Description
Trained on Anima Base v1.0.
FAQ
Comments (3)
This LoRa creation is simply amazing. I have also experimented with LoRa as an AI artist, but the results were quite disappointing. Could you kindly share some of your experiences in training LoRa?
For style loras, i usually get good results by training for at least 1200 steps at batch size 2, with rank 16. Bigger batch size might need more epochs/repeats. For optimizer, adamw is a good choice, just need to set betas = (0.9,0.99) for short runs like this.
n3r2smileSep 10, 2026
@rift1000 Thank you, I will strive to be as successful as you!