Shouting into the void , now with AI.
Over the weekend I built an AI agent. I thought this was an engineering problem, but I now think it is a user experience problem. In regular software, we solve engineering problems with UX. If a page loads slowly, we show a spinner so the user feels like something is happening. In agentic software, we create UX problems with engineering decisions. If we take some data out of the chat history, but…
built a system to teach my kid to read, using a free program called Anki for the “planning” and AI to make content that would lure him in. t worked, he and I enjoyed it immensely, he reads fluently (though mechanically). Along the way I observed and learned a great deal about learning and teaching, young kids, or my young kid. That’s what I want to share with you today. In…
Triton Inference Server has become a popular choice for production model serving, and for good reason: it is fast, flexible, and powerful. That said, using Triton effectively requires understanding where it shines—and where it very much does not. This post collects five practical lessons from running Triton in production that I wish I had internalized earlier. Choose the Right Serving Layer Not…
Before diving into the reasons behind my decision, it’s essential to know that I am a Jewish Israeli atheist living in Berlin. This background might make you wonder why I would even consider building such an app. Despite my core identity, I sold a developer tools company two years ago and vowed, “Never again to build a developer tools company.” Instead, I want to pursue something…
tl;dr RNNS work great for text but convolutions can do it faster Any part of a sentence can influence the semantics of a word. For that reason we want our network to see the entire input at once Getting that big a receptive can make gradients vanish and our networks fail We can solve the vanishing gradient problem with DenseNets or Dilated Convolutions Sometimes we need to generate text. We can…
Update 15.03.2024 I wrote this more than seven years ago. My understanding has evolved since then, and the world of deep learning has gone through more than one revolution since. It was popular back in the day and might still be a fun read though you might learn more accurate and upto date information somewhere else Update 25.1.17 — Took me a while but here is an ipython notebook with a rough…