Trigger words makes lora much stronger.
Recommended settings for Anima:
Sampler:
er_sde | euler | dpmpp_sde_gpu
Scheduler:
beta | linear_quadratic | simple
Lora Strength: 0.6-0.8 | Steps: 25 | CFG: 4
Description
FAQ
Comments (1)
Hey i've been figuring out the settings for lora training on Anima and your loras look really cool! Can i ask about your parameters? I see that you use cosine with restarts, how many cycles do you set for it? It doesn't seem to be recorded in the metadata