why the new version of allennlp's datasetloader doesn't have the __add__ method?

Question

is it canceled? then how can I add two dataset for build a vocab? in allennlp1.0, I can do

reader = StanfordSentimentTreeBankDatasetReader()
train_dataset = reader.read('train.txt')
dev_dataset = reader.read('dev.txt')

for inst in train_dataset + dev_dataset:
    print(inst)
    break

and then build a vocab together, but why allennlp 2.0 canceled it? sorry for being a fresh in allennlp

Answer 1

DatasetReader.read() now just returns an iterator. So you could accomplish the same thing like this:

from itertools import chain

for inst in chain(train_dataset, dev_dataset):
    print(inst)