📄️ Tip 1: To Do and Not To Do
The tips I provide can actually be applied to various scenarios. I explained them in a specific scenario because I believe they are more likely to be used in that scenario, and making them easier to remember. It does not mean that these tips can only be used in that specific scenario. Additionally, mixing multiple tips is also a good approach.
📄️ Tip 2: Include Examples
If you find it challenging to explain a problem or provide instructions with just text, you can include examples within the prompt:
📄️ Tip 3: Use Primer Words to Guide Output
A small tweak for code generation is using primer words at the end to signal the AI can start writing code now.
📄️ Tip 4: Add Roles or Characters
The rewrite example above added a role to "make it easier to understand", a common optimization. Roles like "for a business audience" work too.
📄️ Tip 5: Use Symbols to Separate Instructions and Text
Whether summarizing or extracting, you'll likely input a lot of text. A small tweak is using """ to separate instructions and content. From my testing, this improves accuracy especially for multi-paragraph text (tip via OpenAI's best practices)
📄️ Tip 6: Use Formatting Terms to Specify Output Structure
The above example uses this technique - a variant of Tip 2 more common for text generation. You could also apply it in Scenario 7 for customized summaries.
📄️ Tip 7: Zero-Shot Chain of Thought
To address the third drawback, researchers found a technique called chain of thought prompting.
📄️ Tip 8: Few-Shot Chain of Thought
To address this, we can use few-shot chain of thought.
📄️ Tip 9: Miscellaneous
Here are some small tips that I will compile here.