Results of the pre-test

We used a traffic light system to evaluate the result of the pre-test of the individual models. If some of the examples (2 or 3) are recognized, we assume that fine-tuning could help and award a "yellow" rating. If only 0 or 1 example is recognized, we award a red rating. If 4 or 5 of the examples are recognized correctly, we award a green rating. Models rated yellow and red are not tested further for the time being.