Zero-shot prompts. The model generates responses to new prompts dependant on general training with out certain examples. Prompt fine-tuning requires updating hardly any parameters when accomplishing general performance akin to entire model wonderful-tuning We've got, to this point, lar