miniblog.

← Back to all posts
May 4, 2019 at 12:11
An excellent overview of the different gradient descent algorithms, and a nice example of content that is available as both a responsive website and a PDF on arXiv:
An overview of gradient descent optimization algorithms
An overview of gradient descent optimization algorithms
Gradient descent is the preferred way to optimize neural networks and many other machine learning algorithms but is often used as a black box. This post explores how many of the most popular gradient-based optimization algorithms such as Momentum, Adagrad, and Adam actually work.

Related Posts

Sep 4, 2021 at 16:50
earlyoom is a neat linux tool that kills processes a little earlier, so your system remains responsive: https://github.com/rfjakob/earlyoom The readme is pretty educational too, and even encourages you to `tail /dev/zero` to see it in action!
GitHub - rfjakob/earlyoom: earlyoom - Early OOM Daemon for Linux
GitHub - rfjakob/earlyoom: earlyoom - Early OOM Daemon for Linux
earlyoom - Early OOM Daemon for Linux. Contribute to rfjakob/earlyoom development by creating an account on GitHub.
May 20, 2021 at 05:55
1
Writing a REPL that evaluates-as-you-type, keeping the UI responsive and staying defensive against runaway memory usage: https://scattered-thoughts.net/writing/making-live-repls-behave/ It's a really hard problem for PLs, but even this small live demo has a lot of complexity.
Making live repls behave
Nov 24, 2018 at 00:31
1
I really like Stylus, the Firefox plugin for writing custom CSS for websites. My rules aren't clever though. The vast majority are just adding display: none; for distracting UI elements. The remainder are for excessively responsive sites that become unreadable on wide monitors.