| This folder contains models trained for the following 4 couplings | |
| 1. Suletta and Miorine from Suisei no Majo | |
| 2. Maple and Sally from Bofuri | |
| 3. Nozomi and Mizore from Sound Euphonium | |
| 4. Eila and Sanya from Strike witches | |
| You can prompt them with above names except that for Maple and Sally you should use BoMaple and BoSally. | |
| You can naturally get the trained couplings but it is also possible to pair characters from different series with either the native trained model (above) or lora (below): | |
|  | |
|  | |
| ### Dataset | |
| Total size 1394 | |
| Suremio 285 | |
| - Suletta: 81 | |
| - Miorine: 85 | |
| - Suremio: 119 | |
| MapleSally 235 | |
| - Maple: 67 | |
| - Sally: 22 | |
| - MapleSally 33 | |
| Augmented with face cropping | |
| Nozomizo 272 | |
| - Mizore: 81 | |
| - Nozomi: 81 | |
| - Nozomizo: 110 | |
| Eilanya 326 | |
| - Eila 67 | |
| - Sanya: 78 | |
| - Eilanya: 181 | |
| Regularization 276 | |
| ### Base model | |
| [NMFSAN](https://huggingface.co/Crosstyan/BPModel/blob/main/NMFSAN/README.md) so you can have different styles | |
| ### Native training | |
| Trained with [Kohya trainer](https://github.com/Linaqruf/kohya-trainer) | |
| - training of text encoder turned on | |
| - learning rate 1e-6 | |
| - batch size 1 | |
| - clip skip 2 | |
| - number of training steps 64240 | |
| *Examples* | |
|  | |
|  | |
|  | |
|  | |
| ### LoRA | |
| Please refer to [LoRA Training Guide](https://rentry.org/lora_train) | |
| - training of text encoder turned on | |
| - network dimension 128 | |
| - learning rate 1e-4 | |
| - batch size 6 | |
| - clip skip 2 | |
| - number of training steps 69700/6 (50 epochs) | |
| *Examples* | |
|  | |
|  | |
|  | |
|  | |