PyTorch & Python Basics Cheat Sheet

1. Installation

pip install torch torchvision torchaudio  

Verify Installation:

import torch
print(torch.__version__)  

---

2. Import PyTorch

import torch  
import torch.nn as nn  
import torch.optim as optim  
import torchvision  

---

3. Tensors

Create Tensors:

x = torch.tensor([5.0, 10.0, 15.0])  
y = torch.zeros(3)               # Zero tensor  
z = torch.ones(3)                # Ones tensor  
rand = torch.rand(3, 3)          # Random tensor  

Tensor Shape & Type:

print(x.shape)  
print(x.dtype)  

Reshape Tensors:

x = torch.rand(4, 4)  
x_reshaped = x.view(2, 8)  

---

4. Tensor Operations

a = torch.tensor([1, 2, 3])  
b = torch.tensor([4, 5, 6])  

# Basic Operations  
c = a + b  
d = a * b  
e = torch.dot(a, b)               # Dot product  
f = torch.sum(a)                  # Sum of elements  

Matrix Multiplication:

x = torch.rand(2, 3)  
y = torch.rand(3, 2)  
z = torch.mm(x, y)  

Element-wise Operations:

x = torch.rand(3)  
y = torch.exp(x)                  # Exponential  
z = torch.sqrt(x)                 # Square root  

---

5. Autograd (Automatic Differentiation)

x = torch.tensor(2.0, requires_grad=True)  
y = x**3 + 5*x  
y.backward()                     # Calculate gradients  
print(x.grad)                    # dy/dx = 3x² + 5  

---

6. Neural Network Basics

Simple Neural Network:

class Net(nn.Module):  
    def __init__(self):  
        super(Net, self).__init__()  
        self.fc1 = nn.Linear(4, 3)  
        self.fc2 = nn.Linear(3, 1)  

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

net = Net()  
print(net)  

---

7. Loss and Optimizer

criterion = nn.MSELoss()                # Mean Squared Error  
optimizer = optim.SGD(net.parameters(), lr=0.01)  

---

8. Training a Neural Network

for epoch in range(100):  
    optimizer.zero_grad()               # Zero gradients  
    outputs = net(x)                    # Forward pass  
    loss = criterion(outputs, y)        # Compute loss  
    loss.backward()                     # Backward pass  
    optimizer.step()                    # Update weights  

---

9. Loading Data (Datasets & Dataloaders)

from torchvision import datasets, transforms  

transform = transforms.Compose([  
    transforms.ToTensor(),  
    transforms.Normalize((0.5,), (0.5,))  
])  

train_data = datasets.MNIST(root='./data', train=True, download=True, transform=transform)  
train_loader = torch.utils.data.DataLoader(dataset=train_data, batch_size=64, shuffle=True)  

---

10. GPU Acceleration

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')  
model = Net().to(device)  

---

11. Saving and Loading Models

# Save Model  
torch.save(net.state_dict(), 'model.pth')  

# Load Model  
model = Net()  
model.load_state_dict(torch.load('model.pth'))  
model.eval()  

---

12. Testing the Model

with torch.no_grad():  
    for data in train_loader:  
        images, labels = data  
        outputs = model(images)  
        _, predicted = torch.max(outputs, 1)  

---

13. Common PyTorch Functions

---

14. Plotting with PyTorch

import matplotlib.pyplot as plt  

x = torch.linspace(-10, 10, 100)  
y = torch.sin(x)  

plt.plot(x.numpy(), y.numpy())  
plt.title('Sine Wave')  
plt.show()  

---

15. Example: Linear Regression

# Data
x = torch.rand(100, 1)  
y = 3 * x + 2 + torch.randn(100, 1) * 0.1  

# Model
model = nn.Linear(1, 1)  
criterion = nn.MSELoss()  
optimizer = optim.SGD(model.parameters(), lr=0.01)  

# Training Loop
for epoch in range(500):  
    y_pred = model(x)  
    loss = criterion(y_pred, y)  
    optimizer.zero_grad()  
    loss.backward()  
    optimizer.step()  

# Visualization
plt.scatter(x.numpy(), y.numpy())  
plt.plot(x.numpy(), y_pred.detach().numpy(), color='red')  
plt.title('Linear Regression with PyTorch')  
plt.show()