Based on parlbert-topic-german, this model was trained on german legal data using Masked language Modelling to get a model that can handle legal texts.
This model needs to be finetuned for the final use case (Classification, Inference, ...), it is not meant to be used directly.