What is the best way to approach future improvements for Large Language Models?

[erindreams]

There are several ways to approach Large Language Models, so I wanted to compare experiences.

I am particularly interested in future improvements and common mistakes. I would like to compare approaches that people have actually used rather than just theoretical suggestions.

Are there any settings, tools, or workflow changes that made a noticeable difference?