I have given instructions to prompt to generate 10 questions and answers

i have given instructions to prompt to generate 10 questions and answers of aptitude. problem is it generate wrong answer of some questions. the current accuracy is 70%. how to increase it? (p.s.- i am using gpt4-turbo)

Hi there and welcome!

In order for us to give you some ideas on what might be causing this, you have to help us out and provide a few more specifics. Perhaps you can share the prompt and/or provide details on the type of questions where the model fails?

1 Like

perform following task:-

-Generate 10 aptitude multiple-choice questions with correct answers, covering logical reasoning,

numerical problem solving, and blood relations.

-Do not give image related questions.

-Each question should follow the format with keys “Question”,

“Options”, “Skill”, and “Answer” and your output should be in an array of JSON format,but you should not explicitly mention the JSON structure in your response

-Format your response as follows:

Question: .

Options: [




D) ]



-Strictly ensure the correctness of each question, options and answer.

-Your main focus should be on 100 percent accuracy of answer.

this is the prompt

And so which part does the model get wrong?


it is not giving correct answers.
like in this question it generate the question and give 13 as a answer but answer should be 10.
“Question”: “10. A sequence of numbers is given: 7, 10, 8, 11, 9, 12, … What number should come next?”,
“Options”: [
“A) 7”,
“B) 10”,
“C) 12”,
“D) 13”
“Skill”: “Numerical Problem Solving”,
“Answer”: “D) 13”

I see - the main challenge is that you are asking the model to perform too many things at the same time. The model is not given enough time to reflect on and determine the right answer.

Ideally, you’d be able to break down the task into smaller tasks, either by significantly lowering the number of questions you are asking it to generate in one go or by splitting the question and choice generation from the answer generation.