A strength of around 1.5 for Lola is recommended.
It is possible to generate with just the prompt. The learning is kept weak to minimize the impact on the style as much as possible.