Interesting! I think the advantage of style fine-tuning is that you might not have to provides that much context upfront. Also, it's kind of magical to have an LLM just do something out of the box. I'll compare my local fine-tuned models against the baseline with instructions and see how they fare.