Need Help With Prompts? Ask me*

Prompting
64

yes or write a shorter, better prompt

or split it into multiple prompts

again i need to see what they are doing to advise

65

Hi Josh,
Can you help me with my case?

I want to upload Korean standard documents and make ChatGPT to remember them. When I send a paragraph written by an other language, ChatGPT can translate them to Korean base on the documents that I uploaded.
I’ve tried many Prompts but it does not work well. And I realized it seems that ChatGPT cannot remember previous conversation

Prompts that I used:

  1. Suppose you worked as an interpreter/translator for an architectural office for 20 years. Now that you have studied the contract I uploaded, please translate the contract below into Korean.

  2. I want you to act as an Korean translator, spelling corrector and improver. I will speak to you in any language and you will detect the language, translate it and answer in the corrected and improved version of my text, in Korean. I want you to replace my simplified A0-level words and sentences with more beautiful and elegant, upper level Korean words and sentences. Keep the meaning same, but make them more literary. I want you to only reply the correction, the improvements and nothing else, do not write explanations

Can you give me some advices to resolve my problem? Thank you so much

66

Not sure how well openai supports Korean or how much of that language is in the corpus

It’s all statistical. Give it more Korean and it will produce more… But perhaps not enough.

68

This is likely too much for one prompt to do

You need to break this down into a self aware AI with different mental functions

For more, private message me

69

Hi Josh, I need your help to improve the prompt which is used to rewrite product based on description and product title. The result need to meet some conditions and it must avoid plagiarism from the description provided. Here is the current prompt that I used “”"
Please create only 1 paragraph limited to 250 words based on the given product title and description. Ensure that the paragraph meets the following conditions:
1. Remove any mention of payment details, including methods and options.
2. Omit all information related to return policies and procedures.
3. Exclude any references to feedback, both positive and negative.
4. Leave out any content about the company (“us”) and its background.
5. Do not include any references to copyright or legal information.
6. Remove any mentions of warranty terms or coverage.
7. Exclude all information about shipping, including options and times.
8. Exclude any references to other marketplaces like eBay or Amazon.
9. Remove any mentions of ASIN number.
10. Add bullet points to mention key specifications and features of the product.
11. Remove all brand names except the brand name is Power Wheels if product is not identified as spare part.
12. Replace the shop name with “Azautodetailing”.
13. Specific Instruction: Insert the following HTML element exactly as shown, make it blend with the paragraph. This element must contain the phrase “ride-on” and link to “”:
ride-on
14. Present the output in HTML format.
Product Title: {title}
Product Description: {current_description}
Answer:“”"

70

That’s too complex

You need to break that down into steps

And you cannot tell it what not to do

Only what to do

3 Likes
71

Firstly, thank you for offering this thread to ask questions. I am brand new to ChatGPT (but not to computers in general), and thus far I think I get the basics of the prompting. This thread has been very helpful.

I am trying to get ChatGPT to write an assignment based on this prompt:

“Create an assignment for college sophomores based on the following overview: Research and discuss the impact of historical events on the development of nursing as a profession. Also create a grading rubric for this assignment.”

It gives me the high-risk words error:

"Prompt contains high-risk words

As part of our mission to develop AI in a responsible way, we’re starting to block prompts in Playground that have a high risk of producing unsafe content. We recommend taking similar steps to filter content in your application.

If your use case requires the ability to use these words, contact us through our help center."

I have tried rephrasing this over and over but I can’t figure out what the “high-risk” words or phrases might be.

Any ideas on what in that prompt might be triggering the high-risk words warning?

Thanks!

72

you’d have to ask open AI about that

but i bet they do not let you do anything related to schoolwork in there to prevent students from using it to make assignments

73

In case you are still available, I have a seemingly tricky one: when uploading a file to code interpreter the model will very often take the file and only read the first 500-1000 characters. It will then report something along the lines of “the file is truncated” or keep working with this limited understanding until you suddenly receive a reply that is missing pretty much all relevant context from the uploaded file.
Then I blame the model for not reading the file completely and then it proceeds to do just that.

I could really use a prompt to make Code Interpreter read the whole file the first time, every time I ask it to do so.

edit: word

74

Yeah I have noticed that, could be the file type / format

Could be that openai is trying to save costs

Not sure a way around it but I am using code interpreter a lot these days

You can also use the playground if things are under token counts

75

Oh I see. That makes sense.

What I am trying to prompt it to do is create more details for the assignment. I am a teacher and I am investigating how to get GPT to build a class assignment (not a student asking it to write the essay).

I tried adding more details such as grading criteria and a rubric but it gets stuck with this error.

How might I phrase this so that ChatGPT knows that I am asking it for the assignment details, not for it to do the actual assignment?

Thank you for your help!

76

Preface: ChatGPT is the chatbot product at chat.openai.com
The “playground” at platform.openai.com is an API evaluation tool.

The API playground has a safety filter that scans inputs that cannot be disabled. It is over-reactive, for example flagging a list of 1000 image categories but not when broken into two 500 element lists. The text you show is like what playground produces, not ChatGPT.

You would not see these alerts when developing your own software.

Prompts and outputs (when writing end-user API software) should still be submitted to the separate “moderations” endpoint if allowing unknown users. This is a separate safety system than the one you are triggering, and you can choose to submit playground data to this with the upper-right “Content filter preferences”.

With system prompt “You are a teacher’s assistant”, this nursing prompt ran fine in both ChatGPT and on the Playground for me.

1 Like
77

i understand :slight_smile: Short answer: no idea.

I tried it here in the playground selecting GPT4 and temperature of 0 and get this

Assignment:

Title: The Impact of Historical Events on the Development of Nursing as a Profession

Instructions:

  1. Choose three significant historical events that have influenced the development of nursing as a profession. These events can be from any period in history and from any part of the world.

  2. For each event, provide a detailed description of the event itself, the context in which it occurred, and the immediate impact it had on nursing at the time.

  3. Discuss how each event has influenced the development of nursing as a profession in the long term. Consider aspects such as nursing education, nursing roles, nursing practices, and public perceptions of nursing.

  4. Use at least five scholarly sources to support your discussion. These sources may include books, peer-reviewed journal articles, and reputable online sources.

  5. Your paper should be 8-10 pages long, double-spaced, and formatted according to APA style.

  6. Submit your assignment via the course’s online portal by the due date.

Grading Rubric:

  1. Selection of Historical Events (20 points): The student has chosen three significant historical events that have clearly influenced the development of nursing as a profession.

  2. Description of Events (20 points): The student has provided a detailed description of each event, including the context in which it occurred and its immediate impact on nursing.

  3. Discussion of Long-Term Impact (30 points): The student has thoroughly discussed how each event has influenced the development of nursing as a profession in the long term.

  4. Use of Scholarly Sources (15 points): The student has used at least five scholarly sources to support their discussion.

  5. APA Formatting (5 points): The paper is formatted according to APA style, including in-text citations and a reference list.

  6. Writing Quality (10 points): The paper is well-written, with clear organization, proper grammar and spelling, and effective use of language.

Total: 100 points.

1 Like
78

Thank you both very much!

Yes, I am working on my own software using the playground and by reading your two posts, I am realizing that I have to take a step back and catch up on tutorials. :slight_smile:

79

Just use it in the playground in a work fine for exactly what you want to do just make sure temperature is set at zero and frequency penalty and presence penalty is set at zero

I also have the idea of saying first you are a teaching assistant AI but given that the model is stochastic AKA random that might not always work. This is why I did not recommend it.

The playground would be the most stable area to use

81

You should likely be using dynamic prompts for this

Hard code the hard code parts

Fit it into the prompt so it “knows” what it said

Lmk if you need anymore help

84

that depends on a few different factors

if you want to private message me i can try to help you out further

1 Like
92

Folks, there are no rules being broken on the forum in this thread, so far.

Please keep everything friendly and professional.

1 Like
96

The forum should be a place for all developers of AI products that make use of OpenAI services to come and chat about all aspects of the dev cycle, this will naturally attract those looking for developer services and other developers looking for help on projects.

First and foremost this space should be a place for the free exchange of ideas and best practices around OpenAI services, along with that will be a degree of commercial interest and involvement, those treating the forum as a valuable asset in their journey with AI are welcome, and those that abuse it will be removed.

1 Like
101

We live in a free world and everyone have right to think and be vocal with their opinions and I respect everyone in the community and their opinions.

Apologies, I didnt knew my question can ignite the conversations.
I appreciate the help Josh, was not in the intension of free code service. I did not understood the depth of the work needed. I was hoping for an elaborated guidance.

I am a newbie who is tired getting scammed enrolling in different GPT Prompting Courses online. I thought to look into the community and came across your thread.

It was just an example for learning purpose, thought a project or purpose can make my learning more enjoyable and it will be presentable to share with my professor. I will take a simpler examples.