Duplicates are defined as cursor positions that have an approximate Jaccard similarity greater than 0.5 (over token 5-grams). From the resulting clusters of near-duplicates, we select up to N examples that are maximally different from each other. Release Notes: - N/A