Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. How (and why) to be an effective meat proxy (vivekhaldar.com)
    —discuss
  2. My Six Favorite YouTube IDs (greg.technology)
    —discuss
  3. X Files – Ghost in the Machine (imdb.com)
    —discuss
  4. Almost nobody gets XML sitemaps right (joost.blog)
    —discuss
  5. Read My Damn Paper (wilsoniumite.com)
    2comments
  6. Ex-CIA officer's novel alleges grave dysfunction in deep-cover operations (washingtonpost.com)
    —discuss
  7. Duke Builder: An experimental level editor for Duke Nukem 3D (github.com/aheck)
    1comments
  8. Screw Is Needed for Missiles and Planes. Buying One Can Take Years (nytimes.com)
    1comments
  9. Generate realistic Postgres test data from DDL, without a database (weavori.com)
    —discuss
  10. Why Do Clinical Trials for Cancer Drugs Take So Long? (nytimes.com)
    —discuss
  11. Gitman: Easy config-file based multi repo management tool (github.com/sascha10000)
    1comments
  12. Opus 5.5 remade FL Studio in Rust in about 1 hour (twitter.com/skewbed)
    —discuss
  13. Joint supplement linked to faster Alzheimer's progression (sciencedaily.com)
    —discuss
  14. Show HN: A New Way to Host Design Challenges (eightpotions.com)
    1comments
  15. How the Smithsonian became the latest front in Trump's culture war (propublica.org)
    —discuss
  16. Ask HN: What am I missing with Siri AI?
    1comments
  17. Show HN: iPulse AI – Visitors can inspect disagreement across debating agents (ipulseai.com)
    —discuss
  18. A visual notebook where sound, art and geometry connect (n1m21n.github.io)
    —discuss
  19. Earth chorus waves show limited link to rapid electron loss from radiation belts (phys.org)
    —discuss
  20. Why Are So Many Adults Cutting Off Their Parents? (wsj.com)
    1comments
  21. I hate that I like this song (reddit.com)
    —discuss
  22. Mallard, fastest steam engine train interactive painting experience (echohive.ai)
    —discuss
  23. A classifier is not a classifier is not a classifier (sutro.sh)
    —discuss
  24. Oral History of John Chowning (Computer History) [video] (youtube.com)
    —discuss
  25. Google OpenAI Anthropic Begin Forming SAFA – Standards Authority for Frontier AI (proactiveinvestors.com)
    1comments
  26. I don't read code anymore (duyet.net)
    4comments
  27. Kids turned the comment section of an NPR podcast into a group chat (thisamericanlife.org)
    —discuss
  28. The NSA Is Spending Billions to Test AI Models, Classified Estimates Show (gadgetreview.com)
    —discuss
  29. GPU Glossary (Readable) (modal.com)
    —discuss
  30. Flock (YC 2017) Wants Detailed Map of Its Surveillance Cameras Taken Offline (theintercept.com)
    2comments

Ember-1

48 pointsby 56m agofireworks.ai
11 comments
9m agoHN ↗

The problem: thinking models think too much

Analysis paralysis stifles not just human intelligence, but other intelligences too.

9m agoHN ↗

I don't think the article mentions Pareto frontier enough.

Also, did I miss a memo? Suddenly every article on AI seems to be talking about the Pareto frontier - or have I just not been paying attention?

8m agoHN ↗

Need this done for DeepSeek, ideally one of the Flash models.

4m agoHN ↗

If you have the compute, I have the expertise.

7m agoHN ↗

Well done, and great iteration.

The pareto frontier needs clearer distinction. Benchmarks miss half the story. What, if any, capability is lost by the token reduction (for example, was it like super awesome at Golang before and now kind of sucks? that kind of distinction).

7m agoHN ↗

It looks like it would be similar to GLM 5.3 Flash, had they tested it...

7m agoHN ↗

Ignoring for the moment issues of what "counts" as open, won't open models rapidly advance due to stuff like this in ways that it's less possible for the proprietary ones to do? This is exactly how Linux & Wikipedia, for example, overtook their "frontiers", right?

3m agoHN ↗

Yes, absolutely, but only if people keep contributing in the open.

6m agoHN ↗

On the smaller end, Quen 3.8, while being extraordinarily capable for a small local model, also suffers from extreme thinking. I wonder if the techniques described here generalize to other models too.

4m agoHN ↗

Does anybody know if this would be a good model for creative writing?

3m agoHN ↗

So they trained a model on open weights, and then aren't releasing the weights... am I reading this right?