ModelsLab/ Over-Fitting MIX (BlackSo

虽然听着很荒谬，但它确实做到了理论上模型混合该获得的效果。

但随着混合LORA数量的增多，已混合的LORA的效果便会随着下降。

或许早期训练的LORA并不适合这种混合方式。

Although it sounds absurd, it indeed achieves the effects that the theoretical model mixing should obtain

However, as the number of mixed LoRA increases, the effectiveness of the already mixed LoRA will decline.

Perhaps the early trained LoRA is not suitable for this hybrid approach

LORA：https://civitai.com/models/14517?modelVersionId=17096

https://civitai.com/models/39253?modelVersionId=45171