HNHacker News
TopNewBestAskShowJobs

koakuma-chan

1,454 karma · joined October 14, 2024

light.beach8865@fastmail.com

https://github.com/koakuma-chan

submissionscomments
koakuma-chan··on Show HN: Xmloxide – an agent-made Rust replacement for libxml2
Please stop spreading this "AI evals" terminology. "evals" is what providers like OpenAI and Anthropic do with their models. If you wrote a test for a feature that uses an LLM, it's just a test, there's no need to say "evals." Having a separate term only further confuses people who already have no idea what that actually means.
koakuma-chan··on President Trump bans Anthropic from use in government systems
Does Grok suck? I have never used it.
koakuma-chan··on I am directing the Department of War to designate Anthropic a supply-chain risk
Both?
koakuma-chan··on I am directing the Department of War to designate Anthropic a supply-chain risk
$1,000,000 doesn't seem like a lot of money for them, why would it matter to them?
koakuma-chan··on We gave terabytes of CI logs to an LLM
This seems really weird to me. Isn't that just using LLMs in a specific way? Why come up with a new name "RLM" instead of saying "LLM"? Nothing changes about the model.
koakuma-chan··on In 2025, Meta paid an effective federal tax rate of 3.5%
Too brief, minus 10 marks.
koakuma-chan··on You Want to Visit the UK? You Better Have a Google Play or App Store Account
It works for filling an existing password, but not for creating a new one, iOS still prompts me to fill existing even though I'm on the sign up page. On the web 1Password can also automatically generate a Fastmail masked email address, but I doubt there's any hope for that to work in a native app.
koakuma-chan··on You Want to Visit the UK? You Better Have a Google Play or App Store Account
Well then I don't know whom to blame, 1Password or app developers not marking the fields correctly.
koakuma-chan··on You Want to Visit the UK? You Better Have a Google Play or App Store Account
And why is it broken? Is there a way for a password manager app to somehow inspect other apps and identify forms within them and interact with the forms?
koakuma-chan··on You Want to Visit the UK? You Better Have a Google Play or App Store Account
No, installing apps is just genuinely bad UX, at least for me, because my password manager doesn't work in apps, and I have to manually generate and copy and paste password. If creators of the app are extra stupid, they make it so that you cannot paste into the password input, so I have to enter the generated password character by character.
koakuma-chan··on An autopsy of AI-generated 3D slop
But what business value does high quality code bring? /s
koakuma-chan··on You Want to Visit the UK? You Better Have a Google Play or App Store Account
It's more convenient to open a website than it is to install an app.
koakuma-chan··on You Want to Visit the UK? You Better Have a Google Play or App Store Account
But why do they need to make a native app for that? It makes no sense.
koakuma-chan··on Tell HN: YC companies scrape GitHub activity, send spam emails to users
I have been having the same experience. If you starred a GitHub repo, and they think that their product is similar, they will send you their spam. I condemn this! They should be ashamed!
koakuma-chan··on Show HN: Shibuya – A High-Performance WAF in Rust with eBPF and ML Engine
"The most advanced open-source WAF ever built."

Somehow, the moment I read this, I knew it was AI slop.

koakuma-chan··on Show HN: Shibuya – A High-Performance WAF in Rust with eBPF and ML Engine
What the fuck is this slop?

https://github.com/theghostshinobi/Shibuya-waf-light-version...

koakuma-chan··on Show HN: AI Timeline – 171 LLMs from Transformer (2017) to GPT-5.3 (2026)
Nope, no bigger picture. That's all I meant.
koakuma-chan··on Show HN: AI Timeline – 171 LLMs from Transformer (2017) to GPT-5.3 (2026)
It depends on the instruction, and how many other instructions there are. Models converge on doing things the way that emerged from their training, and with every turn the model cares less and less about your instructions. In practice, this means that after you had the model plan and execute the plan, you almost always end up having to iterate on the output because during the process of outputting the output the model began to derail and ignore instructions. You get things like "In a real app, we would do X, for now, just return null" or various subtle bugs.

It makes sense if you remember that it just predicts, what should probably be the next piece of text?

koakuma-chan··on Show HN: AI Timeline – 171 LLMs from Transformer (2017) to GPT-5.3 (2026)
No, I am making the argument that models have poor capabilities outside of tasks they are RLed for, and their capabilities inside those tasks are only as good as capabilities of people evaluating their responses, i.e. not great. Even if you instruct the model "don't do X" or "do X this way"—you cannot rely on the model following that instruction. This means that there is nothing you can do if model makes "errors."

Not necessarily relevant, but fun, I had the ChatGPT model correct itself mid-response when checking my math work. It started by saying that I was wrong, then it proceeded to solve the problem and at the end it realized that I was correct.

koakuma-chan··on Show HN: AI Timeline – 171 LLMs from Transformer (2017) to GPT-5.3 (2026)
> If you're seeing the model deliberately creating errors so you have something to fix, then that sounds like something is fundamentally wrong in your prompt.

No, all these models are just bad for anything that they weren't RLed for, and decent for things they were. Decent, because people who evaluate them aren't experts.

koakuma-chan··on Why is Claude an Electron app?
Jarred has 4.8K GitHub issues of his own.
koakuma-chan··on Gemini 3.1 Pro
Who they? Do the engineers who actually did that work at Google still? I heard that the guy who made TPUs has his own startup now.
koakuma-chan··on South Korean ex president Yoon Suk Yeol jailed for life for leading insurrection
It is mind blowing. I guess he didn't have enough allies in power? All the corrupt politicians around the world must be laughing at him right now.
koakuma-chan··on South Korean ex president Yoon Suk Yeol jailed for life for leading insurrection
How did they stop him?
koakuma-chan··on Minecraft Java is switching from OpenGL to Vulkan
Have you played Bedrock? It sucks.
koakuma-chan··on What years of production-grade concurrency teaches us about building AI agents
100% agreed
koakuma-chan··on What years of production-grade concurrency teaches us about building AI agents
You Should Just Use Rust
koakuma-chan··on OpenClaw is dangerous
Would they? I don't think anyone cares about security.
koakuma-chan··on Windows: Prefer the Native API over Win32
Let me narrow down the scope here. I am a Rust developer, developing software that will run on my Linux server. Why would I want to use libc? Why does Rust standard library use libc? Zig, for example, doesn't.
koakuma-chan··on Windows: Prefer the Native API over Win32
> It's like coding to the Linux syscall interface instead of libc.

The right thing to do? I don't see why I would want to use libc.

← PreviousPage 3 of 34Next →