01_PyTorch_simple_fullynet.py

"""
Working code of a simple Fully Connected (FC) network training on MNIST dataset.
The code is intended to show how to create a FC network as well
as how to initialize loss, optimizer, etc. in a simple way to get
training to work with function that checks accuracy as well.
"""

# Imports
import torch
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F  # relu, tanh etc
from torch.utils.data import DataLoader
import torchvision.datasets as datasets
import torchvision.transforms as transforms


# Create Fully Connected Network


class NN(nn.Module):
    def __init__(self, input_size, num_classes):  # MNIST 28X28
        super(NN, self).__init__()
        self.fc1 = nn.Linear(input_size, 50)
        self.fc2 = nn.Linear(50, num_classes)

    def forward(self, x):
        x = F.relu(self.fc1(x))
        x = self.fc2(x)
        return x


# Set device
device = torch.device('cuda' if torch.cuda.is_available() else "cpu")

# Hyperparameters
input_size = 784
num_classes = 10
learning_rate = 0.001
batch_size = 64
num_epohs = 1

# Load data
train_dataset = datasets.MNIST(root='dataset/', train=True, transform=transforms.ToTensor(), download=True)
train_loader = DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True)
test_dataset = datasets.MNIST(root='dataset/', train=False, transform=transforms.ToTensor(), download=True)
test_loader = DataLoader(dataset=test_dataset, batch_size=batch_size, shuffle=True)

# Initialize network
model = NN(input_size=input_size, num_classes=num_classes).to(device)

# Loss and optimizer
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=learning_rate)

# Train network
for epoch in range(num_epohs):
    for batch_idx, (data, targets) in enumerate(train_loader):
        data = data.to(device=device)
        targets = targets.to(device=device)

        # Get to correct shape
        data = data.reshape(data.shape[0], -1)
        
        # forward
        scores = model(data)
        loss = criterion(scores, targets)

        # backward
        optimizer.zero_grad()
        loss.backward()

        # gradient descent or adam step
        optimizer.step()


# Check accuracy on training & test to see how good is our model
def check_accuracy(loader, model):
    num_correct = 0
    num_samples = 0
    model.eval()

    with torch.no_grad():
        for x, y in loader:
            x = x.to(device=device)
            y = y.to(device=device)
            x = x.reshape(x.shape[0], -1)

            scores = model(x)
            _, predictions = scores.max(1)
            num_correct += (predictions == y).sum()
            num_samples += predictions.size(0)

        print(f'Got {num_correct} / {num_samples} with accuracy {float(num_correct) / float(num_samples) * 100:.2f}')

    model.train()


check_accuracy(train_loader, model)
check_accuracy(test_loader, model)