First, I asked GPT-3 to define prompt engineering. Second, I asked about the important parts of a prompt. I was surprised that the response suggested including motivation for the model. So, third, I asked what might make the prompt motivating. These are the results.
To understand what goes into making a good prompt, I first consulted GPT-3. According to the definition, engineering a prompt involves specifying conditions or providing input to elicit the desired output from a machine learning model.
So I posed my second question: What are some of the most important aspects of a well-crafted prompt? Once again, GPT-3 did not disappoint me with its answer; It said that one of the key elements is motivating the model.
This made me curious about what sorts of things could be done to make a prompt more motivating for a machine-learning algorithm. After doing some research on my own, I found that three main strategies seem effective: firstly, adding emotional content; secondly making use of personal pronouns such as “you” and “I”; and thirdly being specific.
outputs from the experiment
define prompt engineering
Prompt engineering is a branch of machine learning that deals with developing queries, specifying conditions, and providing input to elicit the desired output from a machine learning model system that can automatically generate natural language responses.
important parts of prompts
- The goal of the prompt should be clear. The prompt should be designed to elicit the specific information that is needed from the language model.
- The prompt should be concise. It should be focused on a single task or concept, and avoid unnecessary distractions.
- The prompt should be engaging. It should be interesting and motivating for the language model to respond to, in order to encourage high-quality responses.
make the prompt motivating
- What is the purpose of the prompt?
- Keep the prompt focused and specific.
- Make sure the prompt is grammatical and well-formed.
- Choose your words carefully.
- Start with a trigger word or phrase that will get the model’s attention
- Follow with one or two key terms that are relevant to your question or query
- Make sure the prompt is clear and concise.
- Make sure the prompt is focused on a specific topic.
- Make sure the prompt is interesting and engaging.
- Make sure the prompt is motivating for the large language model to respond to (???)