Distribution of predicted scores by Zoobot1 (left) and Zoobot2 (right) for lens candidates across all grades. Grade A lens candidates are confidently identified by both networks, shown to the right of the distribution in red at a high value of p_LENS. Grade B predicted scores are shown in blue and are more varied for Zoobot2. The predicted scores for grade C lens candidates have the most variation amongst grades, shown in green, for both networks illustrating that Zoobot1 and Zoobot2 are not as confident in identifying grade C lens candidates.