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

Qwen 3.8

twitter.com

Post reply on HN
Print view View original

611–620 of 793 posts

Re: Qwen 3.8

#611

Post by tough »

behnamoh wrote: post #535
seizethecheese wrote: post #488

Earlier quoted context omitted.

There’s a simpler explanation, which is that this is how Chinese business operates. When I was in China earlier this year the big topic of conversation was “overproduction”. The big example was electric cars, where there were too many companies making too many cars and making revenue but no profit. It was explained to me that generally Chinese firms will compete hard and maximize revenue above all, whereas western fi…

> It was explained to me that generally Chinese firms will compete hard and maximize revenue above all, whereas western firms tend to focus on profit. ok but why?

[flagged]

Re: Qwen 3.8

#612

Post by oofbey »

solenoid0937 wrote: post #456
oofbey wrote: post #375

Earlier quoted context omitted.

These things take months to train. No chance this is a reaction to what just happened.

Distills don't take months to train, they take weeks. Distills are very easy to train.

That depends. There are two wry different processes that both get called distillation. One is where you have a fully trained large model and you are converting it to a smaller model. That kind if vastly cheaper than training a full model. Sure you could do that in weeks maybe even days for a big model. But it requires you to have the weights of the teacher model.

The other kind of distillation is where you record the outputs from a teacher model and use it to train a smaller model from scratch. That kind of distillation is not so cheap. It’s cheaper than training a model fully from scratch - starting with pretraining, then alignment, RLHF, the whole riggamarole. But here you are still starting from nothing and need to figure out how to get trillions of random numbers aligned in a way that makes them act intelligent. This is still gonna take a very long time if you’re talking about trillions of parameters.

Re: Qwen 3.8

#613

Post by 1over137 »

vkou wrote: post #97
maxignol wrote: post #82

Earlier quoted context omitted.

Shouldn’t we fear they start doing only close source like most us labs once they catch up in market shares ?

Yeah, but that mousetrap keeps working for SV startups, what makes you think it won't work for Chinese ones? Uber spent a decade undermining taxis, and once it had market share, it stopped giving away rides and raised prices. It now costs more than a regular taxi, with the quality of the ride being... At best proportionate to the premium in price.

Uber costs more than regular taxi? In what country/region? Not where I am.

Re: Qwen 3.8

#614

Post by postepowanieadm »

jcelerier wrote: post #601
akoboldfrying wrote: post #498

Earlier quoted context omitted.

I'll supply such a comment: Any software open sourced by any for-profit company, including Google, is a calculated move ultimately intended to increase their bottom line, and it's naive to think otherwise.

Isn't it borderline illegal for publicly traded companies to not try to maximize profit? "interest of the company" in theory but in practice

Think about the shareholders!

Re: Qwen 3.8

#615

Post by xiconfjs »

nsbk wrote: post #55
psychoslave wrote: post #30

Earlier quoted context omitted.

What hardware do you have?

I run a 2x 3090 rig, but a single 3090 already provides a great experience at a reasonable quant and context size. On a single card I used to run Qwen_Qwen3.6-27B-Q4_K_M or similarly quantized 35B MoE at 65536 context size

you should look into https://github.com/noonghunna/club-3090

Re: Qwen 3.8

#617

Post by throwa356262 »

cnhwl wrote: post #583

I’m a developer from China. So, is this what Hacker News is all about? Whenever a model comes from China, the comments section stops discussing its technical architecture, optimisation points or real-world performance, and instead starts going on about politics, human rights and all that rubbish? To be honest, we Chinese IT professionals possess a genuine geek spirit. That’s why you’re lagging behind in open-source c…

Yes, and its geting annoying

If you know of another forum with less politics or VC talk and more technical depth please let us know :)

Re: Qwen 3.8

#618

Post by epihelix »

zozbot234 wrote: post #269
baq wrote: post #78

Earlier quoted context omitted.

There’s a Twitter thread making rounds by Dean Ball about deceleration in AI development caused by open models and I can’t understand how people don’t see that it’s true: open models dismantle the frontier lab capex spend potential by reducing the training budget to zero in the limit. Tokens from different providers are not fungible, but customers are nevertheless very price sensitive and close enough is good enough,…

> There’s a Twitter thread making rounds by Dean Ball about deceleration in AI development caused by open models and I can’t understand how people don’t see that it’s true: open models dismantle the frontier lab capex spend potential by reducing the training budget to zero in the limit. If you're worried about an AGI arms race between the U.S. and China putting AI Safety at risk, then the fact that inherently less kn…

This is hypothesised future deceleration, I'm guessing? Because we've seen the exact opposite of deceleration from closed models over the last seven months.

Re: Qwen 3.8

#619

Post by d5lt5 »

WillPostForFood wrote: post #469
asdewqqwer wrote: post #410

Earlier quoted context omitted.

It is hilarious to see people from arguable the most polarized political systems in the world believing the evil 1.5 billion people across the sea share one single mind, either a saint, or a devil.

It will be a great day for China and the world when the Chinese people are free from a totalitarian dictatorship. But until then we have to speak of the policy of the Chinese government as the policy of China, even if many, or most disagree with those policies.

Nickname checks out.

Re: Qwen 3.8

#620

Post by deaux »

jcelerier wrote: post #601
akoboldfrying wrote: post #498

Earlier quoted context omitted.

I'll supply such a comment: Any software open sourced by any for-profit company, including Google, is a calculated move ultimately intended to increase their bottom line, and it's naive to think otherwise.

Isn't it borderline illegal for publicly traded companies to not try to maximize profit? "interest of the company" in theory but in practice

No, it is not. This is a common misconception that the large companies are very happy about.

1. The century-old Ford decision wasn't about this. It was about him refusing to pay dividends to shareholders he was feuding with. I.e. dominant shareholder using his control to starve minority shareholders of returns.

The court still let him keep spending huge sums on factories and price cuts that didn't clearly maximize profits.

2. Even if the case was about profit maximization (it wasn't), the judgment was a Michigan state decision. It doesn't have force outside of it.

3. US business law is in practice actually the opposite. A business can do pretty much whatever it wants. This is fundamental.

4. Even if it were illegal (it's not), anything can reasonably be framed as being in the long-term interest of the company, including donating to causes, raising employee wages and so on. Courts never question this.

Just imagine this was a thing. It's completely untenable for this reason. Who is a court to judge that something isn't in the longterm interest of shareholders unless it's literally spending all the company money on yachts for personal use?

5. No company has ever been prosecuted for this, obviously, because it's not a thing that exists. The closest you can get is that there's a potential duty to seek the best price if a company must be sold or broken up. But that's a very specific situation.

It's 100% a myth. Feel free to copy this and spread it when you see someone saying this.

Post reply on HN