In [10]:
!pip install matplotlib ipywidgets torchsummary

Requirement already satisfied: matplotlib in /usr/local/lib/python3.10/dist-packages (3.10.1)
Requirement already satisfied: ipywidgets in /usr/local/lib/python3.10/dist-packages (8.1.7)
Collecting torchsummary
  Downloading torchsummary-1.5.1-py3-none-any.whl (2.8 kB)
Requirement already satisfied: numpy>=1.23 in /usr/local/lib/python3.10/dist-packages (from matplotlib) (2.2.4)
Requirement already satisfied: python-dateutil>=2.7 in /usr/local/lib/python3.10/dist-packages (from matplotlib) (2.9.0.post0)
Requirement already satisfied: packaging>=20.0 in /usr/local/lib/python3.10/dist-packages (from matplotlib) (24.2)
Requirement already satisfied: pillow>=8 in /usr/local/lib/python3.10/dist-packages (from matplotlib) (11.1.0)
Requirement already satisfied: fonttools>=4.22.0 in /usr/local/lib/python3.10/dist-packages (from matplotlib) (4.57.0)
Requirement already satisfied: pyparsing>=2.3.1 in /usr/local/lib/python3.10/dist-packages (from matplotlib) (3.2.3)
Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.10/dist-packages (from matplotlib) (0.12.1)
Requirement already satisfied: contourpy>=1.0.1 in /usr/local/lib/python3.10/dist-packages (from matplotlib) (1.3.2)
Requirement already satisfied: kiwisolver>=1.3.1 in /usr/local/lib/python3.10/dist-packages (from matplotlib) (1.4.8)
Requirement already satisfied: comm>=0.1.3 in /usr/local/lib/python3.10/dist-packages (from ipywidgets) (0.2.2)
Requirement already satisfied: widgetsnbextension~=4.0.14 in /usr/local/lib/python3.10/dist-packages (from ipywidgets) (4.0.14)
Requirement already satisfied: traitlets>=4.3.1 in /usr/local/lib/python3.10/dist-packages (from ipywidgets) (5.14.3)
Requirement already satisfied: jupyterlab_widgets~=3.0.15 in /usr/local/lib/python3.10/dist-packages (from ipywidgets) (3.0.15)
Requirement already satisfied: ipython>=6.1.0 in /usr/local/lib/python3.10/dist-packages (from ipywidgets) (8.34.0)
Requirement already satisfied: exceptiongroup in /usr/local/lib/python3.10/dist-packages (from ipython>=6.1.0->ipywidgets) (1.2.2)
Requirement already satisfied: matplotlib-inline in /usr/local/lib/python3.10/dist-packages (from ipython>=6.1.0->ipywidgets) (0.1.7)
Requirement already satisfied: jedi>=0.16 in /usr/local/lib/python3.10/dist-packages (from ipython>=6.1.0->ipywidgets) (0.19.2)
Requirement already satisfied: prompt_toolkit<3.1.0,>=3.0.41 in /usr/local/lib/python3.10/dist-packages (from ipython>=6.1.0->ipywidgets) (3.0.50)
Requirement already satisfied: pygments>=2.4.0 in /usr/local/lib/python3.10/dist-packages (from ipython>=6.1.0->ipywidgets) (2.19.1)
Requirement already satisfied: pexpect>4.3 in /usr/local/lib/python3.10/dist-packages (from ipython>=6.1.0->ipywidgets) (4.9.0)
Requirement already satisfied: decorator in /usr/local/lib/python3.10/dist-packages (from ipython>=6.1.0->ipywidgets) (5.2.1)
Requirement already satisfied: stack_data in /usr/local/lib/python3.10/dist-packages (from ipython>=6.1.0->ipywidgets) (0.6.3)
Requirement already satisfied: typing_extensions>=4.6 in /usr/local/lib/python3.10/dist-packages (from ipython>=6.1.0->ipywidgets) (4.13.0)
Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.10/dist-packages (from python-dateutil>=2.7->matplotlib) (1.17.0)
Requirement already satisfied: parso<0.9.0,>=0.8.4 in /usr/local/lib/python3.10/dist-packages (from jedi>=0.16->ipython>=6.1.0->ipywidgets) (0.8.4)
Requirement already satisfied: ptyprocess>=0.5 in /usr/local/lib/python3.10/dist-packages (from pexpect>4.3->ipython>=6.1.0->ipywidgets) (0.7.0)
Requirement already satisfied: wcwidth in /usr/local/lib/python3.10/dist-packages (from prompt_toolkit<3.1.0,>=3.0.41->ipython>=6.1.0->ipywidgets) (0.2.13)
Requirement already satisfied: pure-eval in /usr/local/lib/python3.10/dist-packages (from stack_data->ipython>=6.1.0->ipywidgets) (0.2.3)
Requirement already satisfied: executing>=1.2.0 in /usr/local/lib/python3.10/dist-packages (from stack_data->ipython>=6.1.0->ipywidgets) (2.2.0)
Requirement already satisfied: asttokens>=2.1.0 in /usr/local/lib/python3.10/dist-packages (from stack_data->ipython>=6.1.0->ipywidgets) (3.0.0)
Installing collected packages: torchsummary
Successfully installed torchsummary-1.5.1
WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv

Task¶

Our task is simple, recognize handwritten digits. We will use MNIST dataset for this tutorial.

Import necessary library¶

In this tutorial, we are going to use pytorch, the cutting-edge deep learning framework to complete our task.

In [11]:
import torch
import torchvision
In [12]:
## Create dataloader, in PyTorch, we feed the trainer data with use of dataloader
## We create dataloader with dataset from torchvision,
## and we dont have to download it seperately, all automatically done

# Define batch size, batch size is how much data you feed for training in one iteration
batch_size_train = 64 # We use a small batch size here for training
batch_size_test = 1024 #

# define how image transformed
image_transform = torchvision.transforms.Compose([
                               torchvision.transforms.ToTensor(),
                               torchvision.transforms.Normalize(
                                 (0.1307,), (0.3081,))
                             ])
#image datasets
train_dataset = torchvision.datasets.MNIST('dataset/',
                                           train=True,
                                           download=True,
                                           transform=image_transform)
test_dataset = torchvision.datasets.MNIST('dataset/',
                                          train=False,
                                          download=True,
                                          transform=image_transform)
#data loaders
train_loader = torch.utils.data.DataLoader(train_dataset,
                                           batch_size=batch_size_train,
                                           shuffle=True)
test_loader = torch.utils.data.DataLoader(test_dataset,
                                          batch_size=batch_size_test,
                                          shuffle=True)
In [13]:
# import library
import matplotlib.pyplot as plt
# We can check the dataloader
_, (example_datas, labels) = next(enumerate(test_loader))
sample = example_datas[0][0]
# show the data
plt.imshow(sample, cmap='gray', interpolation='none')
print("Label: "+ str(labels[0]))

Label: tensor(3)
No description has been provided for this image
In [14]:
## Now we can start to build our CNN model
## We first import the pytorch nn module and optimizer
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
## Then define the model class
class CNN(nn.Module):
    def __init__(self):
        super(CNN, self).__init__()
        #input channel 1, output channel 10
        self.conv1 = nn.Conv2d(1, 10, kernel_size=5, stride=1)
        #input channel 10, output channel 20
        self.conv2 = nn.Conv2d(10, 20, kernel_size=5, stride=1)
        #dropout layer
        self.conv2_drop = nn.Dropout2d()
        #fully connected layer
        self.fc1 = nn.Linear(320, 50)
        self.fc2 = nn.Linear(50, 10)
    def forward(self, x):
        x = self.conv1(x)
        x = F.max_pool2d(x, 2)
        x = F.relu(x)
        x = self.conv2(x)
        x = self.conv2_drop(x)
        x = F.max_pool2d(x, 2)
        x = F.relu(x)
        x = x.view(-1, 320)
        x = self.fc1(x)
        x = F.relu(x)
        x = F.dropout(x)
        x = self.fc2(x)
        return F.log_softmax(x)
In [15]:
## create model and optimizer
learning_rate = 0.01
momentum = 0.5
device = "cuda"
model = CNN().to(device) #using cpu here
optimizer = optim.SGD(model.parameters(), lr=learning_rate,
                      momentum=momentum)
In [16]:
from tqdm import tqdm_notebook as tqdm
##define train function
def train(model, device, train_loader, optimizer, epoch, log_interval=10000):
    model.train()
    tk0 = tqdm(train_loader, total=int(len(train_loader)))
    counter = 0
    for batch_idx, (data, target) in enumerate(tk0):
        data, target = data.to(device), target.to(device)
        optimizer.zero_grad()
        output = model(data)
        loss = F.nll_loss(output, target)
        loss.backward()
        optimizer.step()
        counter += 1
        tk0.set_postfix(loss=(loss.item()*data.size(0) / (counter * train_loader.batch_size)))
##define test function
def test(model, device, test_loader):
    model.eval()
    test_loss = 0
    correct = 0
    with torch.no_grad():
        for data, target in test_loader:
            data, target = data.to(device), target.to(device)
            output = model(data)
            test_loss += F.nll_loss(output, target, reduction='sum').item() # sum up batch loss
            pred = output.argmax(dim=1, keepdim=True) # get the index of the max log-probability
            correct += pred.eq(target.view_as(pred)).sum().item()
    test_loss /= len(test_loader.dataset)

    print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
        test_loss, correct, len(test_loader.dataset),
        100. * correct / len(test_loader.dataset)))
In [17]:
num_epoch = 3
for epoch in range(1, num_epoch + 1):
        train(model, device, train_loader, optimizer, epoch)
        test(model, device, test_loader)

/tmp/ipykernel_292/1895196525.py:5: TqdmDeprecationWarning: This function will be removed in tqdm==5.0.0
Please use `tqdm.notebook.tqdm` instead of `tqdm.tqdm_notebook`
  tk0 = tqdm(train_loader, total=int(len(train_loader)))

  0%|          | 0/938 [00:00<?, ?it/s]
/tmp/ipykernel_292/4020281764.py:32: UserWarning: Implicit dimension choice for log_softmax has been deprecated. Change the call to include dim=X as an argument.
  return F.log_softmax(x)

Test set: Average loss: 0.3366, Accuracy: 9000/10000 (90%)


  0%|          | 0/938 [00:00<?, ?it/s]
Test set: Average loss: 0.2137, Accuracy: 9352/10000 (94%)


  0%|          | 0/938 [00:00<?, ?it/s]
Test set: Average loss: 0.1719, Accuracy: 9502/10000 (95%)
In [18]:
from torchsummary import summary
summary(model, (1, 28, 28))

----------------------------------------------------------------
        Layer (type)               Output Shape         Param #
================================================================
            Conv2d-1           [-1, 10, 24, 24]             260
            Conv2d-2             [-1, 20, 8, 8]           5,020
         Dropout2d-3             [-1, 20, 8, 8]               0
            Linear-4                   [-1, 50]          16,050
            Linear-5                   [-1, 10]             510
================================================================
Total params: 21,840
Trainable params: 21,840
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.00
Forward/backward pass size (MB): 0.06
Params size (MB): 0.08
Estimated Total Size (MB): 0.15
----------------------------------------------------------------

/tmp/ipykernel_292/4020281764.py:32: UserWarning: Implicit dimension choice for log_softmax has been deprecated. Change the call to include dim=X as an argument.
  return F.log_softmax(x)
In [ ]:
In [ ]: