santa hat
deerdeer nosedeer glow
Sign In

Come to the conclusion that using different models in the LoRA Trainer doesn't do anything

Come to the conclusion that using different models in the LoRA Trainer doesn't do anything

All it does is not give the LoRAs any new variations if I used the actual LoRAs or if they had their "downloaded model" version and I can't really test or verify with CivitAI's Creator thing because I will unnecessarily bump my models so I decided it wasn't worth the effort, not even knowing if the models even changed.

Honestly feel discouraged at the whole deal. I'll be sticking to the Anime Model for now on.

UPDATE, 12-13-2023: Tried once more because I was trying to make a certain model's style work and figured to use a checkpoint. Unfortunately, I wouldn't know if it worked until I fully cook them so I basically wasted a lot of processing and so, I got struck with the "GPU backend" message.

It is extremely frustrating to play a guessing game with how it works and knowing it may potentially screw me out of making more LoRAs really puts the final nail in its coffin.

0

Comments