Sorry, that’s what I meant. But if they did publish the trained model, they could prune them and/or release models with less parameters (not entirely sure whether that’s the same), basically like llama or really any other LLM (or diffusers. Also I’m pretty sure the gpus are going to cost more...
I usually start reading at the beginning, maybe your chapter title was just that enticing?
Edit: nvm just read TrashyHuman’s reply probably just that (then again who doesn’t just click on the novel and read the synopsis?)