Different visualisations of the metrics used to evaluate the performance of the O’TRAIN model trained on the JAST-STDPipe data cube. Top left: Evolution of the MCC and F1-score as a function of different values of p_t. The value p_t = 0.51 (best threshold) maximises both the MCC and F1-score. Top right: Classification probability distribution of the validation data set at the last epoch of the training. Bottom left: Evolution of the accuracy and loss metrics as a function of the number of training epochs. Bottom right:) ROC curve per bin of the OT candidate magnitudes in order to diagnose the behaviour of the O’TRAIN model for different source brightness.