Torchtext: not able to convert to floats

Similar to #78 on github issue of torchtext.

!cat data/split_test.csv
' tries to force its quirkiness upon the audience .,1
's not nearly enough that 's right,0
entered the bizarre realm,1
torch.__version__, torchtext.__version__
('0.4.0', '0.2.3')
PHRASE =  data.Field(batch_first = True, lower = True)
SENTIMENT = data.LabelField(tensor_type=torch.double, batch_first = True, use_vocab = False, postprocessing=data.Pipeline(lambda x: float(x)))

fields = [('phrase', PHRASE), ('sentiment', SENTIMENT)]
train, valid, test = data.TabularDataset.splits(
                path = 'data',
                train = 'split_train.csv',
                validation = 'split_valid.csv',
                test = 'split_test.csv',
                format = 'csv',
                fields = fields,
                skip_header = True

{'phrase': ['likely', 'to', 'leave', 'a', 'lasting', 'impression'],
 'sentiment': '3'}

i want it to be something like

{'phrase': ['likely', 'to', 'leave', 'a', 'lasting', 'impression'],
'sentiment': 3}

it doesn’t seem to work, or maybe im doing something wrong, i’d be happy to be corrected. Thanks in advance.