Today, I will share 26 practical tips commonly used in practice to make your output more efficient. Let's take a look.
1 - Communicate with large language models without using polite language, without adding "please", "if you don't mind", "thank you", "I want", etc., and directly state your needs.
Purpose: Save time and get to the core of the problem directly.
Example: Instead of saying "Please help me analyze this data", simply say "Analyze this data".
2 - Clearly identify your audience in the prompt, such as experts in the field.
Purpose: Clearly identify the audience to help the model better target the depth and technical level of the answer.
Example: If the audience is financial experts, you can say "Explain this financial model, assuming I am an expert in the field of finance".
3 - Break down complex tasks into a series of simpler prompts in interactive dialogues.
Purpose: Break down complex tasks into simple steps to make the answers easier to understand and implement.
Example: Instead of asking "How to create a website from scratch?", ask step by step, such as "First, how do I choose a platform for the website?"
4 - Use affirmative instructions like "do" and avoid using negative language like "don't".
Purpose: Avoid using negative language to make instructions clear and simplified.
Example: Say "Explain the basic principles of quantum physics" instead of "Don't explain quantum physics using complex terms".
5 - When you need a clear understanding or a deeper insight into a topic, idea, or any information, use the following prompts:
Explain [specific topic] in simple terms.
Explain as if I were an 11-year-old child.
Explain as if I were a beginner in [field].
Write [article/text/paragraph] in simple English, like explaining to a 5-year-old.
Purpose: Use simplified language or metaphors when a deeper understanding is needed to make complex concepts easier to understand.
Example: Instead of asking "Explain relativity", say "Explain relativity as if to an 11-year-old child".
6 - Add "I will give a tip of $xxx for a better solution!"
Purpose: Offer potential rewards to inspire deeper and more innovative answers.
Example: When asking for help with a complex problem, you can say "I'm willing to give a tip of $50 for this solution".
7 - Use example-driven prompts (use a few example prompts).
Purpose: Use specific examples as guidance to improve the relevance and accuracy of answers.
Example: Instead of saying "Write an article about global warming", provide an example like "Write an article like this one about global warming".
8 - When formatting your prompts, start with '###Instruction###', followed by '###Example###' or '###Question###' (if applicable). Then present your content. Use one or more line breaks to separate instructions, examples, questions, background, and input data.
Purpose: Use a structured format to clearly convey instructions and expectations.
Example: You can write a prompt like this: "###Instruction### Write a piece of code to sort an array of numbers. ###Example### For example, [3, 1, 4] should be sorted as [1, 3, 4]."
9 - Include the phrases "Your task is" and "You must".
Purpose: Use clear instructions to make the goal of the task more explicit.
Example: You can say "Your task is to provide an effective market analysis report" instead of just asking "Can you do market analysis?"
10 - Include the phrase "You will be penalized".
Example: When setting challenging tasks, you can say "You will be penalized if the analysis is inaccurate".
11 - Use the phrase "Answer a question in a natural, human-like way" in your prompt.
Example: "Please answer in a natural, human-like way, what is machine learning?"
12 - Use guiding words like "think step by step" when explaining steps to solve a problem.
Example: "Think step by step when explaining the steps to solve the problem."
13 - Add "Make sure your answer is unbiased and not relying on stereotypes" in your prompt.
Example: "When discussing different cultures, make sure your answer is unbiased and not relying on stereotypes."
14 - Allow the model to ask you for precise details and requirements until it has enough information to provide the desired output (e.g., "From now on, I want you to ask me questions to...").
Example: If you need a specific feature in software, you can say "I need an inventory management software. Please ask me questions to understand my specific requirements."
15 - When you want to ask about a specific topic, idea, or any information and want to test your understanding, you can use the phrase "Teach me [any theorem/topic/rule name] and include a test at the end, but don't give me the answer, and then tell me if I answered correctly."
Example: "Teach me the Pythagorean theorem and test me at the end, but don't give me the answer, and let me know if I answered correctly."
16 - Assign a role to the large language model.
Example: "As a nutritionist, tell me a healthy diet plan."
17 - Use separators.
Example: "###Question### What is machine learning? ###Answer### Machine learning is..."
18 - Repeat specific words or phrases multiple times in a prompt.
Example: "Explain photosynthesis. What is the process of photosynthesis? What is the role of photosynthesis in an ecosystem?"
19 - Combine Chain of Thought (CoT) with a few example prompts.
Example: "Solve this math problem: 5x + 3 = 2. Think step by step: First, what do I need to do? Next, how should I proceed?"
20 - Use output lead-in words, i.e., end your prompt with the beginning of the expected output.
Example: "Write an article about environmental protection. The article should start with 'Environmental protection is now more important than ever because...'"
21 - To write a detailed piece of text such as an article, paragraph, or any type of text: "Write a detailed [article/text/paragraph] about [topic] for me, including all necessary information."
Example: "Write a detailed article introducing the importance of renewable energy, including all relevant statistics and case studies."
22 - To correct/change specific text without altering the style: "Try to improve each paragraph provided by the user. You should only improve the grammar and vocabulary, making sure it sounds natural. You should not change the writing style, such as making a formal paragraph casual."
Example: "Please improve the grammar and vocabulary of this paragraph while keeping its formal style intact."
23 - When you have a complex coding prompt that may involve multiple files: "From now on, whenever you generate code that spans multiple files, create a runnable [programming language] script to automatically create the specified files or modify existing files to insert the generated code. [Your question]."
Purpose: Address programming tasks involving multiple files to improve code generation efficiency and accuracy.
Example: "I need a project that spans multiple Java files to manage employee data. Generate a script to automatically create the necessary files and insert the code."
24 - When you want to start or continue a piece of text with specific words, phrases, or sentences, use the following prompt: "I provide you with the beginning of [lyrics/story/paragraph/article...] and insert [lyrics/words/sentence]. Complete it based on the provided vocabulary, maintaining a smooth consistency."
Purpose: Inspire creativity while maintaining coherence and consistency in the text.
Example: "I provide you with the beginning of a story: 'On a sunny morning, a fox...' Please complete the story based on these words."
25 - Clearly state the requirements that the model must follow to generate content in the form of keywords, rules, prompts, or instructions.
Purpose: Ensure the accuracy and relevance of the content through specific keywords and instructions.
Example: "Write an article about global warming. Please make sure to include the following keywords: carbon dioxide, greenhouse effect, sustainability."
26 - To write any text, such as an article or paragraph, that is similar in content to a provided sample, including the following instruction: "Use the same language as the provided paragraph[/title/text/article/answer] to write."
Purpose: Guide the model to generate text with a similar style and content based on the given sample.
Example: "Write a paragraph in a similar style based on this description of medieval Europe."