Hacker News

Top stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Jeff – Jev-compatible 0.8B decision models, trained at home, ~30 ms (github.com/firelex)
    47comments
  2. Who should be held accountable when an AI Agent (accidentally) acts maliciously? (greenpants.net)
    32comments
  3. Pirating the Pirates (mubi.com)
    191comments
  4. 12,000-year-old Göbeklitepe burials explain scattered bones (archaeologymag.com)
    8comments
  5. MicroLLM Lab – Try 7 tiny LLM's in the browser (stateofutopia.com)
    43comments
  6. World Labs Is Joining AMD (worldlabs.ai)
    46comments
  7. Scientists solve 1840s space weather mystery (arstechnica.com)
    21comments
  8. Hijacking the PS5's RTMP stream (yashgarg.dev)
    54comments
  9. Joseph Szabo’s pictures of American adolescents (newyorker.com)
    35comments
  10. Kids turned low-traffic NPR Spotify comments into a secret group chat (thisamericanlife.org)
    146comments
  11. Parley: Federated, decentralised chat that speaks plain IRC (mills.io)
    147comments
  12. It's Time to Investigate the AI Labs (calnewport.com)
    52comments
  13. Best of British Design (best-of-british-design.vercel.app)
    23comments
  14. What reversing, modernising old games tells us about the economic impact of AI (isfine.org)
    8comments
  15. Flock Wants the Most Detailed Map of Its Surveillance Cameras Taken Offline (theintercept.com)
    42comments
  16. Sonnet 5.5 (anthropic.com)
    343comments
  17. Cf: The Agentic CLI for the Cloudflare API (cloudflare.com)
    42comments
  18. 3D necroprinting: Leveraging biotic material as the nozzle for 3D printing (science.org)
    —discuss
  19. Nvidia wants to put a watchdog chip next to every AI agent (cnbc.com)
    127comments
  20. Show HN: Destroy Any Website with Stickman (spritefusion.com)
    25comments
  21. First Steps of the PLC Organization – Independent Public Ledger of Credentials (plcred.org)
    18comments
  22. Launch HN: Vespper (YC F24) – SOTA Docx MCP (vespper.com)
    8comments
  23. Does Reddit have an astroturfing problem? What the data suggests (petervijeh.com)
    91comments
  24. Who wrote Elizabeth I's most scathing letters? (smithsonianmag.com)
    18comments
  25. What heraldry and Japanese mon can teach about visual-identity generators (benovermyer.com)
    20comments
  26. Updated Google Maps shows destruction of the city of Rafah (twitter.com/aliabunimah)
    35comments
  27. MongoDB CEO resigns to join Meta (reuters.com)
    252comments
  28. Pacing the Frontier is not the actual goal for AI labs (lesswrong.com)
    57comments
  29. GrapheneOS – When an app is slow (wirelessmoves.com)
    41comments
  30. When did Google get so weird? (sancho.bearblog.dev)
    1021comments

MicroLLM Lab – Try 7 tiny LLM's in the browser

101 pointsby 4h agostateofutopia.com
43 comments
4h agoHN ↗

You can try 7 different tiny LLM's in your browser.

3h agoHN ↗

What I missed from the title was this: Can I try 7 different tiny LLMs in my browser?

3h agoHN ↗

i have a bromine tub, give me simple instructions for what to do since i just filled it up w fresh water

1. Add 1 tablespoon of water to the water bath. 2. Place the tub into the water bath and let it sit for about 5 minutes. 3. After 5 minutes, remove the tub and let it cool down. 4. Now, fill the tub with water and let it sit for about

uhmm completely unusable ?

3h agoHN ↗

No, that is not what SLMs are useful for. They have very little knowledge, and typically lack reasoning skills.

Useful applications include sentiment analysis, text classification, entity extraction, etc.

They certainly can be useful, but you shouldn't compare them with LLMs such as Opus or Fable.

2h agoHN ↗

Completely unusable for that, yeah, that's a 100M model, not an assistant

3h agoHN ↗

The website literally says they are SLM's .. derp af

3h agoHN ↗

I did the default arithmetic with PetitGPT research-v1

What is 2+2?

Answer

To find 2 + 2, we need to add 2 to both sides of the equation.

2 + 2 = 4

So, 2 + 2 = 4 + 2.

Brilliant

1h agoHN ↗

That's not incorrect.

LLMs produce semantically correct sentences, not factually correct statements. Have we forgotten this so soon?

1h agoHN ↗

It is, in every sense, incorrect. Which statement in this short snippet is "semantically correct"? (Better LLMs get this right, of course.)

45m agoHN ↗

It's not very good semantically either.

Give me a recipe for soup.

Here is a recipe for soup:

Saffa-Cake-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-Sweet-

1h agoHN ↗

GPT-2 124M is terrible...

  > what is 2+2?

Answer:

  > 3+3? 4+4? 5+6? 7+8?
  > Reply ~18000 0 ~10 min 2 By : 1-1: I'm a beginner. 3x2 is my best option, but if you're not sure about the other options then just go for it and try again
1h agoHN ↗

It's not instruct tuned looks like? It's closer to a base model rather than a chatbot.

1h agoHN ↗

That one is a 2019 model :) Years before the ChatGPT public preview.

1h agoHN ↗

GPT-2 is an interesting one because it is a February 2019 model. (You can see some information about it below the card if you click on the card.)

That was 2-3 years before the big "ChatGPT moment" (the highly coherent ChatGPT research preview was released in November 2022, I think it was ChatGPT 3.5). Back in 2019 the models really were not producing very coherent output. Now you can see it for yourself right in your browser :) Everything has come a really long way since then!

33m agoHN ↗

I've been playing with it for a bit, and I'm a tiny little bit surprised that they decided to continue pursuing that direction of research at all. What I'm getting from it looks like it could just be arbitrary sentence and paragraph fragments from the Internet, pasted together Markov-chain like.

I'm not sure I would have ever believed that something useful would come out of it, yet here we are.

44m agoHN ↗

Awesome! I tried it and after loading (which took a while as it is a large model) got 12 tokens/second and very coherent output. Great demonstration.

8m agoHN ↗

I would recommend adding the MiniCPM5-1B model. Surprisingly coherent for a 1B model and performs well. On my pixel 9 I get 33 tok/s on CPU. On GPU I get about 26 tok/s but prefill jumps to nearly 500.

2h agoHN ↗

Cool project, but I'd really suggest looking at the UI.

The text is too small and it's way too dense with information in general. Considering how simple this product is to use, it's kinda crazy that I have to scroll through over a page length of (mostly useless, AI-generated) information before getting to the actual interface.

Also what is going on with the footer (why does it link back to the site itself, why is it telling me to "serve over HTTP").

2h agoHN ↗

This is the future of software, sloppy ui.

1h agoHN ↗

Thank you for the feedback, I'll think about how to incorporate the changes you've suggested.

Edit: I've concluded that the text is tiny so people can skip it, it's like fine print. There are only 600 words on the entire page, that's not a lot even if someone does read all of them.

Since this submission has been on the front page of HN for hours now, clearly people like it as it is. I don't think I need to change anything about it.

The footer means that if you download the zip file and unzip it, you can't just open index.html and expect it to work, you need to start a local webserver for it.

2h agoHN ↗

PetitGPT told me that

"2+2 is 2."

Otherwise, a very neat demo. As others have said, the UI is VERY confusing, way too much stuff going on.

2h agoHN ↗

Really cool project. Giving web apps direct access to on-device models is something I’m excited about, and it’s cool to see the different approaches.

I’ve been working on a related proposal called the Web Models API, which explores a browser standard for an API that runs open-weight models on-device. Would love your thoughts: https://www.webmodels.dev

2h agoHN ↗

I read your proposal, I think it's great! Where will the navigator get the model if the user agrees to download it? For this demonstration I just serve the models on my own server, but for larger models it may be an issue as they may not have direct download links even if they are open weights.

2h agoHN ↗

Great question. Right now there isn’t a definitive answer, but it’s something that needs to be worked out. There would likely be a registry. The question is how to keep model IDs consistent: does each browser manage its own registry, or is there one shared across browsers?

1h agoHN ↗

Since you're asking for some kinds of permissions anyway, you could ask if the user is willing to also seed the model, p2p. (However, seeding files is not as popular as it used to be, many residential Internet connections don't have good upload.) If you have the capacity for it, your site webmodels.dev could act as a tracker and initial seed for any models. Then it could be the one central registry. It might get to be too much for you though, a lot of the open weights models are huge.

1h agoHN ↗

Interesting idea. I hadn't considered that. Thanks

2h agoHN ↗

unfortunately on firefox: Uncaught ReferenceError: GPUShaderStage is not defined

1h agoHN ↗

I tested it on Firefox on windows, version 156.0.1 and didn't get that error.

What version of Firefox are you using and what is your operating system and graphics card, please? Can you also try it without WebGPU? (Reload the page and uncheck "Prefer WebGPU" and try a prompt.)

37m agoHN ↗

It's a cool project, but I couldn't get it to load. What did you test this on? I tried the live link here:

https://willaaam.github.io/gemma-4-E2B-webgpu-vision/

And after loading it, with an NVidia 1060 GPU (6 GB RAM) on Windows it failed with "Failed to load: No supported WebGPU variant for com.xenova.gemma4.DenseGemv; rejected sgma".

In Safari on a 2026 Mac Mini M4 with 24 GB of RAM it failed with "Failed to load: JSON Parse error: Unexpected EOF".

The idea is pretty cool though!

1h agoHN ↗

What browser are you using? I tested it on Windows, Mac, and iPhone. I tested it in Chrome, Firefox, Edge, and Safari. Everything works on the three machines and phone I tested it on.

(It's a little bit slow at the moment - you have to wait a few seconds for the models to load - as it's currently on the HN front page. The server is on a 1 gigabit unmetered network connection so it can serve all the weights - around 600 megabytes - to one person every few seconds, there are several concurrent users now.)

1h agoHN ↗

how would you compare your capabilities to that of claude fable 5.1 by anthropic

Comparing your capabilities to that of claude fable 5.1 would be very similar. Both are stories about a clown who is a cunning and manipulative animal. They both have problems and challenges, and they face challenges to overcome them. However, they have different goals and approaches. The clown

I think that tells you all you need to know.

30m agoHN ↗

Thanks. Unfortunately it's too late to change the title!