0-SD协议，公开训练参数，使用kohya-ss训练脚本进行训练。

• an3xl v1.0版本，训练集共647张图片。

pretrained_model_name = "animagineXLV3_v30.safetensors"
learning_rate = 6e-06
resolution = "1024,1024"
lr_scheduler = "cosine"
train_batch_size = "4"
gradient_accumulation_steps = 4
max_train_steps = "1200"
mixed_precision = "fp16"
seed = "1919810"
optimizer_type = "lion8bit"
network_module = lycoris.kohya
network_args "preset=attn-mlp" "factor=8" "algo=lokr"
network_dim = 100000
multires_noise_iterations = "8" 
multires_noise_discount = "0.6" 
network_train_unet_only = true
min_snr_gamma = 10
enable_bucket = true
bucket_reso_steps = 32
min_bucket_reso = 512
max_bucket_reso = 2048
class_tokens = 'plana\(blue archive\)'
shuffle_caption = true
gradient_checkpointing = true
xformers = true
cache_latents_to_disk = true
persistent_data_loader_workers = true

• delta v1.0版本，训练集共722张图片。

pretrained_model_name = "kohakuXLDelta_rev1.safetensors"
learning_rate = 1e-05
resolution = "1024,1024"
lr_scheduler = "cosine"
train_batch_size = "4"
gradient_accumulation_steps = 4
max_train_steps = "1200"
mixed_precision = "fp16"
seed = "1919810"
optimizer_type = "lion8bit"
network_module = lycoris.kohya
network_args "preset=attn-mlp" "factor=8" "algo=lokr"
network_dim = 100000
multires_noise_iterations = "8" 
multires_noise_discount = "0.6" 
network_train_unet_only = true
min_snr_gamma = 10
enable_bucket = true
bucket_reso_steps = 32
min_bucket_reso = 512
max_bucket_reso = 2048
class_tokens = 'plana\(blue archive\)'
shuffle_caption = true
gradient_checkpointing = true
xformers = true
cache_latents_to_disk = true
persistent_data_loader_workers = true

许可证

Fair AI Public License 1.0-SD