<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">斗破苍穹动漫，萧薰儿脸部Battle Through the Heavens Anime, Xiao Xun'er Face1.触发词/Trigger Words: <pre><code>&lt;lora:xiaoxuner_zhongzhou_outfit_face&gt;,xiaoxuner_zhongzhou_face</code></pre>2. 训练时选用的 <a target="_blank" rel="ugc" href="https://civarchive.com/models/1704957/diva-3d-danira?modelVersionId=2201745&amp;sync-account=green">Diva 3D Danira </a>base mode 无VAE，可搜索Hugging Face下载sdxlVAE_sdxlVAE.safetensorsThe model was trained using the<a target="_blank" rel="ugc" href="https://civarchive.com/models/1704957/diva-3d-danira?modelVersionId=2201745&amp;sync-account=green"> Diva 3D Danira</a> base model (no VAE). You can search on Hugging Face to download sdxlVAE_sdxlVAE.safetensors. 3. 配合服饰 lora<a target="_blank" rel="ugc" href="https://civarchive.com/models/2657777/xiaoxunerzhongzhououtift"> xiaoxuner_zhongzhou_outfit </a>，face lora adtailer使用 Use together with the clothing LoRA<a rel="ugc" href="https://civarchive.com/models/2657777/xiaoxunerzhongzhououtift?modelVersionId=2984403"> xiaoxuner_zhongzhou_outfit</a> and apply the face LoRA with Adetailer. 4. 训练集参考Optional FilesPlease refer to the Optional Files for the training dataset reference. 5. 完整训练参数，感兴趣的朋友尽情调优Here are the complete training parameters, feel free to fine-tune.<pre><code>model_train_type = "sdxl-lora"
pretrained_model_name_or_path = "F:/SD/sd-webui-linux/models/Stable-diffusion/diva3DDanira_v10NoVAE.safetensors"
resume = ""
vae = "F:/SD/sd-webui-linux/models/VAE/sdxlVAE_sdxlVAE.safetensors"
train_data_dir = "F:/SD/training_dataset/btth/xiaoxuner_zhongzhou_face"
prior_loss_weight = 1
resolution = "1024,1024"
enable_bucket = false
min_bucket_reso = 512
max_bucket_reso = 1536
bucket_reso_steps = 64
bucket_no_upscale = true
output_name = "xiaoxuner_zhongzhou_face_pony"
output_dir = "F:/SD/trained_loras"
save_model_as = "safetensors"
save_precision = "bf16"
save_every_n_epochs = 2
save_state = true
save_last_n_epochs_state = 2
max_train_epochs = 24
train_batch_size = 2
gradient_checkpointing = true
gradient_accumulation_steps = 1
network_train_unet_only = false
network_train_text_encoder_only = false
learning_rate = 0
unet_lr = 0.00006
text_encoder_lr = 0.000003
lr_scheduler = "constant_with_warmup"
lr_warmup_steps = 40
optimizer_type = "AdamW8bit"
network_module = "networks.lora"
network_dim = 32
network_alpha = 32
randomly_choice_prompt = false
positive_prompts = "xiaoxuner_zhongzhou_face, 1girl, earrings, jewelry, solo, hair ornament, long hair, black hair, portrait, closed mouth, dress, realistic, expressionless, flower, blurry background, upper body, blue eyes"
negative_prompts = "lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry"
sample_width = 1024
sample_height = 1024
sample_cfg = 6
sample_seed = 2333
sample_steps = 30
sample_sampler = "euler_a"
sample_every_n_epochs = 1
log_with = "tensorboard"
logging_dir = "./logs"
caption_extension = ".txt"
shuffle_caption = true
keep_tokens = 1
keep_tokens_separator = ""
max_token_length = 255
color_aug = false
flip_aug = false
random_crop = false
seed = 1337
mixed_precision = "bf16"
no_half_vae = true
xformers = true
lowram = false
cache_latents = true
cache_latents_to_disk = false
persistent_data_loader_workers = true
vae_batch_size = 1
optimizer_args = [ "weight_decay=0.01" ]</code></pre>

Files with SHA256: 389d1ba3229955ef712ce41acefc0e54c756baff8a558830e37a9deb79c53f64