Make Your First GAN Using PyTorch/

...

/

The Fashion MNIST Classifier Performance

The Fashion MNIST Classifier Performance

Let's see how the network performs with the given configuration.

We'll cover the following...

Training the classifier

Training a classifier neural network is now really simple. That’s because we’ve done all the hard work setting up the dataset class and the neural network class.

Press + to interact
%%time 
# create neural network
C = Classifier()
# train network on MNIST data set
epochs = 3
for i in range(epochs):
    print('training epoch', i+1, "of", epochs)
    for label, image_data_tensor, target_tensor in fmnist_dataset:
        C.train(image_data_tensor, target_tensor)
        pass
    pass

Train the network again for 6 min 44 sec with 3 epochs.

Testing the classifier

Now that we have a fairly well-trained network, let’s ask it to classify images. We’ll switch to the FMNIST test dataset of 10,000 images. These are images our neural network has not yet seen.

Loading the test data

Let’s load the dataset with a new Dataset object.

Press + to interact
# load MNIST test data
fmnist_test_dataset = FMnistDataset('fashion_mnist/fashion-mnist_test.csv')

Classifying the test sample

We can pick a record from the test dataset and see what the image ...