fine-tune task with uploaded jsonl file

[quincy451]

Looking to convert a number of pdf files into jsonl files..upload them into chatgpt run a fine-tune task with them, then run a conversation about what is in the jsonl file. I started with various python scripts that gave a jsonl file, but chatgpt didn't seem to make much sense of that. So I went to more basic functional test. I created this for a jsonl file:
{"prompt": "gunta program 1", "completion": "print hello world program 1"}
{"prompt": "gunta program 2", "completion": "print hello world program 2"}
{"prompt": "gunta program 3", "completion": "print hello world program 3"}
The idea is if you mention gunta to chatgpt after the fine tune task and it acknowledging knowing anything about it...you have had an effect on the system. But alas when I do this chatgpt never knows anything about gunta.
Any advise on how to setup the jsonl file or maybe how to convert the pfd to jsonl. I have found the gpt4 extension askyourpdf which will allow you do upload a pdf get an ID paste that ID into chatgpt and your know talking about your doc subject. But I want to do this in an application embedded situation.