Note 1. Main root dataset of files

Note 2.1. Segmentation of the root dataset into discrete, conceptual chunks

Note 2.2. Combination of multiple chunks, randomly over a uniform distribution to allow multi-hop

Note 3.1. Single hop questions formed By various models on individual chunks

Note 3.2. Multi hop questions formed by various models over chunk combinations

Note 4.1. Dataset of all generated questions, from all models

Note 4.2. Dataset of deduplicated questions, with eps value of 0.18, with over 90% shrinkage while maintaining reasonable diversity