why mix the dev(validate) dataset and the train dataset

[SeekPoint]

    f_train = open("%s/chat.in" % data_dir, 'w')
    f_dev = open("%s/chat_test.in" % data_dir, 'w')
    for line in f_zip:
        f_train.write(line)   ---so all data in train dataset and dev dataset included ????
        if n < 10000:     
            f_dev.write(line)
            n += 1