• To start a training run use **lstm_training.py** with custom parameters like number of LSTM units, dropout, IOB file path etc. . You can call the important scripts with -h to get help. All output of a training run will land in the *modelzoo* directory. To configure and run a training via a parameter grid use (just change for docker) **runner.py**. To get an overview of the performance of the trained models via **runner.py** it will generate an csv formatted file containing metrics that can be visualised with **swarmplot_scores.py**. The *modelzoo* directory contains examples (only one model was committed to this repo).