This is a LoRA for Hiyasaka 冷やさか(@hiyasaka001)'s style.
I will remove this if requested by the original artist.
Info:
Trigger word: @h1y4s4k4,
Recommended NEGATIVE prompt: loli, halo,
Description
Trained on Anima Preview v02.
FAQ
Comments (3)
Hii, I have a little question about how do you train your anima loras. I've looked at the metadata of several loras and noticed that the epochs and steps are often very different from each other, and I don't know if there's a reason for that. Sometimes you leave the epochs halfway, other times you finish them completely, and so on. Is that due to the number of images? Is there a limit on the number of steps you've set? I'm curious because I'm still testing what the best config would be to train my own loras as well. I would greatly appreciate your help if possible <3
The step count in my LoRA metadata does not seem to be written in an intuitive way. I often get confused when checking it myself later. Since the number of epochs and the number of training images are accurate, please calculate the total number of steps by multiplying the number of epochs by the number of training images. I am currently aiming for 5,000 steps for my training.
@toghashie441 Thank you so much for your help!