The release of the OpenAI GPT-2 sample language model from the paper Language Models are Unsupervised Multitask Learners (also see Better Language Models and Their Implications) shows great promise of what is to come. The paper describes how training data was collected by following outbound links from Reddit. This got me thinking about what types of content it has seen. I have experimented with triggering recipe generation from the model by using “recipe” and similar conditioning texts.Continue reading
The Stupid Hackathon is a one-day event where participants conceptualize and create projects that have no value whatsoever. Me and Memeleader made Chat Roulette Kastanjett – a service for all castanet players who want to talk to other castanet players over castanet morse code. We made a small chat website (with the help of socket.io and Heroku), wired up castanets with a micro Arduino controller running morse code debugging code. And we won a prize!
Torch is a scientific computing framework with wide support for machine learning algorithms. Andrej Karpathy has an excellent blog post explaining recurrent neural networks (RNNs) and character level models. With his sample code for Torch it is very easy to get started creating your own RNN by using text from a specific domain. This model learns to predict the next character in a sequence. The model can be used to generate text character by character that looks similar to the original training data. Continue reading
A small hack to simplify the application for temporary parental benefit for care of children from the Swedish Social Insurance Agency (Försäkringskassan). Typically this involves multiple steps in a web browser. Now, moving a doll to a square and the computer will send relevant information to kindergarten and the social insurance agency.
A small crab attacked my GoPro camera.
This post is a summary of some ideas for a lightweight semantic interoperability framework It is mainly a composition of existing open standards to form a framework for organisations to be able to ensure that semantic and technical descriptions stay connected over time. The idea is to provide a framework that allows for an increasing semantic interoperability emerging over time without having a large centralized organisation defining vocabularies. Continue reading
Jonathan Gray of the Open Knowledge Foundation participated in a conference of the Communia project, a European thematic network on the digital public domain. In a great post about the meeting at the OKFN blog, he recommends two improvements to the current PSI work; 1: Broaden the scope of the PSI Directive to include publicly funded cultural heritage organisations and 2: Broaden the evidence base for opening up PSI. Continue reading
A while ago I was offered a position with the Swedish eGovernment Delegation.
I will be working with IT standardisation issues and the next revision of guidelines for public websites.
For those of you that have been following Ola Bini’s work on Ioke, the dynamic language for the JVM, I am happy to report that the current release 0.1.1 is usable enough to solve Project Euler problems with. I wanted to learn more about Ioke and the best way to learn a new language is to use it on your own. So, here is some example Ioke code for some of the simpler Project Euler problems. Continue reading
I have been trying to reduce distractions in my computing environment lately. Apparently, our brains aren’t wired properly for task switching: Continue reading
My son has the dubious pleasure of being the primary beta tester of T-shirt messages. This time it is the first draft of the model that will be called “Ballmer”. Turned out pretty OK…
Spring is in the air and it is time to start using the balcony. Currently it is in a state of decay and mainly used to store old furniture. I am out of ideas what to make of it. Can you help? Measurements below.
This pretty much speaks for itself. If I am not mistaken our own packet pro Patrik Fältström is visible in the audience at the end of the clip.