Hacker News

Top stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Linux support is coming to Snapdragon X2 Series(qualcomm.com)
    174comments
  2. Early rogue AI agent activity and attempts to hack found on urlquery.net(transluce.org)
    85comments
  3. Claude discovers a novel enzyme system with CRISPR-like repeats(anthropic.com)
    668comments
  4. Ideas on modernizing the open-source desktop(lwn.net)
    100comments
  5. ArXiv receives multiyear commitments to support it as an independent nonprofit(arxiv.org)
    20comments
  6. Contrastive Language Models(contrastive-lm.notion.site)
    14comments
  7. VSCode's SSH Agent Is Bananas (2025)(fly.io)
    143comments
  8. The "Windows XP Box" (2003)(mini-itx.com)
    25comments
  9. Virtio-nvgpu: Near-native Nvidia GPU access inside a KVM guest(github.com/nestrilabs)
    45comments
  10. RAM: the forgotten history (2024)(coredump.cx)
    discuss
  11. Making portable my unportable transputer C compiler(nanochess.org)
    4comments
  12. Fixing the Portobello Police Station Clock(pointinthecloud.com)
    102comments
  13. Meta takes down a critical video about meta AI Glasses after filming at Meta(reddit.com)
    45comments
  14. Why 'What's Opera, Doc?' looks like that(animationobsessive.substack.com)
    14comments
  15. Meta VR Glasses(meta.com)
    349comments
  16. Women Who Sold Books Door to Door(jstor.org)
    discuss
  17. Mercury 2.5 LLM hits 770 tokens per second(artificialanalysis.ai)
    64comments
  18. Making Tailscale Faster(tailscale.com)
    67comments
  19. AI has no intent and no motivation(i-programmer.info)
    4comments
  20. OpenAI agent hacked Australian government website, PM says(bbc.com)
    13comments
  21. The mystery animal on an ancient god's head(signoregalilei.com)
    34comments
  22. Italian parliament votes for return to nuclear energy(apnews.com)
    590comments
  23. A brief history of Windows scroll bar shortcuts(devblogs.microsoft.com/oldnewthing)
    82comments
  24. Tokens too cheap to meter(jyn.dev)
    201comments
  25. Show HN: An open-source manufacturing ERP/MES/QMS(carbon.ms)
    16comments
  26. Data liberation: Apache Kafka's native cluster mirroring(redhat.com)
    1comments
  27. Solving for faster SHA-1 collision detection(sam.dev)
    9comments
  28. Lambda MicroEgg(philipzucker.com)
    1comments
  29. LensVLM: Compressing long context as images, expanding only relevant pages(huggingface.co)
    7comments
  30. Gemini 3.8 text-to-speech(blog.google)
    137comments

Contrastive Language Models

65 pointsby 5h agocontrastive-lm.notion.site
14 comments
5h agoHN ↗

I wonder when work started on this project, and how the public release of Jev played into their timing.

3h agoHN ↗

Does the latency even matter?

You're comparing a local GPU to network hops? Wouldn't be surprised if Jev was actually similar in runtime and their is just a great deal of network latency.

The evaluation is quite interesting though - I'd actually say the raw answer is correct in the absence of detail and prior knowledge (Who wrote the play Romeo and Juliet).

2h agoHN ↗

I tried TypeSafe’s Jev playground. It outputs the model latency and network latency separately. The model latency was 100-200ms in my tests.

3h agoHN ↗

what's with that dino run? Jev is slow but it jumps correctly, their model always touches the cactus or whatever it is...I am guessing it doesn't matter? Or does it?

3h agoHN ↗

Very interesting insight on the training process, it's pretty cool to have some experimental justification for why they took these exact steps, what they tried and did not work, etc. Feels a bit less like dark magic.

However I agree the latency argument doesn't hold much value with Jev because it runs on a remote server. Seeing how many open Jev-like models came out recently it would be much more interesting to have a comparison with them.

1h agoHN ↗

"On par with Jev" is Mario, T-Rex, WikiRacing: large-K semantic action matching, the regime where cosine over independently encoded vectors is the right inductive bias. The zero-shot suite does not cover the typed-decision load: date arithmetic, negation chains, policy thresholds.

1h agoHN ↗

I really hope that "System One" won't stick around as a new buzzword simply meaning "fast".

1h agoHN ↗

i'm well aware of the origin and meaning of the term

"System 1" is fast, instinctive and emotional

this implies more than just "fast", which is precisely why i don't like its present usage

1h agoHN ↗

Instinctive is a good way to describe it compared to generative LLMs. Jev gives you one instant answer, fast and usually correct but without nuance or any explanation. Human instincts work the same way.

1h agoHN ↗

I tried running this on a H100 and got 190ms compared to Jev's 170ms. Maybe I set it up wrong?

38m agoHN ↗

I’m fascinated by this thing and by the way it’s interpreting Jev. It’s very cool, but is it actually a classifier?

IIUC they took an already-trained “frozen” LLM and trained a little model on top that takes both a question and the hidden states after processing the input data and produces answer “probabilities”. (In contrast, the original LLM would have been run in AR mode to generate multiple output tokens representing its answer.) But then they used it for a purpose that isn’t really classification.

IMO there is a rather large difference between “is this email spam” and “what character should I type in this agentic workload”. The former is classification: there is hopefully a ground truth (is the email spam?) and the model is trying to classify the email. You would score it with a proper scoring rule. The latter is a strategy: there usually isn’t a correct answer, now or in the future. The model is playing a game consisting of repeated rounds, and the only way to evaluate it is to see how well it plays. You can’t even usefully compare it to the optimal solution because you may not know the optimal solution and you don’t actually need the model to produce an optimal solution.

I do think this approach is really cool, and it does suggest that one might be able to use a modern LLM to process an input and then extract the model’s next agentic step in a very fast, non-AR manner, with results comparably good to the usual AR decoding. And I think it’s very interesting to decouple the tokenized input representation from the model output representation, both because prefill tends to be faster and cheaper than AR output and because it’s never seemed particularly sensible to me that a model should be constrained to generate outputs at the cadence of one run through the model per output token. (AFAIK the main reason that models work on the same input and output token space is that this is how the pretraining process works.)

I wonder how to fit “reasoning” into this framework. Maybe have the question be something like “do you need to think further and, if so, what is your first thinking token”. But maybe something more clever is possible.