
MANA's style. This lora was trained on Civitai without regularization, resulting in poor tag reconition and compatibility. When I have free time I will retrain it with regularization.
If you find the results satisfactory, please provide feedback. I'm really interested to know how Lora performs for other users in terms of generation quality. It's an exciting aspect for the creators of a training model. weight: 0.6-0.8. the weight I used: 0.7.
MANA的风格,用civitai的训练模块试练的,没有正则化,可能泛化不太行,看哪天有时间再更新正则化后的吧,如果你觉得效果还行,请给予我反馈,我真的很想知道这些Lora在其他使用者处表现如何,这对于训练模型的创作者来说是种激励和动力,权重0.6-0.8 我使用的权重:0.7
描述:
test version
训练词语: mana
名称: MANA_02.safetensors
大小 (KB): 73903
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success