数据集来自公开的手办训练集,我对其进行了调整增减,以低拟合度训练出了一个可以让生成图像变得更立体的lora模型。
The dataset is from a public figure figurine training set. I made adjustments, including additions and deletions, to it. With low fitting training, I developed a LORA model that enhances the three-dimensional appearance of generated images.
视情况我可能会上传权重更高的模型（因为他并不是特别好）
Depending on the circumstances, I might upload a model with higher weights (because it's not particularly good).