frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Executing programs inside transformers with exponentially faster inference

https://www.percepta.ai/blog/can-llms-be-computers
85•u1hcw9nx•1d ago

Comments

andy12_•20h ago
This seems a really interesting path for interpretability, specially if a big chunk of a model's behavior occurs pseudo-symbolically. This is an idea I had thought about, integrating tools into the main computation path of a model, but I never imagined that it could be done efficiently with just a vanilla transformer.

Truly, attention is all you need (I guess).

galsapir•18h ago
one of the most interesting pieces I've read recently. Not sure I agree with all the statements there (e.g. without execution the system has no comprehension) - but extremely cool
pennomi•18h ago
It makes sense that a next token predictor could execute assembly code. This is fascinating work, especially with the memory implementation.
koolala•1h ago
I'd like to see this combined with reinforcement learning to optimize models to think computationally. Generating ideas with hypothetical results and then running them in the same thought. Their solution sounded like a lot of tokens though.
behehebd•58m ago
Is this genius? Or just a new binary executable format? Can't tell.
mirekrusin•49m ago
This is brilliant, game changing level.

Hey, give it also access to the dump of its weights and way to propose updates so it can see and tinker its brain directly.

ThouYS•42m ago
what!
bonoboTP•38m ago
This shows the downside of using AI to write up your project. I see the eloquent sentences, but don't get the message.

> This works, but the actual execution happened outside the model. The model specified the computation, then waited for an external system to carry it out. > Our transformer also emits a program, but instead of pausing for an external tool, it executes that program itself, step by step, within the same transformer.

What's the benefit? Is it speed? Where are the benchmarks? Is it that you can backprop through this computation? Do you do so?

Why is it good that it's "inside" the model? Just making it more elegant and nice? The tool was already "inside" the overall hybrid system. What's the actual problem?

famouswaffles•23m ago
>This shows the downside of using AI to write up your project. I see the eloquent sentences, but don't get the message.

Not really sure what this obsession with calling things you don't like AI generated is but it's poor form. If you have something to say about the text then say it. Otherwise leave baseless accusations out of it.

>What's the benefit? Is it speed? Where are the benchmarks? Is it that you can backprop through this computation? Do you do so?....

It's pretty clearly an ideological thing. Some people are firmly on the 'some sort of symbolic logic is necessary' camp. From the article, 'A system that cannot compute cannot truly internalize what computation is.'

Some things are just interesting for the sake of it. This is one of those things. I don't agree with the authors on the above and I'm still glad they shared. It's a very interesting read regardless.

entropi•7m ago
I got the same impression as the parent post. Even if its not AI-generated, the text reads like a politician's speech at a lot of places. Talks a lot, says little.

The idea itself was very cool, so I endured it. But it was not a pleasant read.

andy12_•18m ago
Honestly, the most interesting thing here is definitely that just 2D heads are enough to do useful computation (at least they are enough to simulate an interpreter) and that there is an O(log n) algorithm to compute argmax attention with 2D heads. It seems that you could make an efficient pseudosymbolic LLM with some frozen layers that perform certain deterministic operations, but also other layers that are learned.
ndxone•38m ago
big question is how efficient is this compare to executing assembly on CPU
MattPalmer1086•36m ago
Interesting... But why? What is the benefit, other than increasing our understanding of model architectures?

Our brains can also simulate turing machines, slowly. We automated that with computers that are faster and more reliable. So why not allow a model to use external much faster and reliable tools, just as we do?

Bucketsquatting Is (Finally) Dead

https://onecloudplease.com/blog/bucketsquatting-is-finally-dead
31•boyter•1h ago•16 comments

Willingness to look stupid

https://sharif.io/looking-stupid
316•Samin100•3d ago•110 comments

Executing programs inside transformers with exponentially faster inference

https://www.percepta.ai/blog/can-llms-be-computers
85•u1hcw9nx•1d ago•13 comments

Malus – Clean Room as a Service

https://malus.sh
1260•microflash•19h ago•456 comments

“This is not the computer for you”

https://samhenri.gold/blog/20260312-this-is-not-the-computer-for-you/
438•MBCook•7h ago•172 comments

Vite 8.0 Is Out

https://vite.dev/blog/announcing-vite8
263•kothariji•4h ago•65 comments

Prefix sums at gigabytes per second with ARM NEON

https://lemire.me/blog/2026/03/08/prefix-sums-at-tens-of-gigabytes-per-second-with-arm-neon/
38•mfiguiere•4d ago•3 comments

Hyperlinks in Terminal Emulators

https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
54•nvahalik•6h ago•37 comments

ATMs didn’t kill bank teller jobs, but the iPhone did

https://davidoks.blog/p/why-the-atm-didnt-kill-bank-teller
416•colinprince•18h ago•439 comments

Bubble Sorted Amen Break

https://parametricavocado.itch.io/amen-sorting
332•eieio•16h ago•99 comments

Shall I implement it? No

https://gist.github.com/bretonium/291f4388e2de89a43b25c135b44e41f0
1260•breton•12h ago•473 comments

Reversing memory loss via gut-brain communication

https://med.stanford.edu/news/all-news/2026/03/gut-brain-cognitive-decline.html
308•mustaphah•16h ago•122 comments

Understanding the Go Runtime: The Scheduler

https://internals-for-interns.com/posts/go-runtime-scheduler/
110•valyala•3d ago•11 comments

IMG_0416 (2024)

https://ben-mini.com/2024/img-0416
88•TigerUniversity•3d ago•17 comments

The Met releases high-def 3D scans of 140 famous art objects

https://www.openculture.com/2026/03/the-met-releases-high-definition-3d-scans-of-140-famous-art-o...
282•coloneltcb•17h ago•54 comments

Document poisoning in RAG systems: How attackers corrupt AI's sources

https://aminrj.com/posts/rag-document-poisoning/
118•aminerj•19h ago•46 comments

Worldwide Sidewalk Joy: Adding whimsy to neighborhoods

https://worldwidesidewalkjoy.com
15•NaOH•3d ago•3 comments

Celebrating Interesting Flickr Technologies

https://medium.com/@brightcarvings/celebrating-flickr-technology-3c93c8ddecc2
33•steerpike•1d ago•8 comments

Never Snooze a Future

https://jacko.io/snooze.html
13•vinhnx•4d ago•2 comments

US private credit defaults hit record 9.2% in 2025, Fitch says

https://www.marketscreener.com/news/us-private-credit-defaults-hit-record-9-2-in-2025-fitch-says-...
357•JumpCrisscross•20h ago•411 comments

Specimen Gallery – CC0 transparent specimen PNGs organized by taxonomy

https://specimen.gallery/
4•eclectic_mind05•3d ago•2 comments

Ceno, browse the web without internet access

https://ceno.app/en/index.html?
7•mohsen1•3h ago•2 comments

Grief and the AI split

https://blog.lmorchard.com/2026/03/11/grief-and-the-ai-split/
145•avernet•11h ago•225 comments

Bringing Chrome to ARM64 Linux Devices

https://blog.chromium.org/2026/03/bringing-chrome-to-arm64-linux-devices.html
106•ingve•13h ago•47 comments

Innocent woman jailed after being misidentified using AI facial recognition

https://www.grandforksherald.com/news/north-dakota/ai-error-jails-innocent-grandmother-for-months...
608•rectang•12h ago•309 comments

Big data on the cheapest MacBook

https://duckdb.org/2026/03/11/big-data-on-the-cheapest-macbook
352•bcye•21h ago•278 comments

Can you instruct a robot to make a PBJ sandwich?

https://pbj.deliberateinc.com/
27•mooreds•6h ago•31 comments

WolfIP: Lightweight TCP/IP stack with no dynamic memory allocations

https://github.com/wolfssl/wolfip
128•789c789c789c•17h ago•23 comments

Are LLM merge rates not getting better?

https://entropicthoughts.com/no-swe-bench-improvement
149•4diii•21h ago•135 comments

Launch HN: IonRouter (YC W26) – High-throughput, low-cost inference

https://ionrouter.io
63•vshah1016•14h ago•25 comments