Prompt Engineering via Prompt Patterns — Series Introduction

With large language model based chatbots like ChatGPT and Bard entering the lives of millions of users, people have used these tools to solve a variety of problems they face in their daily lives. Most problems we face in our lives are not unique to us but are shared by many others. Now each person tried their own unique way to solve the common issue using large language models by tweaking and find tuning the prompts, and naturally, some were more successful than others. They boasted their success stories to others and researchers were able to get their hands on these creative solutions to common problems.

They gave it a name — Prompt Patterns.

You can switch to video version of this article and view our YouTube playlist here

Let’s go over the terminologies first before diving into prompt patterns.

Prompt Engineering

Prompt engineering is the process of structuring text that can be interpreted and understood by a generative AI model. While this is the default definition, prompt engineering is also referred to as a technique used to optimize and fine tune large language models for particular tasks and desired outputs by carefully constructing prompts or inputs to enhance their performance on specific tasks. The shorter version being a way of designing or refining prompts, questions, or instructions to tweak the model and elicit specific accurate and desired high quality responses from the model.

Prompt Patterns

Prompt engineering is a very vast field, and this article series is dedicated to a specific, but a very important portion of overall prompt engineering concept i.e. prompt patterns. Prompt patterns, taking cue from the software engineering term design patterns, are reusable solutions to common problems faced in a particular context. This series would be built on an extremely well drafted paper where the authors have tried to organize dozens of patterns people have developed to get really fine-tuned output from large language models. The article series would build upon, describe, and explain the concepts in the paper in a simplified way. For those of you who prefer reading research papers, you can switch to the paper using this link.

What would this series cover

The series is going to go over what prompt engineering and prompt pattern are, and then cover a bunch of useful prompting techniques for large language models (LLMs) with dedicated articles. Hopefully you would be able to use your favorite LLM much more productively after going through this article series. Though the concepts in this series apply generally to all LLMs, they are specifically tried out on ChatGPT.

I would emphasize here that you don’t need a technical background to master prompt patterns and with little effort anyone should be able to use them with ease.

Prompt engineering is a growing field so we won’t be able to cover all techniques upfront. If you want me to research and write about a topic not covered in this series, please let me know via comments, and I’ll try to bring your details as soon as possible.

The current (and growing) list of articles include

• Introduction to Prompt Patterns
• Prompt Engineering Via Prompt Patterns — Categories Of Prompt Patterns
• Prompt Engineering Via Prompt Patterns — Meta Language Creation Pattern
• Prompt Engineering via Prompt Patterns — Flipped Interaction Pattern
• Prompt Engineering via Prompt Patterns — Role/Persona Pattern
• Prompt Engineering via Prompt Patterns — Question Refinement Pattern
• Prompt Engineering via Prompt Patterns — Audience Persona Pattern
• Prompt Engineering via Prompt Patterns — Alternative Approaches Pattern
• Prompt Engineering via Prompt Patterns — Cognitive Verifier Pattern
• Prompt Engineering via Prompt Patterns — Fact Check List Pattern
• Prompt Engineering via Prompt Patterns — The Template Pattern
• Prompt Engineering via Prompt Patterns — Infinite Generation Pattern
• Prompt Engineering via Prompt Patterns — Visualization Generator Pattern
• Prompt Engineering via Prompt Pattern — Game Play Pattern
• Prompt Engineering via Prompt Patterns — Reflection Pattern
• Prompt Engineering via Prompt Patterns — Refusal Breaker Pattern
• Prompt Engineering via Prompt Patterns — The Recipe Pattern
• Prompt Engineering via Prompt Patterns — Few Shots Prompting
• Prompt Engineering — Chain of Thought Prompting
• Prompt Engineering — Self Consistency Chain Of Thought Prompting (CoT — SC)
• Prompt Engineering — ReAct Prompting
• Prompt Engineering via Prompt Patterns — Outline Expander Pattern

Please do clap/share if you like the content. Thank you!!!