I trained a small transformer in 1.5hrs and it beats many LLMs
I trained a small transformer from scratch in 1.5hrs on a 5090 Beats many LLMs, and scores the same as TRM/HRM
The latest blog posts to take off on Hacker News. 3,698 posts.
I trained a small transformer from scratch in 1.5hrs on a 5090 Beats many LLMs, and scores the same as TRM/HRM
In 2004 I played too much RuneScape on a 56k modem that died the moment Mum picked up the phone. A 3D world, up to a couple of thousand players on a server, dozens on screen at once - in the browser, on 5 kilobytes per second. It worked. Let’s follow a single step and see how.
A new paper in Psychological Science (.htm) reports a failure to replicate Study 2 of Ariely and Wertenbroch’s influential article entitled, “Procrastination, Deadlines, and Performance: Self-Contr…
It is difficult to estimate how many web pages created in 1993-1994 made it into the new millenium in their premordial way. If you manage to find something that was put online that time, it would in the best case display a 1995-1996 skin. But there is a way to find pages that live for ever in 1993.
Read this Tcl/Tk hands-up/primer to find out how the open-source Command Tool Language toolkit can help you develop cross-plaform tools for Windows, Linux, macOS and even Android.
Learn how to get a free .arpa domain by abusing IPv6 reverse DNS delegations on Hurricane Electric, configuring deSEC, and hosting a live website.
Google today reached the final milestone in a browser-extension transition that has been years in the making, all remaining Manifest V2 extensions were removed from the Chrome Web Store.
Today, tech folk are scrambling to change their workflows to meet newly inflated 5X productivity quotas, while getting pummeled under the co...
I turned three security cameras into an automatic bird identification system using BirdNet-Go. Now my wife and I can track every bird species that visits our yard in real-time.
An Introduction to File Formats and ImHex by Reverse Engineering FEZ's Save File Format
The frozen pizza aisle has become the national security risk du jour
Memoryfields - a vastly simpler way to do agent memory
A few years years ago the curl project signed up and became a CNA. This means that we are masters of and can allocate our own CVE identifiers. For any security problems within our territory, it is we who decides if the issue should get a CVE or not. No more bogus CVEs. 57 CVEs … Continue reading a CVE dispute →
In 2019 I bet on the no-code revolution. That bet aged so-so. Here is the new one.
In this post, we explore how a simple website summary request hijacks Claude Code Opus 5 in Auto Mode and achieves code execution with 60-80% attack success …
How Wirewiki's domain-name autocomplete renders suggestions next frame for 99% of keystrokes.
It's possible to just connect two computers together with Ethernet and do a bit of IP configuration:
How NAT broke the concept of an internet connection
OpenAI announced ChatGPT Work on July 9th, and have been furiously iterating on it ever since. It is an extraordinarily confusing and very powerful product. Here’s what I’ve figured out …
Graphics for Professionals
A blog about software craft and quality
Fully discrete methods dominated for a few years, but language models based on continuous diffusion are making a comeback.
I once overrode my own website blocker for 444 minutes. After that I stopped relying on willpower and started making the internet boring: app deletions, LeechBlock, grayscale mode, and a custom userscript.
Spacelab was a reusable laboratory that could be carried in the Space Shuttle's cargo bay, providing lab space for astronauts and experiment...
An emoji flipbook presentation by Alex Komoroske about how dysfunctional organizational dyanmics arise even when individuals are well-behaved
A security issue in Omarchy’s default Docker configuration meant that essentially every program running in the user’s desktop session could escalate to root without a password, sudo, or a privilege prompt. If you use Omarchy, the most important takeaway is simple: update to 4.0.1. I reported this issue privately…
An anti-pattern is a commonly used process, structure, or pattern of action that, despite initially appearing to be an appropriate and effective response to a problem, has more bad consequences than good ones.Simeon Simeonov first wrote an introduction to the value of startup anti-patterns back in 2013. To sum it up…
Yesterday I covered the OpenAI technical report on the HuggingFace hack. That report had one key new piece of information, and some good prosaic steps OpenAI will be taking to strengthen its alignm…
How Storyteller’s forced alignment algorithm works.
When something you want is not available for purchase, well, you can always build it yourself! I recently moved and have a dedicated office room now. I wanted …