前言（Preamble）：

此模型仅限学习研究使用，严禁商业用途。

This model is for learning and research purposes only.

It is strictly prohibited for commercial use.

更新日志（Update Log）：

2026.02.01：

• 更新和优化训练集，最新数据截止于 2025 年 12 月。

• The training set has been updated and optimized, with the latest data being current as of December 2025.

2025.05.20：

• 更新和优化训练集，最新数据截止于 2025 年 5 月。

• The training set has been updated and optimized, with the latest data being current as of May 2025.

• 发布两个 LoKr 模型，分别在 Illustrious XL 和 NoobAI XL V-Pred 上训练得到。

• Two LoKr models have been released, trained on Illustrious XL and NoobAI XL V-Pred respectively.

• 触发词：onineko。

• Trigger Words: onineko.

2023.07.17：

• LoHa 模型存在一些过拟合现象，推荐权重：0.6 ~ 0.8。

• This model has some overfitting. Recommended weights: 0.6 ~ 0.8.

• 推荐使用“高清修复”生成 1024px 图像。

• It is recommended to use "Hires. fix" to generate 1024px images.

• 触发词：style_onineko，quality_best。

• Trigger Words: style_onineko, quality_best.

训练集（Training Set）：

number of images = 115

• 2_best quality = 16

• 2_chibi = 9

• 5_masterpiece = 90

epochs = 17

[Illustrious XL]onineko-v2[LoKr][Multires Noise].toml：

model_train_type = "sdxl-lora"
pretrained_model_name_or_path = "D:/AI Files/SD-Trainer/sd-models/Illustrious XL 1.0.safetensors"
train_data_dir = "D:/AI Files/SD-Trainer/train/Onineko XL"
prior_loss_weight = 1
resolution = "1024,1536"
enable_bucket = true
min_bucket_reso = 512
max_bucket_reso = 2048
bucket_reso_steps = 64
bucket_no_upscale = true
output_name = "[Illustrious XL]onineko-v2 EP"
output_dir = "D:/AI Files/SD-Trainer/output/LoKr/[Illustrious XL]onineko-v2[LoKr][Multires Noise]"
save_model_as = "safetensors"
save_precision = "bf16"
save_every_n_epochs = 1
save_state = false
max_train_epochs = 20
train_batch_size = 2
gradient_checkpointing = true
gradient_accumulation_steps = 58
network_train_unet_only = true
network_train_text_encoder_only = false
learning_rate = 1
unet_lr = 1
text_encoder_lr = 1
lr_scheduler = "constant"
lr_warmup_steps = 0
optimizer_type = "Prodigy"
network_module = "lycoris.kohya"
network_dim = 114514
network_alpha = 1
randomly_choice_prompt = false
positive_prompts = "(masterpiece, extremely aesthetic, best quality, amazing quality), newest, highres, absurdres, onineko, sitting, yokozuwari, arm support, depth of field, looking at viewer, 1girl, short hair, grey hair, panda ears, hair between eyes, grey eyes, light blush, expressionless, closed mouth, china dress, black dress, medium breasts, bare arms, hand on own thigh, barefoot, nature, on grass, flower, butterfly, dappled sunlight, blurry background"
negative_prompts = "(worst aesthetic, worst quality, low quality), low details, bad quality, poorly drawn, (logo, watermark, signature, artist name), (bad hands, mutated hands), extra digits, fewer digits, bad feet, multiple views, 4koma, furry, anthro"
sample_width = 1024
sample_height = 1536
sample_cfg = 5.5
sample_seed = 2333
sample_steps = 28
sample_sampler = "euler"
sample_every_n_epochs = 1
log_with = "tensorboard"
log_prefix = "[Illustrious XL]onineko-v2 - "
logging_dir = "./logs/LoKr"
caption_extension = ".txt"
shuffle_caption = true
keep_tokens = 1
max_token_length = 255
multires_noise_iterations = 6
multires_noise_discount = 0.3
flip_aug = true
seed = 1919810
mixed_precision = "bf16"
xformers = true
lowram = false
cache_latents = true
cache_latents_to_disk = true
persistent_data_loader_workers = true
network_args = [
"conv_dim=4",
"conv_alpha=1",
"dropout=0",
"algo=lokr",
"factor=4"
]
optimizer_args = [
"decouple=True",
"weight_decay=0.01",
"use_bias_correction=True",
"d_coef=2.0"
]

[NoobAI XL]onineko-v2[LoKr][V-Pred].toml：

model_train_type = "sdxl-lora"
pretrained_model_name_or_path = "D:/AI Files/SD-Trainer/sd-models/NoobAI-XL V-Pred v1.0.safetensors"
train_data_dir = "D:/AI Files/SD-Trainer/train/Onineko XL"
prior_loss_weight = 1
resolution = "1024,1536"
enable_bucket = true
min_bucket_reso = 512
max_bucket_reso = 2048
bucket_reso_steps = 64
bucket_no_upscale = true
output_name = "[NoobAI XL]onineko-v2 EP"
output_dir = "D:/AI Files/SD-Trainer/LoKr/[NoobAI XL]onineko-v2[LoKr][V-Pred]"
save_model_as = "safetensors"
save_precision = "bf16"
save_every_n_epochs = 1
save_state = false
max_train_epochs = 20
train_batch_size = 2
gradient_checkpointing = true
gradient_accumulation_steps = 58
network_train_unet_only = true
network_train_text_encoder_only = false
learning_rate = 1
unet_lr = 1
text_encoder_lr = 1
lr_scheduler = "constant"
lr_warmup_steps = 0
optimizer_type = "Prodigy"
network_module = "lycoris.kohya"
network_dim = 114514
network_alpha = 1
randomly_choice_prompt = false
positive_prompts = "(masterpiece, extremely aesthetic, best quality, amazing quality), newest, highres, absurdres, onineko, sitting, yokozuwari, arm support, depth of field, looking at viewer, 1girl, short hair, grey hair, panda ears, hair between eyes, grey eyes, light blush, expressionless, closed mouth, china dress, black dress, medium breasts, bare arms, hand on own thigh, barefoot, nature, on grass, flower, butterfly, dappled sunlight, blurry background"
negative_prompts = "(worst aesthetic, worst quality, low quality), low details, bad quality, poorly drawn, (logo, watermark, signature, artist name), (bad hands, mutated hands), extra digits, fewer digits, bad feet, multiple views, 4koma, furry, anthro"
sample_width = 1024
sample_height = 1536
sample_cfg = 5.5
sample_seed = 2333
sample_steps = 28
sample_sampler = "euler"
sample_every_n_epochs = 1
log_with = "tensorboard"
log_prefix = "[NoobAI XL]onineko-v2 - "
logging_dir = "./logs/LoKr"
caption_extension = ".txt"
shuffle_caption = true
keep_tokens = 1
max_token_length = 255
flip_aug = true
seed = 1919810
mixed_precision = "bf16"
xformers = true
lowram = false
cache_latents = true
cache_latents_to_disk = true
persistent_data_loader_workers = true
network_args = [
"conv_dim=4",
"conv_alpha=1",
"dropout=0",
"algo=lokr",
"factor=4"
]
optimizer_args = [
"decouple=True",
"weight_decay=0.01",
"use_bias_correction=True",
"d_coef=2.0"
]
v_parameterization = true
zero_terminal_snr = true
scale_v_pred_loss_like_noise_pred = true
debiased_estimation_loss = true