TLDR: I'm going through a simple POC (Proof of Concept) how companies can setup their own ChatGPT-like professional agents without sending their confidential data to OpenAI or to any other online service. After a general introduction to how LLMs are trained, I show how you can use Dolly (a publicly available and …

Read More

I was about to do some basic natural language processing (NLP) task on a large text file: create a dictionary of unique words and count how many times each word occurs. Easy peasy. This should only take me a minute. So I thought. While coding did not take more than 2 zips of coffee, I needed to wait almost an hour for …

Read More

Most of our modern day computers have more than one processing unit. They are often referred to as cores. For a developer, effectively using multiple cores simultaneously to speed up program execution by parallel processing comes with challenges. One of them is changing the value of a variable in shared memory. In this …

Read More

It’s been more than 5 years that I setup this blog on GitHub Pages. Technology has moved on and there are a few new interesting options available. I ended up choosing Hugo because it works well with GitHub Pages, renders static websites at a great performance, offers lots of templates to choose from, and was relatively …

Read More

In this blog post I show how to use logistic regression to classify images. Logistic regression is a statistical method for binary classification, i.e. for analyzing the dependency of a binary outcome on one or more independent variables. In a previous blog post I described linear regression. If you’re not familiar …

Read More

Regression represents one of the cornerstones of machine learning. Comprehending its logic and math provides a solid foundation for learning more advanced machine learning techniques such as neural networks. Linear regression is a statistical method for modeling the relationship between variables. It’s used for …

Read More

I finally found some time to enhance my neural network to support deep learning. The network now masters a variable number of layers and is capable of running convolutional layers. The architecture is generic, light weight (very small memory footprint) and super fast. :-) In a previous blog post I wrote about a …

Read More

I've extended my simple 1-Layer neural network to include a hidden layer and use the back propagation algorithm for updating connection weights. The size of the network (number of neurons per layer) is dynamic. It's accuracy in classifying the handwritten digits in the MNIST database improved from 85% to >91%. …

Read More

In this post I'll explore how to use a very simple 1-layer neural network to recognize the handwritten digits in the MNIST database. In my previous blog post I gave a brief introduction how neural networks basically work. In this post I want to apply this know-how and write some code to recognize handwritten digits …

Read More

When people talk about artificial intelligence and machine learning, they most often refer to (artificial) neural networks (ANN or NN). Let's explore some machine learning basics, without excessive math, purely from a programmer's perspective. A neural network is a computation model imaging the brain where …

Read More