reviews.append(
sentence2tensor(words, w2i, padding_idx, sent_length))
if len(words) > sent_length:
lengths.append(sent_length)
else:
lengths.append(len(words))
labels.append(int(label))
if count % 100000 == 0:
print('Encoded reviews: ', count)
return reviews, labels, lengths
if __name__ == '__main__':
train_file, validation_file = '../Data/train.csv', '../Data/validation.csv'
w2i = pp.obtainW2i(train=train_file, validate=validation_file)
print('Loaded vocabulary')
w2i['<PAD>'] = 0
vocab_size = len(w2i)
padding_idx = 0
sent_length = 80
translator = str.maketrans('', '', string.punctuation)
embedding_size = 50
embedding = nn.Embedding(vocab_size,
embedding_size,
padding_idx=padding_idx)
print('Embeddings calculated')
encoder_hidden = encoder.initHidden()
input_length = sentence_tensor.size(0)
for ei in range(input_length):
output, encoder_hidden = encoder(sentence_tensor[ei],
encoder_hidden)
output = torch.round(output)
if torch.equal(output, label_tensor):
accuracy += 1
return accuracy / len(test_sentences)
if __name__ == '__main__':
# file name for male reviews and female reviews
vocabulary, w2i, sentences_m, sentences_f = pp.obtainW2i(
"../Data/sample_male", "../Data/sample_female")
train_senetences, train_labels, test_sentences, test_labels = pp.testTrainSplit(
sentences_m, sentences_f)
hidden_size = 20
input_size = len(w2i)
output_size = 1
encoder = Encoder(input_size, hidden_size, output_size)
encoder = encoder.to(device)
batch_train(encoder, train_senetences, train_labels, 3, w2i)
#accuracy = evaluate(encoder,test_sentences, test_labels,w2i)
#print(accuracy)