đī¸ ChatGPT Prompt Framework
After going through the various scenarios discussed in the basic section, you should have a deeper understanding of prompts.
đī¸ Zero-Shot Prompts
In the previous section on inference scenarios, I mentioned the technique of Zero-Shot Prompting. This chapter will provide a detailed explanation of what it is and the techniques involved in using it.
đī¸ Few-Shot Prompting
I mentioned few-shot prompting for inference scenarios in the basics - this chapter covers its pros/cons and techniques.
đī¸ Self-Consistency
Self-consistency is a supplementary technique in Chain of Thought that allows the model to generate multiple chains of thought and select the most frequent answer as the final result.
đī¸ PAL Models
PAL Models stand for Program-Aided Language Models. Note this is quite advanced, even hacky in my opinion. Through testing, I've found current ChatGPT capabilities may not need this technique anymore.
đī¸ Using OpenAI Playground
If you want to have a better understanding of OpenAI's API and often encounter unavailability of ChatGPT, I recommend using OpenAI Playground. It is more stable.