Skip to content

Getting Started with Pytorch

Day 1

I've decided to start the journey with Jumping in pool directly approach. Might be a bad decision, but I tried it the other way few months back. I signed up for Andrew NG's Machine learning course, but my interested dropped as I was getting lost in the theory part. Thus I decided to take another approach of learning as you go.

I'm starting off with dedicating 1 Hr each day to watch

Reason why I picked this up :

  • πŸ˜‡ Comfort - I'm already proficient in #Python, which can save me time learning another programming language.
  • 🀩Popularity - Despite hearing much about TensorFlow, recent surveys indicate many are migrating to PyTorch (source:
  • πŸ‹πŸ»Trial and Error - Even if my decision is incorrect, I can always pivot and explore other options. πŸ˜›


My setup is pretty clean to move forward, I might change as I go ahead, but for now what you need.

  • Good knowledge of python : If you don't know python, I'll request you to pause reading this right now, and take a python course. (Although If I were you, I'd not listed to this guy who wrote this page and still continue πŸ˜›, and take Jumping in pool approach for python as well. )

Here are few places I'd recommend you to start

  • Hardware setup - You don't need fancy setup at all to get started. You can start writing AI code for free on websites like Google Colab , Deepnote or you can also setup things on your local, which is what I'm doing as well. I'm using a conda to manage virtual env.

Once the environment is created , I'm installing notebook so I can create python notebooks.

conda create -n ai pytorch

conda activate ai

pip install notebook

Creating first notebook

Activate environment and start writing python code.

import torch