Discrepancy between "annotation_full.tsv" and the actual reviews

While the article identifiers in "annotation_full.tsv" are in the range of 0-426, the identifiers in the reviews (/data/iclr_2017/train/reviews/) are somewhere between 300-700. When running "assign_annot_iclr_2017.py", it ends up matching annotations to incorrect reviews.