Best model for generating multiple choice questions

I have a project where I randomly generates a multiple choice questions about programming language like c++, and if the student choose the wrong answer then the model should generates a new question similar to the old one.

Q1-What is the best suitable model for this task?

2-I couldn’t find any dataset which contains a multi-choice questions about c++, so I decided to generate one by following these steps:
2.1-I’ll give a smart AI Model like ChatGPT 10-15 MCQuestions about certain topic like pointers.
2.2-ask it to generate 10-20 more questions about the same topic.
2.3-give the generated question to another smart AI Model like copilot and ask whether or not they are good and related to the same topic.
2.4-if yes then I’ll add them into the Dataset file in this format :
Question, Option a, Option b, Option c, Option d, Correct Answer, Topic, Chapter, Level.

Q2-is the previous steps a good way to build a high quality dataset? If not, where do I find one, or How to build one from scratch?

Hi, did you find anything good?

Maybe can check out these models?

  • Phind-CodeLlama-34B-v2
  • T3Q-ko-solar-dpo-v6.0
  • t5-small-squad2-question-generation
  • quora-distilroberta-base
  • t5-base-finetuned-question-generation-ap