这是一个Zimage的反蒸馏模型，可以用于训练lora。以及正常加载lora，可以用正常的CFG以及STEP进行推理。请注意，这并不是官方的模型，而是我从turbo中进行训练提取的base部分，不代表官方最终的base模型

This is a Zimage anti-distillation model that can be used to train lora. As well as loading lora normally, you can use normal CFG and STEP for reasoning. Please note that this is not the official model, but the base part I extracted from turbo for training, and does not represent the official final base model.