Which is better for finetuning h2oai/h2ogpt-4096-llama2-7b-chat or h2oai/h2ogpt-4096-llama2-7b on prepared question-answer pairs?

[sunilswain]

Hey
Let's assume I'm preparing a chatbot which would answer my queries based on my personal data. I've prepared around 4000 question-answer pairs for it. Which model should I choose as the backbone in my finetuning process?

Thanks

[psinger]

There is no clear-cut answer, you would need to try multiple options, but chat model might be the better choice.