text similarity for German texts - which language model to use and how to train it?

Question

I want to compare a German text (1 or more Sentences or whole paragraph of business conversation dokument) with a set of around 1000 predefined German text blocks for similarity