
保留了原模型大部分特征并进行升级,
Retained most features of the original model while upgrading it
难以置信的是原模型含量竟然有百分之45
Incredibly, the content of the original model is as high as 45 percent
描述:
花费了大量时间在一个非常非常非常非常愚蠢的事情上。好了,总结下最近的收获。我之前分享的那个拆解方法会将混合模型进行平均切分,因此不停的进行切割便可以提取到一些有趣的东西。
Spent a considerable amount of time on something very, very, very, very foolish. Alright, let me summarize the recent gains. The disassembly method I shared earlier averages the mixed models, so continuous cutting can extract some interesting elements.
然后我根据混合效果将模型分为了4个阶段,根据观察来看阶段越高混合时的效果越明显。
I then divided the model into 4 stages based on the blending effects. Observations indicate that the higher the stage, the more pronounced the blending effect becomes
现在我需要积攒一些混合经验
I now need to accumulate some mixed experience
训练词语:
名称: kmMIXComicsRMX4_v4.safetensors
大小 (KB): 2414699
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success