def from_files(cls, reviews_file: str, labels_file: str, vocab: Vocabulary = None):
with open(reviews_file, "r") as reviews, open(labels_file, "r") as labels:
review_lines = reviews.readlines()
label_lines = labels.readlines()
examples = [
{
# review is text, stored as a list of tokens
"review": review_lines[line].strip("\n").split(" "),
"label": int(label_lines[line].strip("\n")),
}
for line in range(len(review_lines))
]
# initialize a vocabulary from the reviews, if none is given
if not vocab:
vocab = Vocabulary.from_text_files(
[reviews_file],
special_tokens=(Vocabulary.UNK, SSTClassificationDataset.PAD),
)
return cls(examples, vocab)
