Notes. Shown for a subset of fifteen clusters; the full version with all detections and non-detections of all simulated clusters, as well as additional columns, is available online at the CDS. Clusters may be detected at multiple m_clSize values by HDBSCAN, with each detection of a cluster being assigned a unique three-character detection ID. As HDBSCAN may report identical clusters even at different m_clSize values, these IDs were deduplicated – meaning that one cluster at various m_clSize values may share the same detection ID and parameters if they were identical. Clusters not detected at any m_clSize value have no detection ID. The ‘Best?’ column can be used to cut the cluster list to only the best possible detection of each cluster, which was the dataset used to train our model (in addition to later also excluding some clusters based on their X coordinate; see text).