Matches my experience.

Just have it open a new context window, the other thing I wanted to try is to make a LoRa but im not sure how that works properly, it suggested a whole other model but it wasnt a pleasant experience since it’s not as obvious as diffusion models for images.