Getting Started with PyTorch on Linux

PyTorch is an open-source machine learning library widely used for deep learning applications. It offers a flexible and efficient platform to develop and train neural networks. This tutorial will guide you through the installation and initial setup of PyTorch on a Linux system.

Prerequisites

• A Linux system (e.g., Ubuntu, CentOS).
• Python 3.6 or later installed.
• pip installed for managing Python packages.

1. Installing PyTorch

PyTorch provides a simple way to install the library via pip. Start by updating your package index:

sudo apt update

1.1. Using pip to Install PyTorch

You can find the best installation command for your system and configuration on the PyTorch official website. For example:

pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113

This command installs PyTorch with support for CUDA 11.3. Make sure to adjust the version based on your needs and hardware.

2. Verifying the Installation

Once PyTorch is installed, open a Python shell or a Jupyter Notebook to verify the installation:

import torch
print(torch.__version__)

This should print the installed version of PyTorch without any errors.

3. Using PyTorch for a Simple Example

Let’s create a simple PyTorch tensor to get familiar with how PyTorch works:

import torch

# Create a 2D tensor
my_tensor = torch.tensor([[1, 2, 3], [4, 5, 6]])
print(my_tensor)

You should see the output of the created tensor displayed in the console.

4. Building a Simple Neural Network

Here’s a basic example of how to build a simple neural network using PyTorch:

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

# Define the neural network class
class SimpleNN(nn.Module):
    def __init__(self):
        super(SimpleNN, self).__init__()
        self.fc1 = nn.Linear(3, 3)  # Layer with 3 inputs and 3 outputs

    def forward(self, x):
        return self.fc1(x)

# Initialize the model
model = SimpleNN()
print(model)

5. Training the Model

In a typical workflow, you would prepare your data, define a loss function, and set up an optimizer for training. For example:

criterion = nn.MSELoss()  # Mean Squared Error loss
optimizer = optim.SGD(model.parameters(), lr=0.01)  # Stochastic Gradient Descent

You can then train your model on your dataset using a loop that adjusts the weights based on the loss.

6. Conclusion

By following this tutorial, you have successfully installed PyTorch and run a simple example of how to create tensors and a neural network. PyTorch is a powerful framework for machine learning and deep learning, allowing developers to experiment with various applications. Explore the extensive documentation and resources to expand your knowledge of PyTorch and its capabilities.

About the Author

Dorian Kane

Author

View All Posts