def testAccuracy(self, x_test, y_test, q_pred, headId):
acc = 0
num_pred_samples = 100
for x_test_batch, y_test_batch in self.getBatch(x_test, y_test):
monteCarlo = MonteCarlo(q_pred, num_pred_samples)
y_pred_batch = monteCarlo.computeMonteCarlo(x_test_batch, headId)
_, y_pred_batch = torch.max(y_pred_batch.data, 1)
y_pred_batch = torch.eye(
self.dataGen.get_dims()[1])[y_pred_batch].type(FloatTensor)
acc += torch.sum(torch.mul(y_pred_batch, y_test_batch)).item()
return acc / y_test.shape[0]
parameters = vanillaNN.getParameters()
qPrior = ParametersDistribution(sharedDim, headDim, headCount)
qPrior.setParameters(parameters, 1)
parameters = qPrior.getFlattenedParameters(1)
for i, (images, labels) in enumerate(trainLoader):
images = images.reshape(-1, 28 * 28).to(Device)
yOnehot = _onehot(labels)
qPosterior = maximizeVariationalLowerBound(images,
yOnehot,
qPrior,
taskId=1)
print("Prediction Time :-) ")
with torch.no_grad():
correct = 0
total = 0
for images, labels in testLoader:
images = images.reshape(-1, 28 * 28).to(Device)
labels = labels.to(Device)
monteCarlo = MonteCarlo(qPrior, numSamples)
predicted = monteCarlo.computeMonteCarlo(images, 1)
_, predicted = torch.max(predicted.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
print('Accuracy of the network on the 10000 test images: {} %'.format(
100 * correct / total))