verdverm

17 ms·

571.
by verdverm 2mo ago
actually, bar is true but wait, the models constantly go back and forth on these things in their thinking traces, so it is unclear which self correcting is actually correct
572.
by verdverm 2mo ago
Quantization is typically very cheap and fast. It can even be done on hardware that does not fit the model, by processing the weights layer by layer. I use this project: https://github.com/vllm-project/llm-compressor
573.
by verdverm 2mo ago
I'm unaware of any actual attempts at socialism ever achieving those things, so perhaps socialism is defined otherwise in reality?
574.
by verdverm 2mo ago
vllm PR for DFlash2: https://github.com/vllm-project/vllm/pull/52816
575.
by verdverm 2mo ago
I'm sympathetic to people who take issue with being under the thumb of a regime that cuts up journalists https://en.wikipedia.org/wiki/Assassination_of_Jamal_Khashog... Gamers will complain, or just not buy, if th
576.
by verdverm 2mo ago
What's in a name? That which we call socialism / By any other name would smell as sweet.
577.
by verdverm 2mo ago
this is well studied and that emotional trigger level used to be much lower, iirc - inflation pricing has a half life of 6 months - wage increases raise the emotional trigger level People feel prices through their wallets. If their wallets
578.
by verdverm 2mo ago
I'm running qwen3.8 27B dense on reasonable hardware (oem spark) tbh, I have stopped using MoE in the name of speed, the dense (with more active parameters) makes a real difference in output quality
579.
by verdverm 2mo ago
Taking a % cut on top of every operation
580.
by verdverm 2mo ago
Stripe operates in a very regulated industry where new entrance is difficult OpenRouter is commodity stuff, I've never used it personally and picked alternatives. The sense I got from people who said they used it was that they are on a
581.
by verdverm 2mo ago
for sure, there is a minimum size and knowledge base that is required to be useful at the same time, search may find newer or better alternatives, and you can always specify specific technologies you want to use, I typically do this when st
582.
by verdverm 2mo ago
imo, node is fine, they have done a lot to improve and the menagerie of js runtimes have largely converged
583.
by verdverm 2mo ago
there are linters that check for dead code (depending on your language / ecosystem), it should always be 0%
584.
by verdverm 2mo ago
You're breaking HN rules by submitting this link over and over https://news.ycombinator.com/submitted?id=KiaanKothari Please review the guidelines and faq linked at the bottom of most pages
585.
by verdverm 2mo ago
> EGA V9 achieved 100% detection performance with 0% false-positive and false-negative rates This is a big red flag, you should double check your work or reevaluate your evaluation data. No one gets a perfect score There is no related wo
586.
by verdverm 2mo ago
They didn't afaict https://artificialanalysis.ai/agents/coding-agents?coding-ag... It seems it takes some time to run a new model on all the benchies, not sure they run all models on all of them either
587.
by verdverm 2mo ago
I do not rely on any LLM of any size for general knowledge baked into the weights, they all hallucinate and that is the wrong way to hold them imo I think there is some merit in that smaller models cannot memorize so much of the training da
588.
by verdverm 2mo ago
save qwen3.8 27B which is outclassing much larger models and is in spitting distance of the top 10 in https://artificialanalysis.ai/models#intelligence
589.
by verdverm 2mo ago
or the latest qwen3.8 27B doing so well at ~1/100 the size of K3
590.
by verdverm 2mo ago
depends on if it is a one time purchase or you also have to buy fuel for them, solar is a one and done kind of deal, no need to purchase petrol from belligerents
591.
by verdverm 2mo ago
more times than the number of ships transiting the strait, which is a key fact omitted from the submitted "news", you don't have to control the strait (physically) to control (traffic in) the strait
592.
by verdverm 2mo ago
most americans do not have a 401k
593.
by verdverm 2mo ago
drumming the EA beat... I believe it when it actually happens at scale, because they do do scale in SV still, yea?
594.
by verdverm 2mo ago
hopefully we the people, who pay the taxes for such an idea, will reject it, as I am of the opinion we should not fund investor returns with taxpayer money
595.
by verdverm 2mo ago
> we're working on our own extensions to the protocol what I hear is another path to app centralization, regardless of your intent, this is what comes out of custom extensions to a protocol bluesky should have actually been open abo
596.
by verdverm 2mo ago
> I am just thinking you ... you are incorrect and this is why it's advised not to project the answer but ask a question instead for clarity, I do in fact use github's social features I also believe that the idea around the atm
597.
by verdverm 2mo ago
atproto is not the next thing, there will be more social protocols to come tangled will struggle to have private repos that align with user expectations for RBAC, atproto ACL permission proposal is insufficient, Bluesky is not interested in
598.
by verdverm 2mo ago
than that is the exceptional case, not the default I struggle to think of a case where it is technically impossible to have an API because the website is presumably making API calls under the hood
599.
by verdverm 2mo ago
youtube without ads
600.
by verdverm 2mo ago
Too big to fail? I for one will not trust or use Ai created or supported by my government, it's too corrupt and broken

More ›