ogipote/荻pote Anima
MixTrained-base1.0-test3MixTrained-base1.0-test2MixTrained-base1.0-test1MixTrained-p3NLTrained-base1.0
詳細
ファイルをダウンロード (1)
このバージョンについて
モデル説明
我仍然在继续研究更好的训练Anima的方式，因此需要做更多实验。此模型用于记录我的研究过程
=======================
PS:MixTrained-base1.0-test3需要触发词:@ogipote
不得不承认，Anima在推理时使用lora创造出来的风格十分不稳定，种子和提示词形式对结果影响极大。这一点让我在评估不同标注和训练参数带来的影响时十分吃力
I am still continuing to research better ways to train Anima, so I need to conduct more experiments. This model is used to document my research process.
============================
PS:MixTrained-base1.0-test3 needs a tiger word:@ogipote
It has to be admitted that the style created by Anima using LoRA during inference is highly unstable, and the seed and prompt format have a significant impact on the results. This makes it very difficult for me to evaluate the effects of different annotations and training parameters.