Prompt engineering can be challenging. At times, it can feel impossible to get the model to do exactly what you want. We’ve spent countless hours diving into the latest research and testing prompt engineering methods. Now, we’ve distilled the most effective, tried-and-true techniques into Prompt Enhancers—ready for you to use directly in the Playground.
Key points
Here’s a quick overview of the available Enhancers:
- Generate Persona: Generate high-quality personas from just a task description to improve prompt performance
- Prompt Iterator: Give feedback, and let the LLM refine and improve your prompt with each iteration
- Automatic Chain of Thought: Turn task descriptions into reasoning chains to guide the model toward better, more accurate outputs
- Semantic Alternative: Generate semantically similar variations to explore subtle differences in phrasin
Benefits
- Efficiency: Greatly reduce time spent on prompt engineering
- High quality prompts, instantly: Best practices are built in, so you don’t have to learn them from scratch
- Flexibility: Enhancers are modular, allowing you to choose whichever one you’d like to try
How it works
- Select an enhancer: In the completion Playground, click the three-star icon (next to the Run Test button) to open the Enhancer menu.
- Follow the steps: Each enhancer has slightly different steps
- Enhance and review: Review the outputs from the Enhancer and update your prompt with the new variant
Demo video
There are demo videos on each of the individual Enhancer pages linked above.
Notes
- Currently available on all plans, but will require a paid account in the future.
- Only available for completion projects.
