Getting Started with PyTorch on Fedora 42

Sumantro Mukherjee Version F42 onwards Last review: 2025-05-16

PyTorch is an open-source machine learning framework developed by Meta AI. It provides dynamic computational graphs and automatic differentiation, making it highly flexible and popular in both academia and industry. At its core, the torch module in PyTorch offers powerful tensor computation, similar to NumPy, but with GPU acceleration.

With Fedora 42, PyTorch is now available directly through the system package manager, making installation and updates seamless for users.

PyTorch is officially available only on Fedora 42 and above via the system repositories. Installing on earlier Fedora versions using dnf is not supported.

Installation

You can install PyTorch using the Fedora package manager:

sudo dnf install python3-torch

This installs the core torch package and dependencies required to start developing with PyTorch.

Basic Usage

After installation, you can verify the install and start working with tensors immediately.

Import and Version Check

import torch

print("PyTorch version:", torch.__version__)
print("CUDA available:", torch.cuda.is_available())

Tensor Creation

import torch

# Create a 1D tensor
a = torch.tensor([1.0, 2.0, 3.0])
print("Tensor a:", a)

# Random tensor
b = torch.rand(3)
print("Tensor b:", b)

# Add two tensors
c = a + b
print("Sum a + b:", c)

Learn More

For tutorials, model zoo, and full documentation, visit:

Feedback

If you encounter issues or have suggestions for the Fedora package, file a bug at https://bugzilla.redhat.com or join the Fedora AI SIG.

Want to help? Learn how to contribute to Fedora Docs ›