That is very, very hard to do in an objective manner, as the current LLM benchmark gaming demonstrates.
Sure, you can deploy a smaller model to production to get real-world user data and feedback, but a) deploying a suboptimal model can give a bad first impression and b) the quality is still subjective and requires other metrics to be analyzed. Looking at prediction probabilities only really helps if you have a single correct output token, which isn't what LLM benchmarks test for.
That is very, very hard to do in an objective manner, as the current LLM benchmark gaming demonstrates.
Sure, you can deploy a smaller model to production to get real-world user data and feedback, but a) deploying a suboptimal model can give a bad first impression and b) the quality is still subjective and requires other metrics to be analyzed. Looking at prediction probabilities only really helps if you have a single correct output token, which isn't what LLM benchmarks test for.