Geek Out Time: Playing with a Machine Learning Pipeline with PyTorch and MLflow

I used Airflow many years ago in my first machine-learning project when I was at GFG. This week, I decided to experiment with a few other tools, starting with TFX. Unfortunately, I did not manage to get it installed on my M1 MacBook due the compatibility issues after multiple rounds of troubleshooting. It might be due to the limited support for the M1 chip on TFX’s side. :-( If you know why, I’d be happy to hear. Then, I tried to create a simple machine-learning pipeline using PyTorch for model training and MLflow for experiment tracking. I used a small dataset to keep things manageable and demonstrate the key concepts.

Prerequisites

I have the following installed:

Step 1: Setting Up a Virtual Environment

This is an excerpt — the full article continues on Medium.