def processLine(self, row):
text, label, label_name = row
text = " ".join(text.split()[:self.max_length])
x = get_indices_tensor(text.split(), self.word_to_indx,
self.max_length)
sample = {'text': text, 'x': x, 'y': label, 'y_name': label_name}
return sample
def processLine(self, line, i):
if isinstance(line, bytes):
line = line.decode()
label_t = [float(v) for v in line.split('\t')[0]]
label = int(label_t[0])
text_list = line.split('\t')[-1].split()[:self.max_length]
text = " ".join(text_list)
self.args.num_class = 2
x = get_indices_tensor(text_list, self.word_to_indx, self.max_length)
sample = {'text': text, 'x': x, 'y': label, 'i': i}
return sample
def processLine(self, line, i):
if isinstance(line, bytes):
line = line.decode()
label = float(line[0])
if self.objective == 'mse':
raise NotImplementedError(
"SST2 dataset only allows binary classification")
else:
self.args.num_class = 2
text_list = line[2:].split()[:self.max_length]
text = " ".join(text_list)
x = get_indices_tensor(text_list, self.word_to_indx, self.max_length)
sample = {'text': text, 'x': x, 'y': label, 'i': i}
return sample
def processLine(self, line, aspect_num, i):
if isinstance(line, bytes):
line = line.decode()
labels = [ float(v) for v in line.split()[:5] ]
if self.objective == 'mse':
label = float(labels[aspect_num])
self.args.num_class = 1
else:
label = int(self.class_map[ int(labels[aspect_num] *10) ])
self.args.num_class = 3
text_list = line.split('\t')[-1].split()[:self.max_length]
text = " ".join(text_list)
x = get_indices_tensor(text_list, self.word_to_indx, self.max_length)
sample = {'text':text,'x':x, 'y':label, 'i':i}
return sample