frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Building a Minimal Transformer for 10-digit Addition

https://alexlitzenberger.com/blog/post.html?post=/building_a_minimal_transformer_for_10_digit_addition
24•kelseyfrog•2h ago

Comments

pankajdoharey•1h ago
Looks like a Tiny Analytic transformer, RNN is arguably a better choice if you are gonna handwire an architecture to mechanically do addition. Learning is about discovering the patterns and algorithm from data. Wiring a machine to follow a procedure defeats that purpose.
wizzwizz4•1h ago
Related: https://news.ycombinator.com/item?id=36851494, discussion of https://www.evanmiller.org/attention-is-off-by-one.html (2023).
wizzwizz4•1h ago
I somewhat feel that using floating point arithmetic for what should be a symbol manipulation exercise is cheating. The deserialisation technique is interesting enough that I'm not really upset, though.

> The codex solution reversed the order which makes sense for making carry logic easy, but it is less clean.

That's the approach I'd have gone with. I've long been an advocate of little-endian numerical representations. That said, if there's a maximum number of digits, it's straightforward to implement the circuitry needed to do calculate the most-significant digit of the result in one go; and I somehow doubt the AI-generated solution really took advantage of the tricks that little-endian allows.

> At some point I set claude code on some debugging to my surprise I don’t recall it actually solving any of the bugs, it seemed much more concerned with “correcting” the funky things I was intentionally doing.

It baffles me that somebody capable of this kind of work would find this surprising. The process that allows LLMs to find bugs in code is the same process that entreats them to "correct" such creativity: their understanding of the world begins and ends at statistical plausibility, and they cannot truly comprehend things (though they can do a very good job of pretending, given sufficient training data).

lacunary•53m ago
What's the difference between comprehending and understanding in this context?

Claude surpasses ChatGPT to become the #1 app on the US App Store

https://apps.apple.com/us/iphone/charts
109•byincugnito•18m ago•22 comments

The Windows 95 user interface: A case study in usability engineering (1996)

https://dl.acm.org/doi/fullHtml/10.1145/238386.238611
77•ksec•2h ago•30 comments

MinIO Is Dead, Long Live MinIO

https://blog.vonng.com/en/db/minio-resurrect/
164•zufallsheld•3h ago•56 comments

We do not think Anthropic should be designated as a supply chain risk

https://twitter.com/OpenAI/status/2027846016423321831
104•golfer•3h ago•32 comments

Obsidian Sync now has a headless client

https://help.obsidian.md/sync/headless
360•adilmoujahid•7h ago•129 comments

The happiest I've ever been

https://ben-mini.com/2026/the-happiest-ive-ever-been
280•bewal416•2d ago•129 comments

The war against PDFs is heating up

https://www.economist.com/business/2026/02/24/the-war-against-pdfs-is-heating-up
15•pseudolus•1h ago•14 comments

Block the “Upgrade to Tahoe” Alerts

https://robservatory.com/block-the-upgrade-to-tahoe-alerts-and-system-settings-indicator/
103•todsacerdoti•5h ago•37 comments

Addressing Antigravity Bans and Reinstating Access

https://github.com/google-gemini/gemini-cli/discussions/20632
193•RyanShook•10h ago•155 comments

Verified Spec-Driven Development (VSDD)

https://gist.github.com/dollspace-gay/d8d3bc3ecf4188df049d7a4726bb2a00
138•todsacerdoti•7h ago•66 comments

Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers

https://venturebeat.com/technology/alibabas-new-open-source-qwen3-5-medium-models-offer-sonnet-4-...
160•lostmsu•4h ago•105 comments

Woxi: Wolfram Mathematica Reimplementation in Rust

https://github.com/ad-si/Woxi
229•adamnemecek•3d ago•101 comments

Building a Minimal Transformer for 10-digit Addition

https://alexlitzenberger.com/blog/post.html?post=/building_a_minimal_transformer_for_10_digit_add...
24•kelseyfrog•2h ago•4 comments

New evidence that Cantor plagiarized Dedekind?

https://www.quantamagazine.org/the-man-who-stole-infinity-20260225/
103•rbanffy•3d ago•66 comments

Show HN: Xmloxide – an agent made rust replacement for libxml2

https://github.com/jonwiggins/xmloxide
3•jawiggins•42m ago•0 comments

Werner Herzog Between Fact and Fiction

https://www.thenation.com/article/culture/werner-herzog-future-truth/
60•Hooke•1d ago•11 comments

Deterministic Programming with LLMs

https://www.mcherm.com/deterministic-programming-with-llms.html
6•todsacerdoti•3d ago•2 comments

Show HN: Now I Get It – Translate scientific papers into interactive webpages

https://nowigetit.us
171•jbdamask•10h ago•98 comments

The whole thing was a scam

https://garymarcus.substack.com/p/the-whole-thing-was-scam
540•guilamu•7h ago•155 comments

Pentagon chief blocks officers from Ivy League schools and top universities

https://fortune.com/2026/02/28/pentagon-officer-education-ivy-league-schools-universities-partner...
15•geox•35m ago•1 comments

The archivist preserving decaying floppy disks

https://www.popsci.com/technology/floppy-disk-archivist-project/
42•Brajeshwar•3d ago•4 comments

747s and Coding Agents

https://carlkolon.com/2026/02/27/engineering-747-coding-agents/
117•cckolon•1d ago•55 comments

From Noise to Image – interactive guide to diffusion

https://lighthousesoftware.co.uk/projects/from-noise-to-image/
98•simedw•2d ago•15 comments

Stop Burning Your Context Window – How We Cut MCP Output by 98% in Claude Code

https://mksg.lu/blog/context-mode
204•mksglu•14h ago•49 comments

Ghosts'n Goblins – “Worse danger is ahead”

https://superchartisland.com/ghostsn-goblins/
56•elvis70•3d ago•24 comments

What I learned while trying to build a production-ready nearest neighbor system

https://github.com/thatipamula-jashwanth/smart-knn
16•Jashwanth01•3d ago•7 comments

Unsloth Dynamic 2.0 GGUFs

https://unsloth.ai/docs/basics/unsloth-dynamic-2.0-ggufs
196•tosh•15h ago•51 comments

The Eternal Promise: A History of Attempts to Eliminate Programmers

https://www.ivanturkovic.com/2026/01/22/history-software-simplification-cobol-ai-hype/
224•dinvlad•3d ago•159 comments

Our Agreement with the Department of War

https://openai.com/index/our-agreement-with-the-department-of-war
187•surprisetalk•3h ago•169 comments

The Future of AI

https://lucijagregov.com/2026/02/26/the-future-of-ai/
112•BerislavLopac•13h ago•89 comments