Luke Williams’ Post

OpenAI have just released a really good and comprehensive guide on prompt engineering: https://lnkd.in/eKt64KUi. While I think guides like this are a useful way to improve your prompting once you're already familiar with tools such as ChatGPT, I wouldn't recommend it for anyone just starting out. My main advice for anyone new is just to play around and figure what works and doesn't for you (i.e., f**k around and find out).

OpenAI recently released a prompt engineering guide with 6 specific strategies to help you unlock ChatGPT’s potential

OpenAI recently (and somewhat quietly) released a prompt engineering guide buried in their documentation section! In plain speak, they published guidance on strategies and tactics designed to help users more effectively query ChatGPT to get the best results. They provide strategies along with specific tactics and examples: 1) Write clear instructions 2) Provide reference text 3) Split complex tasks into simpler subtasks 4) Give the model time to think 5) Use external tools 6) Test changes systematically Take a look, it is quite helpfu! https://lnkd.in/eVt77iX6 Alex Haitoglou, Rebecca Rutschmann, Sam Isaacson, Prof. Nicky Terblanche (PhD), Tim Harrison, Anna Gallotti, Dan Fisher, Ph.D., Jon Peters, Jonathan Passmore, Will Guillaume Foussier, Maureen Kennedy, MA, PCC, Shonna Waters, PhD, Moritz Sudhof, Shana Montesol Johnson, PCC, Kelly Griffith, Donna Patricia Ann Eiby, Melvin L. Smith, Ph.D., Alicia Hullinger, PhD, Gwen Cooper, Rasa Katilienė, PhD, Pamela Richarde (She/Her), Cheryl Procter-Rogers, MCC, APR, Fellow PRSA, Magdalena Nowicka Mook, Joel DiGirolamo, Olivier Malafronte, Mara Gerstein, Rae Cohen Bernamoff, Bryan Ackermann, Keith Sonderling, Guillermo Corea

Link to official article : https://t.co/xhNBs9niQx Find below the final roadmap to Prompt Engineering with Chatgpt. #dataengineer #datascientist #dataanalyst

Here is an introductory Prompt Engineering Guide from the makers of ChatGPT https://t.co/xhNBs9niQx

Checkout the 'real' prompt engineering guide, per OpenAI! This is quite different than other material - this is a descriptive guideline of how ChatGPT is supposed to work. It's like the manual that comes with your car. Before this, prompt engineering guides were more of a game of inference based on people playing with a black box. https://lnkd.in/gzdVTeha

The concepts used in RAG are the same you should use to do prompt engineering and make better ChatGPT evals. When you add context to your prompts, you are doing the retrieval manually instead of automatically querying a vector database for documents. This concept can greatly enhance your usage of ChatGPT and save you a lot of time. This is the 2nd of 5 follow-ups to a previous post on RAG + Vision tasks. https://lnkd.in/ggJjaDEk In this OpenAI guide on Prompt Engineering (https://lnkd.in/g49ppNw6), the 2nd strategy (of 6) is to "Provide reference text". You are doing the manual work to provide a meaningful context; we can think of this as "fine-tuning your model" with new data. Here is an example. USER <insert articles, each delimited by triple quotes> Question: <insert question here>

Looks like I missed the Custom GPT documentation update that explains how custom actions work and API schema definition. Critical for building system integrations with ChatGPT and protecting IP. Whilst it might still seem a little intimidating but at least I can upload it into ChatGPT to help interpret, and I've found it pretty good at just generating a schema if you ask for older APIs. https://lnkd.in/ex28dcKS

OpenAI just released a prompt engineering guide to improve your results with ChatGPT. Worth a read! https://lnkd.in/gWWS9nAA #generatieveai #artificialntelligence #chatgptprompts

Interesting guide from OpenAI for writing better prompts when using ChatGPT. https://lnkd.in/defHapVD