It is mentioned in the IRN paper (https://arxiv.org/abs/1801.04726) that the data is split in a ratio 8:1:1. Could someone please confirm whether it was a random split or dataset is split in any specific manner?
Also, are the accuracy values in Table 2 reported in the paper calculated on test data only?
It is mentioned in the IRN paper (https://arxiv.org/abs/1801.04726) that the data is split in a ratio 8:1:1. Could someone please confirm whether it was a random split or dataset is split in any specific manner? Also, are the accuracy values in Table 2 reported in the paper calculated on test data only?