Learn about the tradeoffs of reducing losses using `sum()`, and reducing losses using `mean()`.

The Perfect Solution to Loss Reduction in Deep Learning!

Overview of a common pitfall Keras users hit when writing subclass models

A Common Pitfall When Writing Subclass Models in Keras

Overview of the first ever official Keras object detection API.

API Sneak Peek: Train an Object Detection Model on Pascal VOC 2007 using KerasCV

Efficiently manage latex assets using a Git based workflow.

Authoring Tip: Efficiently Managing Generated Latex Assets

In this post Luke discusses how Keras users write their code, and how he designs APIs accordingly.

How Keras Users Write Code and How to Design APIs Accordingly

The goal of this guide is to teach someone who knows the basic of machine learning how to train a nearly state of the art image classifier from scratch.

The Definitive Guide to Classification

The goal of this guide is to teach someone who knows the basic of machine learning how to train an extremely powerful object detection model from scratch. This guide explains a wide variety of techniques and gives some theoretical backing as to why they are effective.

The Definitive Guide to Object Detection

Typing static props in NextJS apps is a bit annoying.  This post gives you a solution to this common pain point.

Automatically type NextJS static component props

tz is a new command line tool to switch tmux sessions using a fzf menu.

tz: switch tmux sessions with fzf

I love typescript, but lukewood.xyz is almost 10 years old now.  The codebase pre-dates typescript.  In this blog I talk about how I'm incrementally migrating the codebase.

Incrementally migrating a NextJS app to typescript

How I'm using llms in my neovim workflows.

A neovim plugin using an LLM, with the goal of making it easier to write neovim plugins using LLMs.

Neovim is great for automating tedious tasks.  I used it to make my blogging workflow fast.

Streamlining blogging using neovim

Very short post about a shell script I use to manage tmux

A nifty tmux management script

Approximate non one to one mapping using mixture density networks.

Approximating non-Function Mappings with Mixture Density Networks

Construct a powerful object detection pipeline to detect marine animals

Marine Animal Object Detection with KerasCV

Discussion about why the Elixir programming language sticks out.

TensorFlow Loops: To Unroll, or To Not Unroll

Deep dive into the internals of the Keras model class by writing a custom training loop duplicating the behavior of `keras.Model`.

Keras Model Walkthrough

Use Parametric Spectral Filters to construct an autoencoder.

On the Flexibility of Parametric Spectral Filters

Atoms offer some distinct advantages over strings.  Learn more about them in this blog post.

Argument For Using Atoms

TensorFlow has limited complex number support.  Write the operations yourself to maintain automatic differentiation.

Complex Number Deep Learning with Tensorflow

Why I Love Elixir

The Typescript TypeChecker caught a nasty bug causing massive performance issues.

Typescript Saved My Game's Performance

Use event driven programming to minimize the coupling of components in a distributed game engine.

All Posts

The Perfect Solution to Loss Reduction in Deep Learning!

A Common Pitfall When Writing Subclass Models in Keras

API Sneak Peek: Train an Object Detection Model on Pascal VOC 2007 using KerasCV

Authoring Tip: Efficiently Managing Generated Latex Assets

How Keras Users Write Code and How to Design APIs Accordingly