tveita
Joined 14 years ago
Karma
2,788
Submissions
884
Member since
May 5
Recent submissions
Comment · 2 days ago
But the report OP cites at https://phorion.io/blog/offical-claude-code-vsx-compromise/ doesn't make a claim about Visual Studio Marketplace, only "the official Claude Code for VS Code extension bundle on the Cursor Marketplace"
Comment · 2 days ago
How much are these customers paying for support to have Raymond Chen on first level support to debug their batch files?
Comment · 3 days ago
> So even if you’ve never created a User Profile in the browser, just visiting www.microsoft.com is enough for Edge to create one and sign you in automatically.
This is disgusting behaviour, RCE chain or not.
Comment · 12 days ago
Those tickets sound fairly non-dramatic - a trusted root certificate was dual-use, in breach of root certificate policy, and have committed to transition to TLS-only certificates and remove trust in the old ones.
But the Go issue is market urgent, and assigns it a CVE, so clearly there's more to it than that.
Comment · 1 month ago
That's just a search engine, but then you're competing with traditional players like Elasticsearch and Vespa who all have built-in vector support by now, and you have to compete on attributes like price, performance, features, and who can mention 'AI' the most times on their web page.
Comment · 1 month ago
I was wondering the same thing, but playing it you only fill in small border segments at a time, with the rest of the map filled in, so the projection doesn't change much.
Comment · 1 month ago
As they say in TDD, write a test, then write the simplest code that will make it pass.
Clearly supporting multiple functions starting with 'p' would be overengineering.
Comment · 1 month ago
The recurrent depth sounds a lot more like what is described in https://dnhkng.github.io/posts/rys/ - a way to add depth to a network without increasing the number of parameters.
e: While the actual CoT in neuralese paper is Facebook's Coconut https://arxiv.org/abs/2412.06769 - not sure if any production models use that one.
Comment · 2 months ago
You might find the home made batteries of this guy interesting as well https://www.youtube.com/watch?v=eq7fR9ISuCw
Comment · 2 months ago
It should be fine keeping the n-gram embeddings on SSD which lets you run at least the Q1 and Q2 models on 64GB
Comment · 2 months ago
Interesting that he pointed it to the CSB video - those are great explanations for us laymen, but I assume the LLM would prefer a written report.
Luckily it's still up https://youtube.com/watch?v=CFVUSDzHL8A
I was worried the videos would already be disappearing given Trump's attempts to shut it down https://www.chemicalprocessing.com/safety-security/risk-asse...
Comment · 2 months ago
> Not sure if modern models "think" only by outputting <thinking> blocks
That's pretty much it - a small refinement to "Chain of Thought" prompting, where you tell the model explicitly in the prompt to "Think step by step" or similar, so it writes out more steps before giving a final answer, potentially catching some errors. The "thinking" models are tuned to do that without being prompted to, and to output the "thinking" markers around it, so they can be hidden from the user.
Comment · 2 months ago
https://dam.stanford.edu/memory-prices.html
Pretty sure those devs were saying the same thing way before 2017 as well, which seems to be ~ the last time RAM was this expensive.
Now RAM in the cloud, now you're really paying the Java premium.
Comment · 4 months ago
> Most cryptographers would still recommend the hybrid over pure ML-KEM. This RFC (for pure MLKEM) is marked "recommended to implement = N". It is purely for settings where the implementors independently want to use pure ML-KEM for some reason.
That's exactly how it was with Dual_EC_DRBG.
E.g. https://www.schneier.com/essays/archives/2007/11/did_nsa_put...
I don’t understand why the NSA was so insistent about including Dual_EC_DRBG in the standard. It makes no sense as a trap door: It’s public, and rather obvious. It makes no sense from an engineering perspective: It’s too slow for anyone to willingly use it. And it makes no sense from a backwards-compatibility perspective: Swapping one random-number generator for another is easy.
My recommendation, if you’re in need of a random-number generator, is not to use Dual_EC_DRBG under any circumstances.
Except as we know now, after the standardization NSA was able to persuade/bribe vendors to implement it.
RSA is still a viable cryptography vendor, after accepting money to backdoor their product for paying customers. The standardization gave them a fig leaf of plausible deniability. Honest mistake, could happen to anyone, right? If they had needed to implement a "non-standard" backdoor, or if it had been officially struck from the standard, it would have been a lot harder to row away from.
Comment · 4 months ago
It also gives you the option of serialising the RNG states directly instead of using the counter hack.