Deep Learning with PyTorch

This repo is a resource for my Deep Learning with PyTorch talk. It contains all of the code that was demonstrated as well as the deck.

Inspiration

This talk is inspired by a PyTorch tutorial available online. It is also inspired by a silly nine squares image problem.

The purpose of this silly problem is to provide the basic intution behind models, cost, accuracy, and optimization principles behind machine learning.

Overview

Deep Learning Principles

• input
• model function
• cost function
• optimization method

PyTorch

• Tensors
• DataSets
• DataLoaders
• Models
• Loss (or Cost)
• Optimization

Tools

• ONNX
• Azure Machine Learning service