Tips on using AWS EC2 instances, Docker containers and Pytorch’s distributed communication package — Introduction Efficient distributed training of large machine learning models is a hot-topic right now, especially with billion parameter large language models. Given that Pytorch is widely used for deep learning it makes sense to also employ its distributed package to efficiently train these models. …

Part I: Introduction and Training Loop — In this series we are going to be learning about goal-oriented chatbots and training one with deep reinforcement learning in python! All from scratch! The code for this tutorial series can be found here. Contents of Series Part I: Introduction and Training Loop Part II: DQN Agent Part III: Dialogue State Tracker Part…

What does eigenstuff mean for data science? In this article, we will explore conceptual and mathematical explanations of eigenvalues/eigenvectors. We will develop intuition from understanding linear transformations and various data science algorithms that use eigenstuff (things related to eigenvectors) at their core. …

Proximal Policy Optimization (PPO) is an important reinforcement learning algorithm that has many distributed and asynchronous flavors — Introduction What is PPO? This article assumes you have an understanding of Reinforcement Learning (RL) and specifically PPO, but as a quick recap: PPO is a policy gradient method that attempts to keep successive model updates close together to avoid training instability. …

In this article, we will explore a variety of matrix factorization models, and how to optimize them with gradient descent. — This repo/jupyter notebook contains all the runnable code that goes along with this article. Before we start off, this article assumes you have at least a very basic understanding of collaborative filtering, matrix factorization, gradient descent and neural networks. To refresh your memory on these subjects, let’s relate them:

How to effectively deploy a model with AWS — Introduction Deploying a model with AWS SageMaker is a great way to allow users or customers to interact with it. While you can use the many algorithms and models that come with SageMaker’s Python SDK, there are also many cases where you might want to create your own algorithm. …

Does social distancing and wearing a mask actually help prevent infections? — Take a look at the code that goes along with this article. Try running it yourself and messing with parameters. COVID-19 Modeling the spread of a novel virus and its resulting disease is an important step in learning proper preventive measures. So far there have been many new simulations that have…

Steph Curry is paid too much and James Harden is paid just right — The NBA has tons of stats for stat-heads and machine learning lovers to dive into. There are a bunch of websites that have a wide range of NBA data such as nba.com and basketball-reference.com. Exploring this data is also a great way to brush up on your Tableau and SQL…

Part V: Running the Agent and Future Research — In the previous parts we covered all of the parts and code of a goal-oriented agent! This is the fifth and final part in which we will cover how to run the code and future research and improvements! This tutorial series is based off of TC-Bot. The code for this…