I have seen some comments on the web as to why someone would fine-tuning the new Small Language Model from OpenAI.

I guess the biggest reason for someone to do this is if their organisation is already a corporate type user of OpenAI.

But, I must say fine-tuning is really easy and the results are really encouraging. Especially if an organisation have some standard contextual data (which does not change much) imbedded within the model.