---
license: creativeml-openrail-m
---

# 東北ずん子

- [zunko-wd-v1](https://huggingface.co/p1atdev/touhoku/blob/main/zunko-wd-v1.safetensors): WD1.5系向けLoRA

![image.jpg](https://s3.amazonaws.com/moonup/production/uploads/6305db1fcfbde33ef7d480ff/HxXBMs2-FG2OWiJIOTE7D.jpeg)

```bash
masterpiece, best quality, exceptional, best aesthetic, 1girl, zunko, green school uniform, hairband, very long hair, cowboy shot, looking at viewer, white background,
Negative prompt: worst quality, low quality, bad aesthetic, oldest, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry
Steps: 18, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 1267925586, Size: 512x768, Model hash: 4bf1855f4d, Model: wd-1-5-beta2-fp16, Clip skip: 2, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: zunko-test2-2-e10(6f99f6de4941), AddNet Weight A 1: 1, AddNet Weight B 1: 1
```

トリガーワード: `zunko`

ヘアバンドが消える場合は `hairband` を指定してください。お好みで `school uniform` などを指定してください。色が緑じゃなかったら `green school uniform` のように緑にしてください。

<details>
<summary>
学習の詳細
</summary>

学習ベースモデル: wd-15-2.6-186k.safetensors (クローズド)

学習素材: [公式](https://zunko.jp/con_illust.html#illust_07000) の画像とタグのセットそのまま

学習時のパラメータなど:

```bash
accelerate launch --num_cpu_threads_per_process 1 .\train_network.py --pretrained_model_name_or_path "wd-15-2.6-186k.safetensors" --output_dir "\zunko\test2\lora_2" --prior_loss_weight 1.0 --learning_rate 4e-5 --text_encoder_lr 4e-6 --max_train_epochs 10 --lr_scheduler cosine_with_restarts --lr_warmup_steps 60 --lr_scheduler_num_cycles 10 --optimizer_type Lion --xformers --mixed_precision fp16 --save_every_n_epochs 1 --max_token_length 225 --persistent_data_loader_workers --logging_dir "zunko\logs" --log_prefix "test2-2-" --v2 --v_parameterization --dataset_config "dataset.toml" --network_module networks.lora --network_dim 32 --network_alpha 1 --sample_every_n_epochs 1 --sample_prompts "sample.txt"  --seed 2345
```

dataset.toml:

```bash
[general]
caption_dropout_rate = 0.01
caption_extension = '.txt'
caption_tag_dropout_rate = 0.1
enable_bucket = true
keep_tokens = 1
shuffle_caption = true

[[datasets]]
batch_size = 1
resolution = 512

[[datasets.subsets]]
image_dir = 'D:\Documents\Dataset\zunko\test2\train'
num_repeats = 10
```

</details>