
This is my first attempt at making a Lora and I see someone already uploaded one for Tenryuu today. Honestly the one made by Foxtrott gives much better results overall.
Still uploading this as it does look to have a different style and it can generate usable images. Best to use weights between 0.6 and 0.8
Best for making Tenryuu Kai Ni but it can also create her base version, see example pics for fitting prompts.
Updated my dataset a bit but I'm pretty sure I'm doing something wrong somewhere in the whole process...
Trained for 4Epochs and it's a bit overbaked so weights should be lowered to 0.6-0.7
If anyone has any recommended settings for the training let me know!
描述:
训练词语: tenryuu kai ni \(kancolle\),tenryuu \(kancolle\)
名称: TenryuuDataset.zip
大小 (KB): 13272
类型: Training Data
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
名称: TenryuuE4.safetensors
大小 (KB): 147576
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success