Deep Learning Basics with Pytorch - Part 1

This tutorial series is a hands on beginner guide for pytorch.

Let’s get started. This tutorial series will be comfortable to follow on Unix environment. If you are on Windows make sure to use WSL.

Environment Setup

We are going to need our environment. There are many ways to create environment. But we are going to use Conda.

Install miniconda

Go to the anaconda site and install miniconda.

After you setup and activate miniconda. We are going to install all our libraries in this environment. So that our system won’t be dirty.

Activate environment

Conda is a package manager. Our machine learning tools have first class support in conda. Let’s create a basic environment using conda

conda create -n basic-env python=3.13 -y
conda activate basic-env

Check where our conda env located.

conda env list

Now install Jupyter Notebook inside conda.

conda install -c conda-forge notebook -y

conda-forge is community driven channel. notebook just contain basic jupyterlab. install jupyterlab for full package.

If you are on NVIDIA and have CUDA installed.

conda install pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia

run it.

jupyter notebook

Open Browser and browse:
http://localhost:8888/tree

select python3 default kernel. Give any name

Install pytorch and other libraries.

!pip install torch numpy

Next import torch

import torch

What is PyTorch and Tensor?

PyTorch is a library for Processing Tensors. A Tensor is a number or a vector or a matrix or a n-dimensional array.

Tensors

Let’s create a basic tensor.

#Number
t1 = torch.tensor(10)
t1

Let’s create different types of tensors

# Vector
t2 = torch.tensor([10, 20, 30, 40])
t2

another one

# Matrix
t3 = torch.tensor([[50, 60], 
                   [70, 80], 
                   [90, 10]])
t3

All of these are 1 dimensional array

Let’s see 3 dimensional tensor

# 3-dimensional array
t4 = torch.tensor([
    [[10, 120, 130], 
     [130, 140, 150]], 
    [[150, 160, 170], 
     [170, 180, 190.]]])
t4

Now we can actually see the shape of the tensor

print(t3)
t3.shape

The output will be torch.Size([3, 2]) Try this on your own

We can also do arithmatic operation with these tensors.

y = t1 + t2
y

Numpy

Numpy is library for mathmatics and scientific computing in python.

Let’s see how Numpy array looks like

import numpy as np

x = np.array([[10, 20], [30, 40]])
x

So, In this tutorial we have just touch the water. We will be continuing this tutorial.