Type | |
Stats | 716 200 |
Reviews | (103) |
Published | Apr 24, 2023 |
Base Model | |
Trigger Words | Mumu, long hair, brown hair AND green hair, pointy ears, yellow eyes, hair ornament, mole under eye, small breasts, |
Hash | AutoV2 E11F1BAF88 |
Note:
To achieve better results, it is recommended to use lora block weights.
A simple method to set the weights is as follows: CHARACTER:1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0
BACKGROUND:1,1,1,1,1,1,0.2,1,0.2,0,0,0.8,1,1,1,0,0
triggle words :
Mumu, long hair, brown hair AND green hair, pointy ears, yellow eyes, hair ornament, mole under eye, small breasts,
For clothes: open coat, black skirt, white coat, choker,
Do not use muelsyse \(arknights\) or muelsyse, as the model's built-in tag may lead to overfitting or other strange effects.
If you use some negative words describing normal humans, you may lose the features of Mumu (e.g., pointy ears).
Easynegative may exacerbate overfitting; using easynegative may cause Lora to start overfitting without obtaining all of Mumu's features, so it is not recommended, especially for v1.4, as I prefer to acquire more of Mumu's characteristics.
Due to Mumu in the training dataset mostly wearing her lab coat, changing her outfit becomes very difficult. Generally speaking, V1.4 is slightly better than V1.1, as it allows for easier clothing changes. However, V1.4 is more prone to overexposure.