ChatGPT Quick Guide - Prompt Engineering, Plugins, and more!

Master structuring prompts with delimiters and new lines, specify details, provide examples, and enforce output formats to create clear, well-organized prompts.

Explore few shot prompts to guide sentiment analysis by providing a few labeled examples, teaching the model to output negative, positive, or neutral, and illustrate using templates.

Learn to design step-by-step prompts for multi-task workflows, such as translating Spanish text to English and then producing a three-sentence summary with clear line breaks.

Explore prompting models to think step by step and explain their logic and probabilities, then see how step-by-step formatting clarifies solving probability brain teasers like the emperor's proposition.

Learn how to use custom instructions to steer ChatGPT’s inputs and outputs, specify a preferred language like JavaScript, and replace few-shot prompts with persistent response rules.

Identify how large language models hallucinate with confident but false outputs, and apply prompt engineering and certainty prompts, plus code interpreter and plugins, to improve truthfulness.

Explore how to craft ChatGPT prompts for summarization and extraction, using bullet points or tables, enforcing a seven bullet point limit and tone transformations through custom instructions.

Explore how sentiment analysis serves as a form of categorization and extend it to classify any text into topics, such as technical or non-technical, using custom instructions and prompt engineering.

Transform existing text to different tones and formats with ChatGPT, turning Wikipedia style content into travel-friendly copy and markdown, html, or json formats.

Expand text from a seed or generated ideas into a larger piece, exploring secrecy and scale in topics like the Manhattan Project through outline-form expansion, prompt engineering, and custom instructions.

Explore how ChatGPT plugins connect to online services via APIs to perform tasks like booking with Expedia and OpenTable, or fetching information from Wikipedia.

Learn to enable and use ChatGPT plugins to interact with a PDF document, install the PDF plugin, and generate per-page summaries while noting sensitive information precautions.

Compare the default GPT-4 response with the Wikipedia plugin to access direct Wikipedia content. Use the plugin to pull articles, images, and related links for richer, cited information.

Discover how to connect a large language model to travel services with Expedia and Wikipedia plugins, search flights from SFO to Tokyo, and view booking links.

Explore the Wolfram Alpha plugin, a symbolic computing knowledge engine using the Wolfram Language to perform algebra, plotting, and calculations via natural language requests, with API and code interpreter links.

The code interpreter connects ChatGPT to a sandbox to upload files, generate and run Python code, and iteratively refine results with self-checks until questions are answered or information is needed.

Upload a csv and analyze it with the code interpreter to compute the average tip, group by day, and visualize results using pandas.

Visualize data with the code interpreter by crafting prompts that match your dataset’s actual column names, then generate distributions per day and export to csv for download.

Upload pdfs with the code interpreter to extract text and ask questions about the document. Learn to summarize and research topics, like the GPT-2 paper, while examining risks and misuse.

Show how code interpreter treats images as files, runs python code to draw bounding boxes and convert to black and white, and distinguishes image processing from recognition, noting multimodal future.