Why use test_index as val_mask

[ElektraChen]

Totally confused for there:
train_mask[train_index] = True
test_mask[test_index] = True
val_mask[test_index] = True
In pretrain_and_tune.py, line 290-292.