I am imagining this as a histogram plot with the (actual_positives / number of candidates) ratio as x axis and frequency in the dataset as y axis, maybe with a vertical line indicating the mean. This is a nice supplemental figure to show that generally most candidates are not in the sentence, illustrates the naive baseline for per-sample accuracy.
I am imagining this as a histogram plot with the (actual_positives / number of candidates) ratio as x axis and frequency in the dataset as y axis, maybe with a vertical line indicating the mean. This is a nice supplemental figure to show that generally most candidates are not in the sentence, illustrates the naive baseline for per-sample accuracy.