Every blog post that took off on Hacker News, ranked by the votes of its best submission. 3,091 posts.
I shared this internally at Meta in response to a deluge of clout-chasing posts celebrating PMs landing prod diffs. The response was positive and the message generalizes, so here we are on the open www. PMs shouldn’t waste time landing … Continue reading →
AI is impressive because we’ve failed at semantic web and personal computing | exotext
In the glorious future, every website will be chock-full of semantic metadata. Restaurants won't have a 50MB PDF explaining the chef's vision for organic cuisine - instead, they'll have little scraps of data on the HTML page like: Copied JSON to 📋⧉ JSON"hasMenuItem":{ "@type":"MenuItem", "name":"Dodo In A Bun"…
Lately, I've been playing around with LLMs to write code.
Language models assign probability values to sequences of words. Those three words that appear right above your keyboard on your phone that try to predict the next word you’ll type are one of the uses of language modeling. In the case shown below, the language model is predicting that “from”, “on” and “it” have a high…
Previous major advancements in Computer Science and its applications always felt empowering to me, AI feels different.
A collection of posts by Joe Magerramov
TL;DR: OpenAI has a new code generating model that’s improved in a number of ways and can handle nearly two times as much text (4,000 tokens.) I built several small games and applications without t…
The Case for Bayesian Deep Learning Andrew Gordon Wilson Abstract Bayesian inference is especially compelling for deep neural networks. The key distinguishing property of a Bayesian approach is margin
I’ve been building out a small suite of command-line tools for working with ChatGPT, GPT-4 and potentially other language models in the future. The three tools I’ve built so far …
A first attempt at identifying and cataloging DeepSeek's glitched tokens
After maintaining a total stance of public silence for months, I recently publicly announced my unexpected divorce on Instagram. I shared a picture of the divorce cake that my friends got for me, and shared a brief essay I had drafted the day before about the news. I had to edit it down slightly from my original draft…
Today I was setting up Hermes to see how it does with web research. I chose DeepSeek V4 because I know it is cheap, but seeing it’s pricing next to Anthropic and OpenAI ‘frontier’ models is crazy. Nearly a 50x price increase based on tokens alone, let alone how much pondering any of their models might fall into (using…
Last year I created a fun little experiment where I asked a bunch of LLMs to rank 97 hackernews users using their comment history based on whether they would be good candidates for the role of “software engineer at google”. (yes yes, seems silly I know, you can read part 1 and part 2 but […]
[Twitter thread, Hacker News discussion, Le Monde]
Of all the forms of tech debt, decoherence is the worst. It compounds exponentially with a system's rate of change, and AI is now producing it faster than ever, leading us deeper and deeper into the abyss...
I trusted a lot today. I trusted my phone to wake me on time. I trusted Uber to arrange a taxi for me, and the driver to get me to the airport safely. I trusted thousands of other drivers on the road not to ram my car on the way. At the airport, I trusted ticket agents and maintenance engineers and everyone else who…
We need to augment our command line tools and design APIs so they can be better used by LLM Agents. The designs are inadequate for LLMs as they are now – especially if you're constrained by the tiny context windows available with local models.
This post covers my experience training image classification models with Intel’s PyTorch extension on the Arc A770 GPU.
With the problem of Image Classification is more or less solved by Deep learning, Text Classification is the next new developing theme in deep learning. For those who don’t know, Text classification is a common task in natural language processing, which transforms a sequence of text of indefinite length into a…
AI plays well at complex strategy games with uncertainty and incomplete information.
Lessons learned from building Codescribble with a little help (and a lot of frustration) from an AI
How Google Home provides an interesting insight into the curiosity of a 4-year-old.
Detailed report on the Thoughtworks AI retreat, should we write a manifesto for AI-enabled development, is spec-driven development the return of waterfall, LLMs perform better in healthy codebases
I had an interesting discussion at work today (among people I think would all call themselves statisticians!) about the distinction(s) between statistics and machine learning. It is something I am …
HN Discussion
One day in the mid-1920s, a Moscow newspaper reporter named Solomon Shereshevsky entered the laboratory of the psychologist Alexander Luria. Shereshevsky's boss at the newspaper had noticed that Shereshevsky never needed to take any notes, but somehow still remembered all he was told, and had suggested he get his…
