Prompt Engineering Tips: How to Prompt an LLM

Hey friends! I’m Tasia Custode and in this video, I’m sharing some handy prompt engineering tips. From general knowledge things you should know when prompting an LLM to the crucial 5-part prompt formula, I’m showing you how to prompt an LLM properly. Think of this like your Prompt Engineering 101 class!

Here are the 5 key parts of a prompt formula: [Give a role] [Define the goal] [Speficy the audience] [Set restrictions/style] [Format the result]

⏰ TIMECODES:
00:00 - Prompt engineering tips
00:48 - Be specific
01:20 - Paint a picture with words
02:49 - Ask follow-up questions
03:52 - Format the result
04:35 - 5-part prompt formula
07:52 - Examples using the prompt formula

👀 MORE CONTENT:

📷 YOUTUBE CAMERA GEAR:

✅ RECOMMENDATIONS:
🎦 I use Tubebuddy to grow my YouTube Channel. Learn more:

⚙️ DOWNLOAD MY DIGITAL TOOLKIT:

🤝 CONNECT WITH ME:

*Some of these links are affiliate links, meaning I may make a small commission if you sign up or purchase a product using these links. Don’t worry, this is at no extra cost to you! Thank you so much!

Did you find this video helpful? Have questions or comments? Leave your thoughts in the comment section below.

#PromptEngineering

If you like this video, give a like, a share, or subscribe for more tech tips and app reviews.

[TasiaCustode]

Hey friends! Have you tried this 5-part formula? Comment below!

[RoyBlackwell]

Fantastic! This video helps me understand how to write good prompts.

[sreedharanbaskaradass]

My AI queries have always been very generic and the response has been a mixed bag. Now I know why. Will now try this 5-part formula. You are the best when it comes to explaining complex topics in an easy and fun way. Awesome video. Thank you very much.

[farhanahmed]

WOW Bravo! Hey Custode, that was really amazing and informative. I will definitely try it.

[ronm]

Thanks Tasia. I have not tried this 5-part 😉

[js]

Tasia, your quick engineering tips are invaluable! I like the straightforward explanation of the 5-part prompt formula and the examples of how it guides language model interactions. Great job!

[joanpablojimenez]

Excellent video Tasia. A simple and effective way of explaining prompt engineering in LLM. It was very useful to me. Thanks.

[ICECOLD]

Wow Tasia this is unexpected content from you 😅
👍👍❤❤

[JohnSmith-ngds]

LLMs are definitely not "built to build off your prompts." Each request is stateless. However, both ChatGPT and Gemini include previous parts of your questions (and summaries of the LLM's output) to give the illusion of an ongoing conversation.

[Kalashree.abhilsh]

🎯 Key points for quick navigation:

00:00 *📐 Using a specific formula can improve how you prompt LLMs like ChatGPT and Gemini.*
00:26 *📋 The way you prompt LLMs greatly affects response quality; detailed prompts yield better results.*
00:52 *📝 Being specific in prompts ensures more informative responses from LLMs.*
01:21 *🎨 Use vivid details to help LLMs recreate your imagination; a short paragraph is effective.*
02:17 *🌌 Less detailed prompts allow LLMs to add their own creativity to the output.*
02:43 *🔄 Follow-up questions enhance the responses you get from LLMs.*
03:35 *📊 Specify the format to get the desired output; this helps streamline information.*
04:28 *📖 The 5-part prompt formula includes assigning a role, defining the goal, specifying the audience, setting restrictions/style, and formatting the result.*
05:56 *👥 Specifying the audience tailors LLM responses to be more relevant.*
06:47 *🗒️ Setting restrictions, styles, or tones leads to more nuanced outputs.*
07:42 *📑 Formatting results in specific ways saves time and makes outputs actionable.*
08:34 *🥗 Example: A detailed prompt for a high-protein, plant-based meal plan resulted in a comprehensive and formatted response.*
09:31 *🏒 Example: Prompting an article on a sports topic showed how multiple components of the formula can generate engaging content.*

Made with HARPA AI

[MichaelDomer]

Role? Audience? Most people just have a question.

[adamtajyar]

Surprised this isn’t anything google based in this video lol