本模型为模型作者做的一次测试，非pony版本，只在个别模型上生效。

建议权重1

tag：sakimichan

在kohaku-xl-epsilon-rev3上训练，在相同大模型的不同版本和sanaexlAnime上同样生效。

个人感觉，大佬的大模型感觉还是不太好用，但潜力不错，但是怎么释放潜力，为此我还有一个测试要做，要是失败了。。。那我还是再玩玩pony吧。就算pony的动作固化可能会对别人去优化它有点隐患，但良品率高，对眼睛好。要是下一个测试成功，那我就有新玩具了。

This model is a test conducted by the model author, not a Pony version, and only works on individual models.

Suggested weight 1

tag：sakimichan

Trained on Kohaku-xl-epsion-rev3, it also works on different versions of the same large model and SanaexlAnime.

In my personal opinion, the big model of the expert still doesn't seem very user-friendly, but its potential is good. However, how to unleash its potential? I still have a test to do for this, and if it fails... Then I'll play Pony again. Even though Pony's fixed actions may pose some risks for others to optimize it, it has a high yield rate and is good for the eyes. If the next test is successful, then I will have a new toy.