Hacker News: dev_hugepages

Are We Harold Bloom?

dev_hugepages — Wed, 10 Jun 2026 12:12:52 +0000

Article URL: https://abner.page/post/are-we-harold-bloom/

Comments URL: https://news.ycombinator.com/item?id=48475118

Points: 1

# Comments: 0

New comment by dev_hugepages in "My Agent Skill for Test-Driven Development"

dev_hugepages — Sat, 06 Jun 2026 09:04:34 +0000

Then, you are obsolete in the AI age.

New comment by dev_hugepages in "How LLMs work"

dev_hugepages — Sat, 06 Jun 2026 08:55:08 +0000

Predicting a word is the final objective, as in the output of the model is a probability distribution of the next token. However, choosing the right token is more complicated than just regurgitating the training data (and you won't encounter an exact example in the training data, so you need to interpolate). This makes the model learn abstract representation of things that it is able to manipulate before outputting this back into token. RL also complicates this because the "fitness" is now some arbitrary metric computed over an entire sequence of tokens.

Many portions of Magnifica Humanitas appear to be AI-written

dev_hugepages — Wed, 27 May 2026 13:07:37 +0000

Article URL: https://www.lesswrong.com/posts/GbWwesBnetyiomxEH/many-portions-of-magnifica-humanitas-appear-to-be-ai-written

Comments URL: https://news.ycombinator.com/item?id=48293720

Points: 3

# Comments: 1

New comment by dev_hugepages in "The main thing about P2P meth is that there's so much of it (2021)"

dev_hugepages — Sat, 16 May 2026 08:01:03 +0000

He's talking about the derivatives (nicknamed Pyros because they're related to Pyrovalerone) like α-PVP, MDPV and others.

New comment by dev_hugepages in "Switchlab.dev – Learn switching and routing through interactive labs"

dev_hugepages — Thu, 14 May 2026 05:45:05 +0000

So this is obviously vibe coded, may we obtain the source to see how that works?

New comment by dev_hugepages in "DeepSeek 4 Flash local inference engine for Metal"

dev_hugepages — Sun, 10 May 2026 17:35:37 +0000

Probably a provider thing. Looking at https://help.kagi.com/kagi/ai/llms-privacy.html, they're using deepinfra.

Looking at https://openrouter.ai/deepseek/deepseek-v4-flash/providers tells us that the deepseek provider achieves 49tps of throughput while deepinfra 19tps.

New comment by dev_hugepages in "RaTeX: KaTeX-compatible LaTeX rendering engine in pure Rust"

dev_hugepages — Thu, 07 May 2026 17:32:56 +0000

If anyone else is looking to make graph with typst, this can be done with https://cetz-package.github.io/ -- which is inspired by Tikz from the LaTeX world -- or https://lilaq.org/ which seems more appropriate for this type of data plotting

New comment by dev_hugepages in "RaTeX: KaTeX-compatible LaTeX rendering engine in pure Rust"

dev_hugepages — Thu, 07 May 2026 15:40:38 +0000

I took a look at the repo and it's probably the fault of the the SVG of the graphs, not of typist itself. Now, you could have used typst libraries to generate the graphs but back then (2 years ago I think?) it was probably a struggle.

New comment by dev_hugepages in "Three Inverse Laws of AI"

dev_hugepages — Wed, 06 May 2026 08:30:45 +0000

The problem is that humans use this as a coping mechanism for things they don't understand: I don't understand why the printer doesn't work, so I give it a mind of its own.

This is harmless for inconsequential stuff like a chair, but when it's an LLM, people should at least understand it's behavior so they don't get trapped. That means not trusting it with advice meant for the user or on things it has no concept of, like time or self-introspection (people ask the LLM after it acted, "Why did you delete my database?" when it has limited understanding of its own processing, so it falls back to, "You're right, I deleted the database. Here's what I did wrong: ... This is an irrecoverable mistake, blah, blah, blah..."

New comment by dev_hugepages in "DeepClaude – Claude Code agent loop with DeepSeek V4 Pro"

dev_hugepages — Mon, 04 May 2026 05:38:52 +0000

And now, because we all upvoted and commented on it, the vibe coded slop of the new user is on the front page now.

Text Files as a User Interface

dev_hugepages — Sun, 03 May 2026 09:49:51 +0000

Article URL: https://ratfactor.com/cards/text-files-as-ui

Comments URL: https://news.ycombinator.com/item?id=47995287

Points: 2

# Comments: 0

New comment by dev_hugepages in "Functional programmers need to take a look at Zig"

dev_hugepages — Thu, 30 Apr 2026 07:33:44 +0000

Unrelated, but I was pleased at how fast the page opened: it felt pretty much instantaneous!

I opened the network log, disabled cache and reloaded to see it only transferred 8kb.

Keep up the good work!

Sloppy Copies

dev_hugepages — Fri, 24 Apr 2026 13:52:58 +0000

Article URL: https://www.markround.com/blog/2026/04/19/sloppy-copies/

Comments URL: https://news.ycombinator.com/item?id=47890339

Points: 70

# Comments: 19

New comment by dev_hugepages in "An update on recent Claude Code quality reports"

dev_hugepages — Fri, 24 Apr 2026 11:45:16 +0000

No, the cache is a few GB large for most usual context sizes. It depends on model architecture, but if you take Gemma 4 31B at 256K context length, it takes 11.6GB of cache

note: I picked the values from a blog and they may be innacurate, but in pretty much all model the KV cache is very large, it's probably even larger in Claude.

New comment by dev_hugepages in "Alberta startup sells no-tech tractors for half price"

dev_hugepages — Wed, 22 Apr 2026 16:59:57 +0000

On HN, I often see comments like this, complaining about Cloudflare blocking access to pages. It makes me wonder if it’s due to a particular setup that triggers bot detection – like Tor or no-JS – that HN readers often use, or if Cloudflare has too many false positives.

New comment by dev_hugepages in "GPT‑5.4 Mini and Nano"

dev_hugepages — Wed, 18 Mar 2026 07:07:05 +0000

Time to gather a dataset and train your own model!

New comment by dev_hugepages in "this css proves me human"

dev_hugepages — Sat, 07 Mar 2026 06:32:35 +0000

Yeah the habit of discarding typography and polish as a "proof of humanity" is worrying to say the least

New comment by dev_hugepages in "AI generated music barred from Bandcamp"

dev_hugepages — Wed, 14 Jan 2026 10:00:05 +0000

I’m more hopeful that MIDI completion/in-filling models will be easier for musicians to control and use. But right now, the most popular tools are things like Suno, where you barely have any control and it spits out an entire, possibly mediocre song. It’s the same vein as ChatGPT image generation vs. Stable Diffusion, where you can do much more controllable inpaints with the latter.

New comment by dev_hugepages in "There's a ridiculous amount of tech in a disposable vape"

dev_hugepages — Wed, 14 Jan 2026 08:57:39 +0000

Nicotine itself is carcinogenic in the mouth:

> Nicotine in tobacco can form carcinogenic tobacco-specific nitrosamines through a nitrosation reaction. This occurs mostly in the curing and processing of tobacco. However, nicotine in the mouth and stomach can react to form N-nitrosonornicotine, a known type 1 carcinogen, suggesting that consumption of non-tobacco forms of nicotine may still play a role in carcinogenesis