'Generate MCQ's using GPT-J

I have multiple pdf's from which I am extracting text into a text file. I then convert the text file into a one column csv file which I am feeding to GPT-J. My aim is to generate MCQ's based on the text inside the csv file. But the model spits out gibberish text during inferencing.

The reason is because the text inside the csv file needs to be in a specific format in order for the model to generate proper MCQ's. The issue is that I don't know what exact data format the model is expecting. If anyone has worked on QnA generating using GPT models, your advice would really be helpful.

Thanks for your time!



Sources

This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.

Source: Stack Overflow

Solution Source