PyTorch Introduction I

Installation

1. Download Anaconda or Miniconda.
2. After installation, open a terminal.
3. Create a dedicated virtual environment.
4. Install PyTorch according to your system and CUDA version.

# create a new environment
(base) $ conda create -n pytorch python=3.11
(base) $ conda activate pytorch

# install PyTorch (example with CUDA 12.6)
pip install torch==2.7.1 torchvision==0.22.1 torchaudio==2.7.1 --index-url https://download.pytorch.org/whl/cu126

Tensors and variables

A tensor is the main data structure in PyTorch. It can hold numeric values, be moved to CPU or GPU, and support automatic differentiation.

import torch

X = torch.randn(4, 5)
print(X.shape)
print(X.dtype)
print(X.device)

Common tensor creation APIs

torch.eye(4)
torch.zeros((4, 5))
torch.ones((4, 5))
torch.rand((4, 5))
torch.tensor([[1, 2], [3, 4]])

Tensor attributes

• X.dtype
• X.shape
• X.device
• X.numpy()

Summary

PyTorch is a strong deep learning framework with an intuitive API, efficient tensor operations, and automatic differentiation for training neural networks.