Hello everyone!I've been a bit busy lately and haven't had much time to updateToday, the project experiment was finally completed, I haven't seen it for a long timeThe platform sent a private message, I didn't expect many classmates to send me private messages to update“gpt4.0 prompt words" series.
I didn't expect everyone to like this series,Today, I will continue to bring you "GPT4."0 prompt words" series.
It is recommended that you take ten minutes to look down, there will be unexpected results!
Note: Today we bring you the latest editing prompts for December, as well as GPT40sci Writing Prompts!
Don't miss it, pay attention to the collection Like **Oh!
GPT should not be understood as a machine that only receives commands。In the course of a conversation, it's not just you who ask, it answers. It is perfectly available as oneInterlocutorand engage with yourIn the course of workCome, I can ask you questionsGuide you to think.
Sometimes, it's hard to figure out how to use itGPT touching, the most critical reasons are:
You see something and you think something is wrong, but you don't know how to change it. 」
You just think there's something wrong with this passage, but you don't know what's wrong!
At this time, some are neededPrompts to guide thinking.
prompt: here’s the section of my **that i’d like to polish:
[insert **section here].
my main focus is on improving the quality of writing in this section, but i’m currently not sure how to effectively improve it. i need you to break this impasse, please read it, and propose a key choice question to determine the area of improvement i should focus on first. i will answer this question, and then you can further narrow the scope of the question based on my answer, and continue to propose more specific choice questions. after two rounds of choices, determine the priority polishing strategy, and help me complete the english polishing.
Hint: This is the ** clip I want to polish:
[Insert snippet here].
My main focus is on improving the quality of writing for this snippet, but I'm not sure how to improve it effectively at the moment. Now it's up to you to break this out of the way, so please read on and ask a core selection question to identify areas of improvement that I should focus on first. I'll base my answer on that question, and then you can narrow down the question further based on my answer and move on to more specific choice questions. After the two-step selection, I decided on a priority editing strategy and helped me with the English editing.
The following is the specific application demonstration:
That's the benefit of this kind of guided questioning, that is, before that, you may not even know what to expect fromThink about it from these two angles!
Therefore, the ultimate goal of this kind of guided questioning is to find the one in your heart"If you want to describe, you can't describe it".requirements.
Now, you can follow the question and move on:
Once you've selected this step, you're ready to executeFinal Editing Orders。This is because I made it clear in the promptAfter a two-step selectionYou can flexibly adjust it according to your needs.
How many good answers in the world are waiting for a good question. ”
This kindThe core of the idea that leads to questioning, in fact, it is throughGPT bootstrappingto help us come up with a good question.
In the example above, you can edit the selection.
Don't underestimate this little feature.
It can do the following three things.
1) Time Machine (Session Copy).
The answers between the different branches are independent of each other, andWe can accomplish different tasks under different branches。And each branch can open multiple branches indefinitely, so that the session can be copied.
2) Regret Pill (Infinite Restart).
The effect of this prompt word is not particularly satisfactory, so that's itYou can edit the prompt directly in the dialog boxRe-answer.
3) Keep the context clean.
A very important pointNote that the GPT output works by taking the previous context (including questions and answers) as input for the next time。And most criticallyGPT's window (so to speak, memory) is limited。So in the same dialog,Questions and answers accumulated earlier, will be the input for the next time, so sometimes we canReuse some key nodes, rather than indefinitely, which reduces the distraction of some preceding irrelevant information.
So let's say you've already debugged onePrompt words, for example:Simplify long and difficult sentences.
Then for the second sentence, in general, you might put the sentence below it, or sometimes add an extra prompt, for example"Follow the way above and modify the sentence below".and so on.
So the first way:
If A and B are independent, that may not be the best option. Because in any case, all the previous information will affect the quality of answer B. Because according to the previous principle,The input at this point contains the long difficult sentence A, the answer A, and possibly the additional prompt. Then it is possible to have an impact on the quality of answer b, and ifThe content is too longFor example, it is followed by c, d, e..., and possiblyCauses the memory length to be exceeded and the initial prompt word is forgotten.
Then to make use of branches, you only need to edit the placeReplace the long and difficult sentence A with B, and you can achieve two parallel branches.
After that, it can be based on answer A or answer BAsk questions in a constant way.
Note:This is just a simple demonstration, but there is no standard answer. For example, in the actual operation process, if the answer A is satisfactory to you, then the branch will be opened after the whole operationYou can start with a long difficult sentence b, because we not only have a prompt word, but also an expected exampleThat is, the long and difficult sentence A, which provides a frame of reference for the generation of answer B. After all, GPT's output is random and completely uncontrollable every time, then a high-quality answer sometimes needs to be cherished!
gpt 4.0 has a longer memory window, which allows for more efficient editing of long texts.
Here's a better hint:
prompt: in the upcoming process of **polishing, i will provide you with sections of the **in batches. each batch of **sections will start with a specific number, which consists of a ‘#’ symbol followed by an arabic numeral, such as ‘##8’, etc. your task is to carefully record each batch of **sections and their corresponding numbers. please note that while recording these sections, you do not need to perform any form of analysis or provide summaries; simply reply with ‘recorded’ after recording each batch. when i send the instruction ‘’it means that all sections of the **h**e been fully provided, at which time you should reply with ‘all content recorded’. please focus on accurate recording and **oid performing any actions beyond the scope of these instructions. now, please prepare to receive and record the **sections that will be provided.
Tip: I'll provide you with the relevant paragraphs in batches during the upcoming editing work. Each batch of ** paragraphs will begin with a specific number, which consists of a ' symbol and a subsequent Arabic numeral, e.g. '8', etc. Your task is to carefully write down each batch of ** paragraphs and their corresponding numbers. Please note that there is no need for any form of analysis or summary to be provided when recording these paragraphs, only to reply 'documented' after each batch of information has been recorded. When I send the '*Entry End' command, it means that all ** paragraphs are fully available, and you should reply 'Everything has been recorded'. Focus on accurate records and avoid doing anything beyond this indication. Now, be prepared to receive and record the ** paragraph that will be provided.
An added benefit of this is:GPT can be easily located。As mentioned earlier, this is equivalent to "naming" each piece of contentThen you can ask questions about the content of the specified number, which is more accurate and labor-saving.
However, it is still important to note that the memory space of GPT is limited.
You'd better be rightWhat has already been enteredto make a check, for example:
If there is nothing wrong with the inspection at this time,Then you can open a new branch and ask questions and polish them.
If the text is still too long to be observant,Let it answer what exactly has been changed.
Specifically, add such a paragraph at the end of the prompt.
prompt：furthermore, list all modifications and explain the reasons for doing so in the markdown table.
Hint: In addition,List all changes in markdown**and explain the reasons for these modifications.
Then, after giving the results of the editing, some before and after comparisons will be listed.
When dealing with complex revisions involving multiple paragraphs, sentences, or text, clear positioning and description are a must.
If it is involved in the course of a revisionMultiple paragraphs, multiple sentences, multiple texts, etc., need to be located.
For example, in the course of revision of paragraphs involving different versions,Avoid vague expressions such as "previous", "previous", or "original".
Instead, a more explicit identification should be adopted, such as:
Use "target paragraph", "source paragraph", "reference paragraph" to explicitly refer to.
Or specify it as "Paragraph A", "Paragraph B", "Paragraph C".
If you want GPT to polish the content according to the style of the journal meeting.
prompt: if i wish to publish a **at a 「x」conference, please polish the above content in the style of a xxx article.
Tip: If I wish to publish ** in an X Conference journal, please polish the content in the style of the X article.
Note: The journal or conference should be well-known.
For example, top journals such as Nature, Science, and The Lancet, I believe GPT can handle it well enough.
If the target journal or conference is relatively less well-known, two strategies can help:
1) Find a well-known journal with a similar style to the journal.
If the style of the target journal is not widely known, a well-known journal with a similar style can be found as a reference.
2) Provide relevant content directly, as an example.
prompt: here is a sample introduction from an article in「x」journal: “insert the introduction of the sample article here]”.please polish the introduction of my **following the style and structure of this sample.
Tip: This is a sample article for an introduction section of the X journal: "[insert introduction section of sample article here]". Please refer to the style and structure of this sample to polish my ** introduction.
Understand that whether it is in the style of the journal or providing examples, it is essentially givingGPT's response provides a frame of referenceThis is also the most basic principle to guide GPT answers.
It's a good idea to check the logic of your article after polishing it.
i want you to act as a fallacy finder. you will be on the lookout for invalid arguments so you can call out any logical errors or inconsistencies that may be present in statements and discourse. your job is to provide evidence-based feedback and point out any fallacies, faulty reasoning, false assumptions, or incorrect conclusions which may h**e been overlooked by the speaker or writer.
Hint: I want you to play the role of the finder of the fallacy. Be aware of invalid arguments and point out any logical errors or inconsistencies that may exist in your statements and discourses. Your job is to provide evidence-based feedback pointing out any fallacies, faulty reasoning, false assumptions, or incorrect conclusions that the speaker or writer may have overlooked.
You think it's gpt4 now0 evolved to now only be used for editing
Then you areBig mistakeNow I can write it for you directly!
Let's try these prompts!
chatgpt prompts for academic writing
This is a repository on github, very detailed and easy to use prompts for academic writing, basically every feature has a side-by-side prompt, in no particular order, try and then decide)
Write the title topic sentence
1. suggest 5 titles for the followingabstract: [abstract paragraph]
2. write a topic sentences forthis paragraph: [paragraph]
Write keywords
1. provide 5 keywords for this: [paragraphs]
Write an abstract
1. generate an abstract for a scientific***based on this information for:[paragraphs]
Write an outline
1. generate an outline for [topic sentence]
2. i want to write a journal article about [topic sentence]. give me an outline for the article that i can use as a starting point.
Write an introduction
1. come up with an introduction for the followingresearch topic: [topic sentence]
Write a literature review
1. conduct a literature review on [topic sentence]and provide review **references
2. provide me with references and links to **s in [parapgraph]
3. summarize the scholarly literature includingin text citations on [paragraphs]
4. write this in standart harvardreferencing [paragraph]
5. convert this [bibliography] from mla to apa style.
6. compare and contrast [theory1]and [theory2] in the context of[research domain]:
Written by: **methodology
1. create objectives and methodologyfor [topic sentence]
2. write a detailed methodology forthe topic: [topic sentence]
3. analyze the strengths and weaknesses ofthis methodology: [paragraphs]
4. write objectives for this study: [topic sentence]
5. what are the limitations of using[topic sentence] in [research domain]?
6. create a recipie for the methods used in this [paragraphs]
7. suggest interdisciplinary approaches to [topic sentence]
8. explain how qualitative/quantitative research methodscan be used to address [research questions]
9. recommend best practices for data collectionand analysis in [topic sentence]
Write experiments sections
1. design an experiment that [action]
Write the results section results
1. write a result section for the following paragraphs.please write this in third person. [paragraphs]
Write a discussion
1. discuss this results: [result paragraphs]
Write a conclusion
1. generate a conclusion for this: [paragraphs]
2. give recommendations and conclusion for: [paragraphs]
Thanks for reading!
If it really helps you, please don't be stingy with likes, favorites and comments, and welcome to more friends.