frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Ask HN: Analog Model of Transformers

7•JPLeRouzic•1d ago
(Sorry if this a stupid question)

Roughly 80 years ago many computers used analog logic, for example an amplifier with variable resistors could multiply and divide.

The matrix multiplication seems to be central in Transformer architecture.

I wonder if it would make sense to create a sort of Transformer on a vaguely similar concept?

If you wonder why this question, after all there are still guys who design digital computers with tubes or even mechanical relays, then why not analog Transformers?

https://hackaday.com/2023/12/05/a-single-board-computer-with-vacuum-tubes/

https://hackaday.io/project/189725-homebrew-16-bit-relay-computer

Comments

SaberTail•21h ago
The term I've heard for this sort of thing is "Physical Neural Networks" or "PNN"s. My impression is that one of the big things holding them back is that because we can't manufacture components to perfect tolerances, you can't train a single model and reuse it like you can with digital logic. Even if you can get close, every single circuit needs some amount of tuning. And we haven't worked out great ways to train them.

There's a lot of research going on in this space though, because yeah, nature can solve certain mathematical problems more efficiently than digital systems.

There's a decent review article that came out recently: https://www.nature.com/articles/s41586-025-09384-2 or https://arxiv.org/html/2406.03372v1

mikewarot•19h ago
Deep networks these days can have hundreds of layers. Any non-digital system is going to pick up noise and gain variations at each stage. It might be possible to get around this with threshold circuits.

The problem then becomes training. The algorithm of choice is back propagation, which requires determining derivatives across the whole network. Doing training on an analog system would require tweaking each value as a batch of values is input multiple times to find the slope. This is impractical for large networks, as training usually requires a billion rounds.

Ask HN: Is Antigravity code search dropping results recently?

3•sankalpnarula•35m ago•0 comments

Ask HN: How do you feel when your coding assistant loses context?

3•noduerme•3h ago•8 comments

Tell HN: Slow Down

10•jacquesm•3h ago•4 comments

Claude Is Down Again

3•Venkymatam•1h ago•0 comments

Lazy Tmux – Lazy-loading tmux sessions with a tree view

2•Alchemmist•5h ago•0 comments

Claude Is Down

12•haebom•2h ago•8 comments

LLMs learn what programmers create, not how programmers work

37•noemit•1d ago•14 comments

Ask HN: AI productivity gains – do you fire devs or build better products?

107•Bleiglanz•3d ago•199 comments

Ask HN: Any recommended engineering/dev related Slack channels?

2•Kuraptka•11h ago•1 comments

Ask HN: How do you offload all coding to AI?

9•makingstuffs•15h ago•11 comments

Tell HN: Russians may soon lose access to the global internet

32•taminka•20h ago•14 comments

Ask HN: Do you feel less happy when coding with agent?

4•zane__chen•16h ago•9 comments

Does nobody care about not being able to copy from Slack anymore?

5•neal_caffrey•16h ago•3 comments

Is Trusttunnel easy for people to use?

2•AnonyMD•17h ago•0 comments

Ask HN: Founders of estonian e-businesses – is it worth it?

11•udl•1d ago•4 comments

Ask HN: Does the World need more software?

3•Vektorceraptor•18h ago•9 comments

Ask HN: Is anyone here also developing "perpetual AI psychosis" like Karpathy?

29•jawerty•1d ago•24 comments

Ask HN: Is using AI tooling for a PhD literature review dishonest?

9•latand6•1d ago•26 comments

Does it make sense to ask Blackberry to re-license ancient QNX sources?

4•ymz5•23h ago•3 comments

Ask HN: Analog Model of Transformers

7•JPLeRouzic•1d ago•2 comments

Ask HN: $50 monthly budget, which coding models would you recommend now?

10•klueinc•1d ago•18 comments

Ask HN: How does one get rich in 2026?

7•roschdal•4h ago•4 comments

Tell HN: H&R Block tax software installs a TLS backdoor

150•yifanlu•4d ago•12 comments

Ask HN: Is the AI software developer demand destruction narrative accurate?

4•RyanShook•15h ago•2 comments

Tell HN: MS365 upgrade silently to 25 licenses, tried to charge me $1,035

24•davidstarkjava•3d ago•8 comments

Anonymize / de-identify LLM chat history export, post-processing

2•msiraj1•1d ago•1 comments

Veevo Health – book a CT angiogram to see plaque buildup in your arteries

5•arvindsr33•1d ago•3 comments

Ask HN: If there has been no prompt injection, is it safe?

7•sayYayToLife•2d ago•8 comments

SparkVSR: Video Super-Resolution You Can Control with Keyframes

3•steveharing1•2d ago•0 comments

Anyone know how long it will take to re-start Qatar's helium plants?

9•megamike•3d ago•6 comments