frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

xAI to Repay $17.5B Debt as SpaceX IPO Nears

https://finance.yahoo.com/news/musk-xai-buy-back-3-173531157.html
1•andsoitis•38s ago•0 comments

Show HN: StreamHouse – Open-source Kafka alternative

https://github.com/gbram1/streamhouse
1•gbram•59s ago•0 comments

Tell HN: AI coding is not for the impatient

1•ghoshbishakh•1m ago•0 comments

Show HN: Partnership Intel – Find partners for your products, faster

https://partnershipintel.com
1•tejas3732•2m ago•0 comments

We Scanned 50 Cursor Rules Files From GitHub. 6 Had Hidden Instructions.

https://agentseal.org/blog
2•voxadam•2m ago•0 comments

'Virtual cell' captures most-basic process of life: bacterial division

https://www.nature.com/articles/d41586-026-00786-4
2•bookofjoe•3m ago•1 comments

Meta's Race to Scale AI Chips for Billions: Four Chips in Two Years

https://ai.meta.com/blog/meta-mtia-scale-ai-chips-for-billions/?_fb_noscript=1
2•surprisetalk•5m ago•0 comments

Go in 9×9 is Awesome

https://entropicthoughts.com/go-in-9x9-is-awesome
3•surprisetalk•5m ago•0 comments

A Recursive Algorithm to Render Signed Distance Fields

https://pointersgonewild.com/2026-03-06-a-recursive-algorithm-to-render-signed-distance-fields/
3•surprisetalk•5m ago•0 comments

Work hard vs. play hard (2024)

https://taylor.town/work-hard-play-hard
3•surprisetalk•5m ago•0 comments

Live Nation CEO urged by ticked-off judge to settle with states

https://nypost.com/2026/03/10/business/live-nation-ceo-urged-by-frustrated-judge-to-settle-with-s...
3•1vuio0pswjnm7•5m ago•0 comments

25 Years of ADSL Speed

https://brainbaking.com/post/2026/03/25-years-of-adsl-speed/
2•speckx•5m ago•0 comments

UK MPs give ministers powers to restrict Internet for under 18s

https://www.openrightsgroup.org/press-releases/mps-give-ministers-powers-to-restrict-entire-inter...
4•robtherobber•7m ago•0 comments

Ig Nobel Prize flees US for Switzerland after 35 years over safety concerns

https://www.theregister.com/2026/03/11/ig_nobel_prize_leaves_us/
3•beardyw•7m ago•1 comments

Show HN: PayrollEngine – Open-source regulation-based payroll framework (.NET)

https://payrollengine.org/
3•payrollengine•7m ago•0 comments

WA income tax clears House after 24-hour debate

https://www.seattletimes.com/seattle-news/politics/wa-income-tax-passes-house-after-24-hour-debate/
5•garbawarb•7m ago•0 comments

Workflow to build context for coding agents

2•wek•11m ago•0 comments

Show HN: AIRiskCalc – AI-Powered Health Risk Calculators

https://www.airiskcalc.com
2•mosbyllc•11m ago•0 comments

Show HN: Trackless Links – clean URLs, no tracking (Safari)

https://github.com/aloth/trackless-links
2•xlth•12m ago•0 comments

Why Apple Rejected a Clamshell-Style Foldable iPhone

https://www.macrumors.com/2026/03/10/apple-clamshell-foldable-iphone/
2•mgh2•12m ago•0 comments

9legacy is a continuation of Plan 9 from Bell Labs

http://9legacy.org/
3•cestith•13m ago•0 comments

Show HN: Liteparser – a complete SQLite parser in C

https://marcobambini.substack.com/p/liteparser-a-fast-embeddable-sqlite
3•marcobambini•13m ago•0 comments

Scientists revive activity in frozen mouse brains for the first time

https://www.nature.com/articles/d41586-026-00756-w
5•tzury•15m ago•2 comments

Show HN: AgentSign – Open-source zero trust engine for AI agents

https://github.com/razashariff/agentsign
2•AskCarX•18m ago•0 comments

A Snapshotable WASM Interpreter

https://github.com/friendlymatthew/gabagool
2•friendlymatthew•18m ago•1 comments

Agentic Engineering: The good, the bad, the ugly

https://substack.com/home/post/p-189354081
2•ixeption•18m ago•0 comments

Show HN: MIDI Visualizer

https://decompiled.dev/apps/fractured-jukebox/
2•decompiled_dev•20m ago•0 comments

I reverse engineered FORM swim goggles to push workouts over BLE

https://www.reachflowstate.ai/blog/form-goggles-reverse-engineering
4•garrickhgan•20m ago•1 comments

Senna-Driven 1986 Lotus F1 Car in JPS Livery Is Perfection for Roughly $10M

https://www.thedrive.com/news/senna-driven-1986-lotus-f1-car-in-jps-livery-is-perfection-for-roug...
2•PaulHoule•20m ago•0 comments

Attractor: Build your own software factory agents

https://github.com/strongdm/attractor
2•kentf•21m ago•0 comments
Open in hackernews

Microsoft BitNet: 100B Param 1-Bit model for local CPUs

https://github.com/microsoft/BitNet
96•redm•1h ago

Comments

QuadmasterXLII•1h ago
headline hundred billion parameter, none of the official models are over 10 billion parameters. Curious.
Tuna-Fish•1h ago
The project is an inference framework which should support 100B parameter model at 5-7tok/s on CPU. No one has quantized a 100B parameter model to 1 trit, but this existing is an incentive for someone to do so.
152334H•1h ago
but there is no trained 100b param model? "can run a 100B BitNet" is about the inference implementation, not about the existence of any such model
syntaxing•1h ago
Misleading title but this is pretty exciting. Interesting how this is based on llama cpp. Its nice to see some momentum since they released the paper in 2023
radarsat1•1h ago
I'm curious if 1-bit params can be compared to 4- or 8-bit params. I imagine that 100B is equivalent to something like a 30B model? I guess only evals can say. Still, being able to run a 30B model at good speed on a CPU would be amazing.
LuxBennu•1h ago
The title is misleading — there's no trained 100B model, just an inference framework that claims to handle one. But the engineering is worth paying attention to. I run quantized 70B models locally (M2 Max 96GB, llama.cpp + LiteLLM), and memory bandwidth is always the bottleneck. The 1.58-bit approach is interesting because ternary weights turn matmuls into additions — a fundamentally different compute profile on commodity CPUs. If 5-7 tok/s on a single CPU for 100B-class models is reproducible, that's a real milestone for on-device inference. Framework is ready. Now we need someone to actually train the model.
rustyhancock•59m ago
Yes. I had to read it over twice, it does strike me as odd that there wasn't a base model to work with.

But it seems the biggest model available is 10B? Somewhat unusual and does make me wonder just how challenging it will be to train any model in the 100B order of magnitude.

wongarsu•44m ago
Approximately as challenging as training a regular 100B model from scratch. Maybe a bit more challenging because there's less experience with it

The key insight of the BitNet paper was that using their custom BitLinear layer instead of normal Linear layers (as well as some more training and architecture changes) lead to much, much better results than quantizing an existing model down to 1.58 bits. So you end up making a full training run in bf16 precision using the specially adapted model architecture

wongarsu•55m ago
I've also always though that it's an interesting opportunity for custom hardware. Two bit addition is incredibly cheap in hardware, especially compared to anything involving floating point. You could make huge vector instructions on the cheap, then connect it to the fastest memory you can buy, and you have a capable inference chip.

You'd still need full GPUs for training, but for inference the hardware would be orders of magnitude simpler than what Nvidia is making

regularfry•44m ago
You only need GPUs if you assume the training is gradient descent. GAs or anything else that can handle nonlinearities would be fine, and possibly fast enough to be interesting.
embedding-shape•53m ago
> Framework is ready. Now we need someone to actually train the model.

If Microslop aren't gonna train the model themselves to prove their own thesis, why would others? They've had 2 years (I think?) to prove BitNet in at least some way, are you really saying they haven't tried so far?

Personally that makes it slightly worrisome to just take what they say at face value, why wouldn't they train and publish a model themselves if this actually led to worthwhile results?

gregman1•43m ago
Cannot agree more!
throwaw12•30m ago
Because this is Microsoft, experimenting and failing is not encouraged, taking less risky bets and getting promoted is. Also no customer asked them to have 1-bit model, hence PM didn't prioritize it.

But it doesn't mean, idea is worthless.

You could have said same about Transformers, Google released it, but didn't move forward, turns out it was a great idea.

embedding-shape•23m ago
> You could have said same about Transformers, Google released it, but didn't move forward,

I don't think you can, Google looked at the research results, and continued researching Transformers and related technologies, because they saw the value for it particularly in translations. It's part of the original paper, what direction to take, give it a read, it's relatively approachable for being a machine learning paper :)

Sure, it took OpenAI to make it into an "assistant" that answered questions, but it's not like Google was completely sleeping on the Transformer, they just had other research directions to go into first.

> But it doesn't mean, idea is worthless.

I agree, they aren't, hope that wasn't what my message read as :) But, ideas that don't actually pan out in reality are slightly less useful than ideas that do pan out once put to practice. Root commentator seems to try to say "This is a great idea, it's all ready, only missing piece is for someone to do the training and it'll pan out!" which I'm a bit skeptical about, since it's been two years since they introduced the idea.

wongarsu•8m ago
On the one hand, not publishing any new models for an architecture in almost a year seems like forever given how things are moving right now. On the other hand I don't think that's very conclusive on whether they've given up on it or have other higher priority research directions to go into first either
GorbachevyChase•20m ago
The most benign answer would be that they don’t want to further support an emerging competitor to OpenAI, which they have significant business ties to. I think the more likely answer which you hinted at is that the utility of the model falls apart as scale increases. They see the approach as a dead end so they are throwing the scraps out to the stray dogs.
cubefox•53m ago
LLM account
orbital-decay•47m ago
Funny enough I now involuntarily take RTFA as a slight slop signal, because all these accounts dutifully read the article before commenting, unlike most HNers who often respond to headlines.
yorwba•43m ago
Not all of them do: https://news.ycombinator.com/item?id=47335156 There are evidently lots of people experimenting with different botting setups. Some do better at blending in than others.
PeterHolzwarth•25m ago
Interesting - the account you mention, and the GP, are both doing replies that are themselves all about the same length, and also the same length between the two accounts. I get what you mean.
cubefox•38m ago
Yeah. It correctly pointed out that the editorialized HN title is wrong, there is no 100B model.
vova_hn2•33m ago
First they claimed that if you use em dashes you are not human

And I did not speak out

Because I was not using em dashes

Then they claimed that if you're crammar is to gud you r not hmuan

And I did not spek aut

Because mi gramar sukcs

Then they claimed that if you actually read the article that you are trying to discuss you are not human...

K0balt•14m ago
I’ve been rounded up for things I wrote two decades ago because of my em dashes lol. The pitchfork mentality gives me little hope for how things are going to go once we have hive mind AGI robots pervasive in society.
hrmtst93837•30m ago
I browsed through the history of the user and confirm this statement. I know that there are users who say they used em-dashes even before the rise of ChatGPT and HN statistics support that. For example, one prominent example is dang.

However this user uses — in almost all his posts and he had a speed of 1 comment per minute or so on multiple different topics.

Springtime•30m ago
Hmm, the user joined in 2019 but had no submissions or comments until just 40 minutes ago (at least judging by the lack of a second page?) and all the comments are on AI related submissions. Benefit of doubt is it'd have to be a very dedicated lurker or dormant account they remembered they had.

Edit: oh, just recalled dang restricted Show HNs the other day to only non-new users (possibly with some other thresholds). I wonder if word got out and some are filling accounts with activity.

nkohari•12m ago
I would love to understand the thought process behind this. I'm sure it's a fun experiment, to see if it's possible and so on... but what tangible benefit could there be to burning tokens to spam comments on every post?
butILoveLife•51m ago
>. I run quantized 70B models locally (M2 Max 96GB, llama.cpp + LiteLLM), and memory bandwidth is always the bottleneck.

I imagine you got 96gb because you thought you'd be running models locally? Did you not know the phrase Unified Memory is marketing speak?

WithinReason•45m ago
> a fundamentally different compute profile on commodity CPU

In what way? On modern processors, a Fused Multiply-Add (FMA) instruction generally has the exact same execution throughput as a basic addition instruction

actionfromafar•32m ago
Bitnet encoding more information dense per byte perhaps? CPUs have slow buses so would eke out more use of bandwidth?
august11•41m ago
In their demo they're running 3B model.
nickcw•1h ago
> bitnet.cpp is the official inference framework for 1-bit LLMs (e.g., BitNet b1.58). It offers a suite of optimized kernels, that support fast and lossless inference of 1.58-bit models on CPU and GPU (NPU support will coming next).

One bit or one trit? I am confused!

drsopp•1h ago
"1-bit LLMs" is just marketing. The Shannon entropy of one letter with a 3 symbol alphabet (-1, 0, 1) is 1.58.
Dwedit•49m ago
Log Base 2 of 3 = ~1.5849625, so that's the limit to how well you can pack three-state values into bits of data.

For something more practical, you can pack five three-state values within a byte because 3^5 = 243, which is smaller than 256. To unpack, you divide and modulo by 3 five separate times. This encodes data in bytes at 1.6 bits per symbol.

But the packing of 5 symbols into a byte was not done here. Instead, they packed 4 symbols into a byte to reduce computational complexity (no unpacking needed)

rasz•32m ago
>1-bit model

>packed 4 symbols into a byte

microslop, typical bunch of two-bit frauds!

cubefox•58m ago
Yeah, "1.58 bit" is 1 trit with three states, since log2(3)≈1.58.

So it's not a inference framework for 1-bit models (two states per parameter) but for 1.58 bit models (three states per parameter). Annoying that they try to mix up the two.

silon42•7m ago
I always hope for "just a bunch of if statements" ... this is not it.
itsthecourier•1h ago
https://github-production-user-asset-6210df.s3.amazonaws.com...

demo shows a huge love for water, this AI knows its home

_fw•59m ago
Also, very influenced by the literature of Jenkins (2010).
giancarlostoro•56m ago
One of the things I often wonder is "what will be the minimally viable LLM" that can work from just enough information that if it googles the rest it can provide reasonable answers? I'm surprised something like Encyclopedia Britanica hasn't yet (afaik) tried to capitalize on AI by selling their data to LLMs and validating outputs for LLM companies, it would make a night and day difference in some areas I would think. Wikipedia is nice, but there's so much room for human error and bias there.
embedding-shape•52m ago
Your worry about Wikipedia is that there is "much room for human error and bias", yet earlier you seem to imply that a LLM that has access to the www somehow would have less human error and bias? Personally, I'd see it the other way around.
giancarlostoro•5m ago
When GPT 3.5 became a thing, it had crawled a very nuanced set of websites, this is what I mean. You basically curate where it sources data from.
utopiah•52m ago
> validating outputs for LLM companies

How? They can validate thousands if not millions of queries but nothing prevent the millions-th-and-one from being a hallucination. People who would then pay extra for a "Encyclopedia Britanica validated LLM" would then, rightfully so IMHO, complain that "it" suggested them to cook with a dangerous mushroom.

uniq7•47m ago
Since Google Search already includes an AI summary, your minimally viable "LLM" can be just an HTTP GET call
intrasight•46m ago
It's not so much a "minimally viable LLM" but rather an LLM that knows natural language well but knows nothing else. Like me - as an engineer who knows how to troubleshoot in general but doesn't know about a specific device like my furnace (recent example).

And I don't think that LLM could just Google or check Wikipedia.

But I do agree that this architecture makes a lot of sense. I assume it will become the norm to use such edge LLMs.

thinkingtoilet•31m ago
Wikipedia has proven to be as accurate as encyclopedias for decades now. Also, I'm betting AI companies have illegally trained their models on the Encyclopedia Britanica's data by now.
bee_rider•18m ago
Isn’t that sort of what a RAG is? You’d need an LLM “smart” enough to turn natural-user prompts into searches, then some kind of search, then an LLM “smart” though to summarize the results.
giancarlostoro•6m ago
Yeah, I think RAG is the idea that will lead us there, though its a little complicated, because for some subjects, say Computer Science, you need a little more than just "This is Hello World in Go" you might need to understand not just Go syntax on the fly, but more CS nuances that are not covered in one single simple document. The idea being having a model that runs fully locally on a phone or laptop with minimal resources. On the other hand, I can also see smaller models talking to larger models that are cheaper to run in the cloud. I am wondering if this is the approach Apple might take with Siri, specifically in order to retain user privacy as much as possible.
Arcuru•40m ago
It's good to see this getting some continued development. I looked into it last year[1] and I thought it showed a lot of promise so I've been very disappointed that I never saw a newer model.

[1] - https://jackson.dev/post/dont-sleep-on-bitnet/

algoth1•35m ago
Headline: 100B. Falcon 3 family: 10B. An order of magnitude off
bee_rider•31m ago
What’s the lower limit on the number of bits per parameter? If you use CSR-style sparse matrices to store the weights can it be less than 1?
simonw•29m ago
Anyone know how hard it would be to create a 1-bit variant of one of the recent Qwen 3.5 models?
nikhizzle•24m ago
Almost trivial using open source tools, the question is how it performs without calibration/fine tuning.
wongarsu•14m ago
The results would probably be underwhelming. The bitnet paper doesn't give great baselines to compare to, but in their tests a 2B network trained for 1.58bits using their architecture was better than Llama 3 8B quantized to 1.58bits. Though that 2B network was about on par with a 1.5B qwen2.5.

If you have an existing network, making an int4 quant is the better tradeoff. 1.58b quants only become interesting when you train the model specifically for it

On the other hand maybe it works much better than expected because llama3 is just a terrible baseline

philvas•21m ago
steve jobs would have loved the microsoft repo with demo on mac
rarisma•18m ago
No 100b model.

My disappointment is immeasurable and my day is ruined.

devnotes77•12m ago
The compute throughput question (whether FMA equals ADD on modern CPUs) is accurate — that's not where the gain is. The real win is memory footprint.

A 100B ternary model packs to roughly 20-25GB (100B params at ~1.58 bits each). FP16 would be ~200GB, INT4 ~50GB. That difference is what moves the "doesn't fit" threshold. You go from needing HBM or multi-GPU NVLink to running on a workstation with 32GB DDR5.

DDR5 at ~100 GB/s is still much slower than HBM at ~3 TB/s, so memory bandwidth is still the inference bottleneck — but bandwidth is only a problem once the model actually fits. For 100B-class models, capacity was the harder constraint. That's what 1.58-bit actually solves.

WhitneyLand•10m ago
If they had a big result like, native 1.58-bit quality clearly matches top peers, they would be saying that prominently in the repo.

The engineering/optimization work is nice, but this is not what people have been waiting for, as much as, can’t the Bitnet idea that seemed promise really deliver in a competitive way.