2611
Understanding large language models: A cross-section of the relevant literature
A Cross-Section of the Most Relevant Literature To Get Up to Speed
New
The latest blog posts to take off on Hacker News. 3,824 posts.
2612
The Military-Industrial Stock Buyback Complex
Why is the US military ceding ground to China? As a new DOD report shows, big defense contractors are middlemen whose main purpose is stock buybacks and dividends.
2613
Load Balancing
A bottom-up, animated guide to HTTP load balancing algorithms.
2614
Nibbles of Rust – Restructuring Patterns
In Rust, there is this feature known as “pattern matching”, whereby you can take apart a piece of structured data by writing out patterns which bind to parts of it. This process is known as “destructuring”, because you’re breaking a structure into parts.
2615
OpenSSH versus SSH
You're probably reading this page because you've attempted to access some part of my blog (Wandering Thoughts) or CSpace , the wiki thing it's part of. Unfortunately you're using a browser version that my anti-crawler precautions consider suspicious, most often because it's too old (most often this applies to versions…
2616
Tabi: The shoe that became a sock and then a shoe again
Stockholm, 1912 At 35° C it was an unusually hot day for Stockholm. Shizo Kanakuri stood at the start line of the marathon in the 1912 Stockholm Olympics. Shizo was carrying the hopes, dreams, and pride of Japan, a country attempting to become a “modern” nation. There was something that
2617
Efficiency trades off against resiliency
What’s the “right” level of CPU utilization for a server? If you look at a monitoring dashboard from a well-designed and well-run service, what CPU utilization should we hope to see, averaged over a day or two? It’s a very general question, and it’s not clear it should have a single answer. That said, for a long time…
2618
Digital clutter: Learning to let go and stop hoarding terabytes
Learning to let go and stop hoarding terabytes: Sharing my journey of downsizing my digital storage, letting go of terabytes of data, and moving away from a NAS setup. I decided to use an external Thunderbolt SSD enclosure and an M.2 NVMe SSD to store my data locally. For cloud backup and encryption, I used the Arq…
2619
2620
33TB of text data for a 1T-parameter model
Important: This page summarizes data scaling only, using tokens to parameters as a ratio, and as derived from large language models like GPT-3, Chinchilla, and beyond, linked to the Compute-Optimal scaling laws like Kaplan and Hoffmann/Chinchilla. Please note that compute scaling laws are outside the scope of my…
2621
Algebraic graph calculus (2021)
We describe a graph-theoretic analogue of vector calculus. The linear operators of vector calculus (gradient, divergence, laplacian) correspond to the matrices naturally associated to graphs (incidence matrix, adjacency matrix). This analogy is useful for formalizing the discretization of some problems in image and…
2622
Generalizations of Fourier analysis (2021)
When I first learned about Fourier series and integrals, I hated it because it seemed like a collection of many ad-hoc definitions, formally related but very different.
2623
Remote code execution vulnerability in Google they are not willing to fix
This is a story about a security vulnerability in Google that allowed me to run arbitrary code on the computers of 50+ Google employees. Although Google initially considered my finding a serious security incident, later on, it changed its mind and stated that my finding is not, in fact, a vulnerability, but the…
2624
Stop Building Closed Ecosystems
I'm sick of it. In the world of mobile development (though likely also elsewhere), we sink so much engineering talent into making tools that are great...
2625
Systems design 2: What we hope we know
Someone asked if I could write about the rise of AI and Large Language Models (LLMs) and what I think that means for the future of people, t...
2626
Hypervisor Development in Rust
In the ever-evolving field of information security, curiosity and continuous learning drive innovation. This blog series is tailored for those deeply engaged in experimental projects, leveraging Rust’s capabilities to push the boundaries of what’s possible. The focus on Rust, after exploring various programming…
2627
The inner beauty of a Tesla Stropkov DS 3600 telephone from the early 1980s
This is a DS 3600 analog telephone made by Tesla Stropkov in about 1982. It's still functional and to my mind it's absolutely beautiful. Bu...
2628
DOOM maps to SVG to laser cutter
I’ve heard a lot about classic Doom’s data format and decided to write some Rust code to extract its maps and convert that to vector graphics I could laser cut.
2629
Low Code Software Development Is a Lie
I’ve been writing custom software for a long time and one of the things that annoys me most is when a client adopts the position that there is a silver bullet which will reduce or remove the inherent complexity of this task. This happens more often than you’d think and guess what? They are almost always wrong. Perhaps…
2630
The Decade of Deep Learning
As the 2010’s draw to a close, it’s worth taking a look back at the monumental progress that has been made in Deep Learning in this decade.[1] Driven by the development of ever-more powerful comput
2631
Surviving Burnout (2015)
the curious life of Ms. Audrey M. Eschright2023311 points162 commentsLife, Learning & Writing details
The first thing I know about burnout is this: burnout happens when we become locked in a cycle of caring about the results of our actions but having no meaningful control over those outcomes. I first learned this while reading about social workers, but the problem is abdundant in tech. I’ve seen it play out in every…
2632
Prompt injection: what’s the worst that can happen?
Activity around building sophisticated applications on top of LLMs (Large Language Models) such as GPT-3/4/ChatGPT/etc is growing like wildfire right now. Many of these applications are potentially vulnerable to prompt …
2633
2634
Zig build system
As the popularity of zig cc, more developers may realize Zig is not only a programming language, but also a toolchain to help maintaining C/C++/Zig project. Today I want to discuss one vital part of this toolchain: Zig's build system, build.zig the first non-trivial Zig program for newcomers. Note: This post is based…
2635
2636
Building LLM Applications for Production
[Hacker News discussion, LinkedIn discussion, Twitter thread]
2637
The least impactful way to spend $300M?
Buying status is as close to bad as a good act can be.
2638
SQL:2023 is finished: Here is what's new
SQL:2023 has been wrapped. The final text has been submitted by the working group to ISO Central Secretariat, and it’s now up to the ISO gods when it will be published. Based on past experience, it could be between a few weeks and a few months.
2639
2640
Don’t record your social life on an append-only social network (2022)
The tech behind peer-to-peer (P2P) social networks such as Secure Scuttlebutt enforces permanency. Absolute permanency isn’t what you want in a social network.