Hacker Newsnew | past | comments | ask | show | jobs | submit | xiphias2's commentslogin

> Temperature / top-k sampling in verify. Currently greedy-only

This is interesting, doesn't greedy-only decoding slow down speculative decoding significantly?

In theory the probability of needing resampling (rejection) is (p_real-p_sample)+, which should be much smaller with non-greedy distribution


It was from GPT-2 and Dario was part of the developers of that model while he was working in OpenAI, not Sam Altman, it's his playbook

> It was from GPT-2

Prior to the released of GPT-5, Sam said he was scared of it and compared it to the Manhattan Project.


Not just Altman. Buffett said it also, more generally.

https://youtu.be/vZlMWF6iFZg


This is pretty much correct, but Mustafa Suleyman has probably been doing it longer.

Not just part of the developers, but rather "led the development of large language models like GPT-2 and GPT-3" as per his website.

https://darioamodei.com/


It's against xAI Colossus datacenter in Memphis


AI is unstoppable as the algorithmic changes are at least as fast as compute improvements. One proof of it is that even older GPUs' value is going up.

They have a choice: study chess and beat Magnus. Until then I will care about Magnus (and lichess) more then those businesses.

The best thing they did was that they bought an amazing domain name.


Isn't treesitter integrated to nvim anyways at this point, even if it's experimental support?


You could probably make things work without nvim-treesitter, but it's an additional maintenance burden you're taking on. As the repo itself says, it's an abstraction layer. You don't _need_ it, but it's nice to have.


Because it wouldn't look good for the EU


Ah yes you choose Switzerland. The country that is famously so affordable to live in.

Last time I have been there food has costed literally double than in my home country which is also in the Alps, also has that standard of living, etc.

But hey, you moderately save on taxes if you finally manage to settle there, which csn be a bit of a challenge since switzerland isn't the most easy country to get permanent residence in.

As a European who has traveled to nearly all European countries and lived in 6 different ones, the idea to only look at the taxrate when chosing your future home country sounds so ridiculously simplistic and money-focused that it could only have come from an American author.

The point of having more money is to lead a happy live. In some countries you need more money to do that than in others. And depending on your character, hobbies and goals in life some countries will make it much more expensive than others.

For example if you like enjoying a beer in the sun and you start living in Island because of the tax rate, your first winter depression will make you question how smart that really was. Or you will just travel to southern Europe on a regular basis, but then why not live there in the first place.


If your country is Italy that might be the case, but groceries are at most 30% more expensive than France, and some are nearly the same price (vegetables). Meat and fish do cost an arm and a leg (100% tax on border crossing).

Meanwhile, median net salary in CH is 5'000-5'500 per month, double to triple its neighbors. So food is actually very affordable.

The food that costs more is the one someone cooked for you, which is logical considering the cook is likely paid more than your engineer (assuming that's your case) salary. But then again, minimum wage Italians are not eating out at the restaurant with any frequency. If you were an engineer in Switzerland instead, you could afford eating out there. The restaurants and terraces are never empty, anyways.

Now, if you want to enjoy a beer in the sun, you can get a 2CHF can at the supermarket and go fire up a barbecue at the lake of Zurich, I see people doing that all the time.


Another project without running real benchmarks. It's very easy to generate tokens, it's much harder to solve tasks locally.


Here is a reference https://www.sharpai.org/benchmark/ For specific tasks, local model could achieve workable level.


It doesn't look like RAM, CPU GPU or bandwidth is getting cheaper if that helps you, quite the opposite.


Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: