I am thinking of fine-tuning a model which is used predominantly for function calling.
There are some functions I haven’t added to the model yet, but want to add further down the line.
Because I wouldn’t be finetuning the model on any data or examples of this function, would I be able to add this functionality further down the line without causing any problems, or would the model be unable to adapt to a new function that it wasn’t specifically trained on, if that makes sense?