Why do we have 4 splits train.tsv, train_dev.tsv, devel.tsv, test.tsv? And is the train_dev.tsv merged by train.tsv and devel.tsv? And in the run_ner.py for example, why the train_dataset is transformed from train_dev.txt and eval_dataset is transformed from eval.txt? Thanks.
Why do we have 4 splits train.tsv, train_dev.tsv, devel.tsv, test.tsv? And is the train_dev.tsv merged by train.tsv and devel.tsv? And in the run_ner.py for example, why the train_dataset is transformed from train_dev.txt and eval_dataset is transformed from eval.txt? Thanks.