Microsoft’s Automatic Prompt Optimization Improves Prompts to Boost LLM Performance

In the new paper Automatic Prompt Optimization with "Gradient Descent" and Beam Search, a Microsoft research team presents Automatic Prompt Optimization, a simple and general prompt optimization algorithm that automatically improves prompts for large language models, significantly reducing the time and energy spent on manual prompting approaches.

ai algorithm artificial intelligence automl boost deep-neural-networks energy general gradient language language models large language model large language models llm machine learning machine learning & data science microsoft microsoft research ml optimization paper performance prompt prompting prompts research research team search team technology