Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

YaLM-100B: Pretrained language model with 100B parameters

github.com

Post reply on HN
Print view View original

91–100 of 666 posts

Re: YaLM-100B: Pretrained language model with 100B parameters

#91

Post by ggktk »

alexb_ wrote: post #13

I have to wonder if 10 years down the line, everyone will be able to run models like this on their own computers. Have to wonder what the knock-on effects of that will be, especially if the models improve drastically. With so much of our social lives being moved online, if we have the easy ability to create fake lives of fake people one has to wonder what's real and what isn't. Maybe the dead internet theory will rea…

I'm predicting that the upcoming Mac Pro will be very popular among ML developers, thanks to unified memory. It should be able to fit the entire model in memory.

Combine that with the fact that PyTorch recently added support for Apple silicon GPUs.

Re: YaLM-100B: Pretrained language model with 100B parameters

#93

Post by kamray23 »

espadrine wrote: post #50
alexb_ wrote: post #13

I have to wonder if 10 years down the line, everyone will be able to run models like this on their own computers. Have to wonder what the knock-on effects of that will be, especially if the models improve drastically. With so much of our social lives being moved online, if we have the easy ability to create fake lives of fake people one has to wonder what's real and what isn't. Maybe the dead internet theory will rea…

> I have to wonder if 10 years down the line, everyone will be able to run models like this on their own computers. Isn’t that already the case? Sure, it costs $60K, but that is accessible to a surprisingly large minority, considering the potency of this software.

You're grossly overestimating. People who make 60k annually are getting a bit rarer nowadays, it's not like everyone can afford it. For the majority of people it'd be a multi-decade project, for a few it might only take 7 years, very few people could buy it all at once.

Re: YaLM-100B: Pretrained language model with 100B parameters

#94

Post by kamray23 »

throwaway4good wrote: post #92
f311a wrote: post #78

Earlier quoted context omitted.

Nvidia DGX

I can see it is about 150.000 USD for such a machine. Is this the cheapest option out there?

Well you can custom-build a suitable system for the middle five digits. It's still not something every idiot can run, but most medium to large companies can set up their own for sure.

Re: YaLM-100B: Pretrained language model with 100B parameters

#95

Post by yarandex »

Aardwolf wrote: post #20
julienfr112 wrote: post #14

Earlier quoted context omitted.

What about 250gb of ram and use a cpu ?

Way too slow on CPU unfortunately But this does make me wonder if there's any way to allow a graphics card to use regular RAM in a fast way? AFAIK built-in GPU's inside CPU's can but those GPU's are not powerful enough

Assuming running on CPU is memory-bandwidth limited, not CPU-limited, it should take about 200GB / (50GB/sec) = 4 seconds per character. Not too bad.

Re: YaLM-100B: Pretrained language model with 100B parameters

#96

Post by kamray23 »

toyg wrote: post #45
f6v wrote: post #8

Earlier quoted context omitted.

The achievements aren’t in any way tainted by their nationality, citizenship, sex, sexual orientation, age, etc.

Well... I'm sorry if I reach for the reductio at Hitlerum, but any achievements Nazi scientists might have reached in concentration camps are definitely tainted. Similarly, achievements in the field of online consumer analysis in a country where consumer-privacy protections are nonexistent, surely should be considered tainted...?

They are and we use them all the same. Rockets fly almost every week now, jet engines are the most common form of propulsion, tons of medicine forcibly tested on innocent people is on the market, and to pass up any of that technology would be pure idiocy.

Re: YaLM-100B: Pretrained language model with 100B parameters

#97

Post by tehsauce »

ggktk wrote: post #91
alexb_ wrote: post #13

I have to wonder if 10 years down the line, everyone will be able to run models like this on their own computers. Have to wonder what the knock-on effects of that will be, especially if the models improve drastically. With so much of our social lives being moved online, if we have the easy ability to create fake lives of fake people one has to wonder what's real and what isn't. Maybe the dead internet theory will rea…

I'm predicting that the upcoming Mac Pro will be very popular among ML developers, thanks to unified memory. It should be able to fit the entire model in memory. Combine that with the fact that PyTorch recently added support for Apple silicon GPUs.

upcoming mac pro will have pretty poor ML performance when compared to even an old nvidia gpu sadly.

Re: YaLM-100B: Pretrained language model with 100B parameters

#98

Post by ptnxlo »

tremarley wrote: post #63
csee wrote: post #49

Earlier quoted context omitted.

Your comparison fails a test of facts. Yandex actively censors any perspective not approved by the Kremlin. Google does not do anything comparable to this.

Google absolutely does the same thing.

Care to elaborate?

Re: YaLM-100B: Pretrained language model with 100B parameters

#100

Post by egorfine »

danuker wrote: post #9
egorfine wrote: post #6

I have huge respect for developers at Yandex. It's kind of sad that achievements like these are tainted by the fact that they come from Russia (and I speak as a Ukrainian). I wonder if the permissive license is able to mitigate that.

Coming from Russia doesn't mean you agree with government policy. If you saw people get arrested as soon as they start protesting, what would you do?

I am asking this question myself for 120 days and I still don't have an answer.
Post reply on HN