4141
4142
Emergent abilities of large language models
Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
4143
How China's “debt traps” work
Raiding the Global South like a piggy bank.
4144
Don't write clean code, write CRISP code
“Clean code” is so over: all the cool kids are switching to CRISP code . I’ll show you how to achieve it in five catchy, T-shirt-friendly slogans.
4145
Real vs. fake AirPods with industrial CT
CT scans of counterfeit AirPods Pro and MagSafe chargers reveal pouch cell batteries in circular cavities, internal weights, and a fake ground pin.
4146
How to reduce stress and overcome anxiety as a software developer
I share my story about stress and anxiety and how I managed to overcome it. I also share practical tips and resources that will help you begin your healing journey today!
4147
Phoenix 1.7 for Elixir: Edit a Form in a Modal
In the final part of our series, we'll implement an edit modal.
4148
Gitlab and Google Cloud Partner to Expand AI-Assisted Capabilities
GitLab AI-powered security feature to leverage Google Cloud's foundation models to empower customers to identify and remediate vulnerabilities more efficiently while maintaining privacy standards...
4149
4150
Don't let event-driven architecture buzzwords fool you
I see a lot of new terms like Command-Event, Event-Based Compute, etc., presented around Event-Driven Architecture. Let me clear that up because there are no such things…
4151
H100 GPUs Set Standard for Gen AI in Debut MLPerf Benchmark
In a new industry-standard benchmark, a cluster of 3,584 H100 GPUs at cloud service provider CoreWeave completed a massive GPT-3-based benchmark in just 11 minutes.
4152
Rust's Golden Rule
I find myself thinking about a particular design principle of Rust today. I’m not sure I’ve ever seen it named specifically before, but it gets referred to from…
4153
Improving mathematical reasoning with process supervision
We’ve trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning (“process supervision”) instead of simply rewarding the correct final answer (“outcome supervision”). In addition to boosting performance relative to outcome supervision, process…
4154
Everything I've learned building the fastest Arm desktop
This is the fastest Arm desktop in the world, yes, even faster than the M2 Ultra Mac Pro. And today, I made it even faster. I upgraded everything: Faster RAM, 128 core CPU, 40 series GPU, I did it all, and we'll see how much we can obliterate the M2 Mac Pro. 128 cores—that's five times more cores, I'm also going to…
4155
Godot: A Collaboration with Google and the Forge
We are excited to announce that we have partnered with Google and The Forge to bring some helpful performance optimizations to our Vulkan mobile backend.
4156
Bayes is guaranteed to overfit
The myth A few years ago I saw a StackOverflow question (but I cannot find it now): allegedly Andrew Gelman blogged that Bayesian models would not overfit and would only underfit (I also cannot find Andrew’s blog post now but I think what he meant by “Bayesian models only underfits” is that you can never fully model…
4157
Why write?
As a reaction to the previous article on how to write a scientific paper, I got some reactions from friends and colleagues, the underlying message being that they would rather like to understand why writing can be important. There is indeed a tendency to see academic writing as a tedious task that one should try to…
4158
Session recording for Tailscale SSH in beta
This guide shows you how to set up session recordings for Tailscale SSH. Learn how to record the terminal output whenever someone on your tailnet initiates a Tailscale SSH connection. Use these recordings to detect threats, investigate security incidents, and remain compliant with your network security policies.
4159
8086 Processor's microcode pipeline from die analysis
Intel introduced the 8086 microprocessor in 1978, and its influence still remains through the popular x86 architecture. The 8086 was a fai...
4160
Unpacking Elixir: Real-Time and Latency
Elixir was built on Erlang. Erlang was built to provide "consistently low latency" and a few other audacious goals. Note, this is not a hard realtime constraint. It is soft, squishy and yet, important and real. It makes Erlang unusually suitable to systems where latency matters and where a near-realtime experience is…
4161
Trying chDB, an embeddable ClickHouse engine
A lightweight in-process SQL OLAP engine, powered by ClickHouse.
4162
US homelessness up 12% from 2022, hits highest level since 2007
A data-driven examination of the 12% increase in US homelessness, the structural causes behind the numbers, and what the data tells us about housing policy failures.
4163
Unions work
The WGA's win may pave the way for more.
4164
There is no hard takeoff
Back in 2014, Elon Musk referred to AI as summoning the demon. And it wasn’t hard to see that view. Soon, Go agents would beat top humans learning from self play. By the end of 2017, the same algorithm mastered Chess and Shogi. By 2020, it didn’t even need tons of calls to the simulator, and could play Atari too.
4165
Awareness at Death
How many cardiac arrest survivors later have a conscious recollection of their death experience—and are their recollections accurate?
4166
4167
4168
4169
You can serve static data over HTTP
This will be short. I need to let some steam off and flush this thing out of my mind because it's already hanging around for too long. The story goes like this: we're prototyping small tool for whatever and we're at the point where there is a visible need to store some (most likely) static data. Oh, the thing is in…
4170
