PinnedPublished inBetter ProgrammingAttach a Visual Debugger to ML-training Jobs on KubernetesAs machine learning models grow in size and complexity, cloud resources are more and more often required for training. However, debugging…Apr 16, 2023Apr 16, 2023
Published inTDS ArchiveDeViSE Zero-shot learningDeViSE combines computer vision with language modeling and can make predictions about thousands of image labels not observed during…Mar 23, 2019A response icon3Mar 23, 2019A response icon3
Published inTDS ArchiveHow to visualize convolutional features in 40 lines of codeDeveloping techniques to interpret convnets is an important field of research. This article explains how you can visualize their features.Jan 20, 2019A response icon22Jan 20, 2019A response icon22
Published inTDS ArchiveRetinaNet: how Focal Loss fixes Single-Shot DetectionObject detection is a tremendously important field in computer vision needed for autonomous driving, video surveillance, medical…Nov 25, 2018A response icon14Nov 25, 2018A response icon14
A journey to the icy world of Saturn’s rings and then 450 lightyears furtherIn this article I will explain to you how astrophysicists use tiny icy moons in Saturn’s rings to learn about young growing planets in…Sep 22, 2018Sep 22, 2018
The stunning beauty of Chaos theoryHave you ever wondered what chaos theory is? If you give me a few minutes of your time, I will explain you the basics of one of my favorite…Sep 16, 2018A response icon7Sep 16, 2018A response icon7
Published inTDS ArchiveHow to match DeepMind’s Deep Q-Learning score in BreakoutIf you are as fascinated by Deep Q-Learning as I am but never had the time to understand or implement it, this is for you: In one Jupyter…Aug 26, 2018A response icon18Aug 26, 2018A response icon18
Published inTDS ArchiveJupyter and tensorboard in tmux and sshIn recent posts, I described how you can set up your personal deep learning workstation and how you can switch it on and access it…Aug 2, 2018A response icon1Aug 2, 2018A response icon1
Published inTDS ArchiveWhy AdamW mattersAdaptive optimizers like Adam have become a default choice for training neural networks. However, when aiming for state-of-the-art results…Jun 3, 2018A response icon7Jun 3, 2018A response icon7
Accessing your Deep Learning station remotely and setting up wake on lanI recently described how I set up my personal Deep Learning Station station.May 20, 2018May 20, 2018