It's impressive how all of this is quickly picking up steam thanks to the Stable Diffusion model being open source with pertained weights available. It's like every week there's another breakthrough or two.
I think the main issue here is the computational cost, as - if I understand correctly - you basically have to do training for each concept you want to learn. Are pretrained embeddings available anywhere for common words?
I think the main issue here is the computational cost, as - if I understand correctly - you basically have to do training for each concept you want to learn. Are pretrained embeddings available anywhere for common words?