Page 182 - Demo

Contextual framing: start your prompt with sufficient backgroundinformation. This includes who you are, the project you%u2019re working on, yourgoals, your resources, and the specific context in which your project exists.This allows the language model to generate a response that aligns with yourspecific needs and constraints. For instance, you could start with, %u201cI am astudent in the CitEuroPass programme, working with a diverse team todevelop frugal biotechnology solutions for air purification in urban areas%u201d.Clarity and precision: be clear and precise about what you want from themodel. If your request is vague, the model%u2019s response will likely be too broador off-target. For example, instead of asking, %u201cHow can biotechnology helpwith air pollution?%u201d, you might ask %u201cCan you list innovative waysbiotechnology could be used to monitor and reduce air pollutants in a busyurban environment?%u201dDirective instructions: guide the model towards the type of response youwant. If you want a list, ask for it. If you want a step-by-step plan, request itexplicitly. If you want the model to debate pros and cons before giving ananswer, instruct it to do so.Engaging language: while this isn%u2019t a requirement, using engaging languagecan help make the interaction more enjoyable and the responses moreinteresting. Feel free to use a conversational tone, ask open-ended questions,or even use a bit of humour.Token consideration: each interaction with the model, including both yourprompt and the model%u2019s response, consumes a certain number of tokens.Long prompts will leave less room for the model%u2019s response, so it%u2019s a goodpractice to make your prompt concise and to the point.Remember, crafting a good prompt often involves iteration. You may not getthe perfect response on your first try, and that%u2019s okay. Review the model%u2019sresponses, refine your prompts, and try again. With practice, you%u2019ll get a feelfor how to interact with the model in a way that works best for you and yourspecific needs.n 6. 1. 3. Iterating on promptsOne of the fundamental aspects of working with language models likeChatGPT is understanding that prompt iteration can significantly improve thequality of responses. Iteration is a process of refining, adjusting, and tryingagain, which is vital in achieving desirable outcomes. Here%u2019s a more in-depthlook into the process of prompt iteration:1826 ChatGPT and Prompt Engineering