HNHacker News
TopNewBestAskShowJobs

0-_-0

4,368 karma · joined October 10, 2018

submissionscomments
0-_-0··on Zuckerberg says AI agent development going slower than expected
Even if you had all that you would get completely different weights out at the end, and you also don't have the resources to "compile" an LLM because the compilation can cost $100M. If you were given the training data but not the weights, would you consider that open source?
0-_-0··on Zuckerberg says AI agent development going slower than expected
Source code makes it easier to modify a binary, training data doesn't make it easier to modify a trained network.
0-_-0··on “Beyond the limit”: Satellites and mirrors in space pose threat to the night sky
If we can launch 1M satellites, how many telescopes can we launch?
0-_-0··on “Beyond the limit”: Satellites and mirrors in space pose threat to the night sky
And what are they reflecting when they are in Earth's shadow?
0-_-0··on TurboQuant can reduce vector index size by 10x at 100M Row Scale
32 bits vs 4 bits it looks like
0-_-0··on Opening up 'Zero-Knowledge Proof' technology to promote privacy in age assurance
Browser fingerprinting can already pinpoint you exactly. We should focus on that.
0-_-0··on DeepSeek V4 Peak Valley Pricing Change
That math makes no sense
0-_-0··on The UK's new under-16 social media ban will cause more harm than it prevents
These never ever accomplish anything
0-_-0··on Swiss parliament lifts ban on new nuclear power plants
Not if it's mass produced
0-_-0··on Mercedes‑Benz starts large‑scale production of electric axial flux motor
No, it means the motor is smaller and it can be put into the wheel
0-_-0··on How much of Thermo Fisher's antibody data has been manipulated?
And this is just fraud that was done with incompetence, so easily caught. How much is done competently?
0-_-0··on Speculative KV coding: losslessly compressing KV cache by up to ~4×
This can't be used to save VRAM in practice. To generate a new token with the primary model, you first need to decompress the cache, which involves regenerating the whole sequence from scratch. I.e. generate 1 million tokens with the small model to generate 1 with the large.
0-_-0··on Speculative KV coding: losslessly compressing KV cache by up to ~4×
You can use the original model to compress the kv cache and get ∞x compression, since the prediction is perfect. The cost is time, and I don't see how this could be worth it.
0-_-0··on The Website Specification
Agents don't buy stuff they see in an ad
0-_-0··on Orthrus-Qwen3: up to 7.8×tokens/forward on Qwen3, identical output distribution
I thought it can do multiple tokens at a time
0-_-0··on Orthrus-Qwen3: up to 7.8×tokens/forward on Qwen3, identical output distribution
3.6 already supports multi token generation AFAIK
0-_-0··on EU Parliamentary Research Service calls VPNs "a loophole that needs closing"
How about next year?
0-_-0··on Xbox CEO ends Copilot AI development and overhauls leadership
What else would it be?
0-_-0··on 245TB Micron 6600 ION Data Center SSD Now Shipping
I don't think he wants to buy one
0-_-0··on Biscuit
I wonder why that's not supported by Biscuit despite dozens of other functions being present. E.g. download and display an image once an hour. Can't wake from sleep regularly?
0-_-0··on LLMs Are Not a Higher Level of Abstraction
You're being downvoted, but you're right. Determinism is a different concept and doesn't characterise LLMs well. You can have deterministic random number generators for example.
0-_-0··on When Dawkins met Claude – Could this AI be conscious?
https://en.wikipedia.org/?title=Emergent_behavior
0-_-0··on Why does it take so long to release black fan versions?
Some of my friends are beige
0-_-0··on Noctua releases official 3D CAD models for its cooling fans
Obscurity makes no sense on a world with patents.
0-_-0··on I benchmarked Claude Code's caveman plugin against "be brief."
How about caveman+be brief?
0-_-0··on The biggest insect ever was a "dragonfly"
I have no idea how much that is so it doesn't terrify me, but a 47 cm dragonfly is terrifying!
0-_-0··on The biggest insect ever was a "dragonfly"
There was a recent HN post where they checked the oxygen hypothesis (insects grew this large due to higher atmospheric oxygen levels) and it didn't hold up
0-_-0··on Microsoft and OpenAI end their exclusive and revenue-sharing deal
What I hear is that getting your network to work on AMD is a huge pain.
0-_-0··on Show HN: AI memory with biological decay (52% recall)
It's the cumulative weighting based on the softmax output? Is it per layer?
0-_-0··on DeepSeek v4
Training algorithm is the programmer, weights are the code that you run in an interpreter
← PreviousPage 2 of 34Next →