Introducing the Prompt Engineering Toolkit

Uber has developed a comprehensive prompt engineering toolkit to enhance interactions with large language models (LLMs). The toolkit includes features for exploring LLM capabilities, creating and iterating prompt templates, and evaluating their effectiveness. It supports both development and production stages, with safety measures and version control. Key components include a Gen AI Playground, auto-prompt builder, revision control system, and evaluation pipelines. The toolkit is being used for various applications at Uber, including offline batch processing, online services, and performance monitoring.