Not sure I understand fully re: alignment, but fine-tuning is usually the most performant way to specialize when you have the data for it.

→ View original post on X — @goodside