Microsoft AI Research Develops APO Framework for LLM Prompt Optimization

Microsoft AI Research has introduced Automatic Prompt Optimization (APO), a general and nonparametric prompt optimization algorithm inspired by numerical gradient descent. APO automates and improves the process of prompt development for Large Language Models (LLMs) by using mini-batches of training data to obtain natural language "gradients" that describe the flaws in a given prompt. APO consistently outperformed the baselines on various NLP tasks, including jailbreak detection, hate speech detection, fake news detection, and sarcasm detection, highlighting its potential to raise the efficiency of big language models.