模型有8维(8dim)和32维(32dim)两个版本。我建议您下载体积更小、效果更好的8dim版本。32dim版本的存在是为了证明SDXL的二次元Lora模型,并为测试需求提供。需要注意的是,维度越高并不一定意味着结果更好。有测试需求的人可以下载体积更大的32dim版本。
heita_32 to 8是将32dim通过supermerger插件将dim值从32降到8。
训练模型:SDXL1.0官模
测试模型:Kohaku-XL
建议权重:0.6~0.8
触发词:Herta,dresschoker,chain,long sleeves,frills,shoulder cutout,white dress,frilled dress,clothing cutout,beret,black dress,
The model has two versions: an 8-dimensional (8dim) version and a 32-dimensional (32dim) version. I recommend downloading the 8dim version, as it has a smaller size and better performance. The existence of the 32dim version is to demonstrate the SDXL's two-dimensional Lora model and to provide it for testing purposes. It's important to note that higher dimensionality does not necessarily mean better results. People with testing requirements can download the larger 32dim version.
Heita_ 32 to 8 is to reduce the dim value from 32 to 8 through the supermerger plugin.
Testing model: Kohaku-XL
Recommended weights: 0.6~0.8
keyword:Herta,dresschoker,chain,long sleeves,frills,shoulder cutout,white dress,frilled dress,clothing cutout,beret,black dress,