108 lines
3.3 KiB
Python
108 lines
3.3 KiB
Python
import pandas as pd
|
|
import torch
|
|
from gensim.models import KeyedVectors
|
|
import nltk
|
|
import csv
|
|
import numpy as np
|
|
|
|
|
|
def tokenize_data(data):
|
|
data_tokenize = [nltk.word_tokenize(x) for x in data]
|
|
|
|
for doc in data_tokenize:
|
|
i = 0
|
|
while i < len(doc):
|
|
if doc[i].isalpha():
|
|
doc[i] = doc[i].lower()
|
|
else:
|
|
del doc[i]
|
|
i += 1
|
|
return data_tokenize
|
|
|
|
|
|
class NeuralNetwork(torch.nn.Module):
|
|
def __init__(self, input_size, hidden_size, num_classes):
|
|
super(NeuralNetwork, self).__init__()
|
|
self.l1 = torch.nn.Linear(input_size, hidden_size)
|
|
self.l2 = torch.nn.Linear(hidden_size, num_classes)
|
|
|
|
def forward(self, x):
|
|
x = self.l1(x)
|
|
x = torch.relu(x)
|
|
x = self.l2(x)
|
|
x = torch.sigmoid(x)
|
|
return x
|
|
|
|
|
|
r_in = './train/train.tsv'
|
|
|
|
r_ind_ev = './dev-0/in.tsv'
|
|
tsv_read = pd.read_table(r_in, error_bad_lines=False, quoting=csv.QUOTE_NONE, sep='\t', header=None)
|
|
tsv_read_dev = pd.read_table(r_ind_ev, error_bad_lines=False, sep='\t', quoting=csv.QUOTE_NONE, header=None)
|
|
tsv_read_test_in = pd.read_table('./test-A/in.tsv', error_bad_lines=False, quoting=csv.QUOTE_NONE, header=None)
|
|
|
|
y_train = tsv_read[0].values
|
|
X_train = tsv_read[1].values
|
|
X_dev = tsv_read_dev[0].values
|
|
X_test = tsv_read_test_in[0].values
|
|
|
|
X_train = tokenize_data(X_train)
|
|
X_dev = tokenize_data(X_dev)
|
|
X_test = tokenize_data(X_test)
|
|
|
|
|
|
model = KeyedVectors.load("./word2vec/word2vec_100_3_polish.bin")
|
|
|
|
X_train = [np.mean([model[word] for word in content if word in model] or [np.zeros(100)], axis=0) for content in X_train]
|
|
x_train_vectors = np.array(X_train, dtype=np.float32)
|
|
x_train_tensor = torch.tensor(x_train_vectors.astype(np.float32))
|
|
|
|
X_dev = [np.mean([model[word] for word in content if word in model] or [np.zeros(100)], axis=0) for content in X_dev]
|
|
x_dev_vectors = np.array(X_dev, dtype=np.float32)
|
|
|
|
|
|
X_test = [np.mean([model[word] for word in content if word in model] or [np.zeros(100)], axis=0) for content in X_test]
|
|
x_test_vectors = np.array(X_test, dtype=np.float32)
|
|
|
|
model = NeuralNetwork(100, 200, 1)
|
|
criterion = torch.nn.BCELoss()
|
|
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)
|
|
|
|
batch_size = 12
|
|
|
|
for epoch in range(6):
|
|
loss_score = 0
|
|
acc_score = 0
|
|
items_total = 0
|
|
model.train()
|
|
for i in range(0, y_train.shape[0], batch_size):
|
|
X = x_train_vectors[i:i + batch_size]
|
|
X = torch.tensor(X.astype(np.float32))
|
|
Y = y_train[i:i + batch_size]
|
|
Y = torch.tensor(Y.astype(np.float32)).reshape(-1, 1)
|
|
|
|
Y_predictions = model(X)
|
|
acc_score += torch.sum((Y_predictions > 0.5) == Y).item()
|
|
items_total += Y.shape[0]
|
|
|
|
optimizer.zero_grad()
|
|
loss = criterion(Y_predictions, Y)
|
|
loss.backward()
|
|
optimizer.step()
|
|
|
|
loss_score += loss.item() * Y.shape[0]
|
|
|
|
|
|
|
|
predictions = model(torch.tensor(x_dev_vectors.astype(np.float32)))
|
|
predictions = predictions.cpu().detach().numpy()
|
|
predictions = (predictions > 0.5)
|
|
predictions = np.asarray(predictions, dtype=np.int32)
|
|
predictions.tofile('dev-0/out.tsv', sep='\n')
|
|
|
|
|
|
predictions = model(torch.tensor(x_dev_vectors.astype(np.float32)))
|
|
predictions = predictions.cpu().detach().numpy()
|
|
predictions = (predictions > 0.5)
|
|
predictions = np.asarray(predictions, dtype=np.int32)
|
|
predictions.tofile('test-A/out.tsv', sep='\n') |