>Evaluators validate each tag individually — for example, protein, preparation, or health, individually rather than judging the item as a whole. Am I reading t

>Evaluators validate each tag individually — for example, protein, preparation, or health, individually rather than judging the item as a whole. Am I reading this right that the jury is multiple LLMs each iterating through each tag and voting on each? Why wouldn't you tune one LLM to be really competent at a single tag? Like a single "spicy evaluator LLM" or "protein evaluator LLM"?