[willy_k]

I don’t know much about the space, but couldn’t LoRA be used to (moderately) efficiently use a relevant and individual fine-tuned model alongside a general LLM?