Hey all, would love some advice, I am planning to build a system which takes in some text as an input and identifies sections of that text which might fall foul of some specific legal regulations.
This system will obviously need to be trained or have access to the files/data containing the regulations in order to make a judgement about the input text. The tool is designed to be used internally by members of the team so there won’t be any prompting from the public or end users.
I’ve seen that I could go two ways, using fine-tuning or using an Assistants.
So my questions are:
- What are the benefits of using the fine-tuning vs Assistants API?
- In the case that regulations change, which is easier and cheaper to “retrain”?
- The input data could be sensitive, but the tool is only going to be used internally. Is it possible that the input files could leak outside of my organisation?
Any advice or help on this would be much appreciated!