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

Promising results from DeepSeek R1 for code

simonwillison.net

Post reply on HN
Print view View original

741–750 of 765 posts

Re: Promising results from DeepSeek R1 for code

#741

Post by troad »

kazinator wrote: post #740
troad wrote: post #734

Earlier quoted context omitted.

> Or you could just say, "Yes, white lives matter" and move on. Which people will interpret as a support for the far-right. You may not intend that, but that's how people will interpret it, and your intentions are neither here nor there. You may not care what people think, but your neighbours will. "Did you hear Jim's a racist?" "Do we really want someone who walks around chanting 'white lives matter' to be coaching…

> Which people will interpret as a support for the far-right. Those people might as well join the far right. > your intentions are neither here nor there If intentions really are neither here nor there, then we can examine a statement or question without caring about intentions. > Do we really want someone who walks around chanting 'white lives matter' to be coaching the high school football team? Well, no; it would…

> Well, no; it would have to be more like: Do we really want someone who answers "yes" when a racist asks "do white lives matter?" to be coaching the high schoool football team?

Uh huh, this is definitely a distinction Jim's neighbours will respect when deciding whether to entrust their children to him. /s

"Look Mary Sue, he's not racist per se, he's just really caught up on being able to tell people 'white lives matter'. World of difference! Let's definitely send our children to the man who dogmatically insists on saying 'white lives matter' and will start a fight with anyone who says 'yeah, maybe don't?'."

> I would say that looking for negative motivations and interpretations in everyone's words is a negative personality trait that is on par with racism, similarly effective in feeding divisiveness. It's like words have skin color and they are going by that instead of what the words say.

And I would say that you're engaged in precisely what you condemn - you're ascribing negative personality traits to others, merely on the basis that they disagree with you. (And not for the first time, I note.)

I would also firmly say none of what we're discussing comes anywhere near being on par with racism. (Yikes.)

Finally, I would say that a rabid, dogmatic insistence on being able to repeat the rallying cries of race-based trolling (your description), whenever one chooses and with absolutely no consequences, everyone else be damned, is not actually anything to valourise or be proud of. (Or is in any way realistic. You can justify it six ways till Sunday, but going around saying 'white lives matter' is going to have exactly the effect on the people around you that that rallying cry was always intended to have.)

>> You likely can, and yet I think the answer to their question is yes, white lives do matter, and someone in charge of children which include white children must think about securing a future for the white ones too.

I have nothing to say to someone who hears the Fourteen Words, is fully informed about their context, and then agrees with them. You're so caught up in your pedantry you're willing to sign up to the literal rhetoric of white nationalist terrorism. Don't be surprised when you realise everyone else is on the other side. And they see you there. (And that's based on the generous assumption that you don't already know very well what it is you're doing.)

Re: Promising results from DeepSeek R1 for code

#742

Post by kazinator »

troad wrote: post #741
kazinator wrote: post #740

Earlier quoted context omitted.

> Which people will interpret as a support for the far-right. Those people might as well join the far right. > your intentions are neither here nor there If intentions really are neither here nor there, then we can examine a statement or question without caring about intentions. > Do we really want someone who walks around chanting 'white lives matter' to be coaching the high school football team? Well, no; it would…

> Well, no; it would have to be more like: Do we really want someone who answers "yes" when a racist asks "do white lives matter?" to be coaching the high schoool football team? Uh huh, this is definitely a distinction Jim's neighbours will respect when deciding whether to entrust their children to him. /s "Look Mary Sue, he's not racist per se, he's just really caught up on being able to tell people 'white lives mat…

> On the basis that they disagree with you

On the basis that they are objectively wrong. I mean, they are guessing about the intent behind some words, and then ascribing that intent as the unvarnished truth to the uttering individual. How can that be called mere disagreement?

> being able to repeat the rallying cries

That's a strawman extension of simply being able to agree with the statement "white lives matter", without actually engaging in the trolling.

> I have nothing to say to someone who hears the Fourteen Words, is fully informed about their context, and then agrees with them.

If so, it must be because it's boring to say something to me. I will not twist what you're saying, or give it a nefarious interpretation, or report you to some thought police or whatever. I will try to find an interpretation or context which makes it ring true.

No risk, no thrill.

I actually didn't know anything about the Fourteen Words; I looked it up though. It being famous doesn't really change anything. Regardless of it having code phrase status, it is almost certainly uttered with a racist intent behind it. Nevertheless, the intent is hidden; it is not explicitly recorded in the words.

I only agree with some of the words by finding a context for the words which allows them to be true. When I do that, I'm not necessarily doing that for the other person's benefit; mainly just to clarify my thinking and practice the habit of not jumping to hasty conclusions.

Words can be accompanied by other words that make the contex clear. I couldn't agree with "we must ensure a future for white children at the expense of non-white children" (or anything similar). I cannot find a context for that which is compatible with agreement, because it's not obvious how any possible context can erase the way non-white children are woven into that sentence. Ah, right; maybe some technical context in which "white", "black" and "children" are formal terms unrelated to their everyday meanings? But that would be too contrived to entertain. Any such context is firmly established in the discourse. Still, if you just overhear a fragment of some conversation between two people saying something similar, how do you know it's not that kind of context? Say some computer scientists are discussing some algorithm over a tree in which there are black and white nodes, some of those being children of other nodes. They can easily utter sentences that have a racist interpretation to someone within earshot, which could lead them to the wrong conclusion.

Re: Promising results from DeepSeek R1 for code

#743

Post by KarraAI »

Been testing Deepseek R1 for coding tasks, and it's really impressive. The model nails Human Eval with a score of 96.3%, which is great, but what really stands out is its math performance (97.3% on MATH-500) and logical reasoning (71.5% on GPQA). If you're working on algorithm-heavy tasks, this model could definitely give you a solid edge.

On the downside, it’s a bit slower compared to others in terms of token generation (37.2 tokens/sec) and has a lower output capacity (8K tokens), so it might not be the best for large-scale generation. But if you're focused on solving complex problems or optimizing code, Deepseek R1 definitely holds its own. Plus, it's incredibly cost-effective compared to other models on the market.

Re: Promising results from DeepSeek R1 for code

#744

Post by Vegenoid »

kragen wrote: post #733
Vegenoid wrote: post #722

Earlier quoted context omitted.

The point was not to be very accurate, it was to make the point that it has existed for a very short amount of time on the scale of humanity. Quibbling over whether software engineering started in the 40s or the 50s and whether that is greater or less than an average life expectancy is beside the point.

People posting comments without caring whether they are true or false undermines the presumption of good faith that underlies rational discourse. Please stop posting such comments on this site. Instead, only post comments that you have some reason to believe are true.

It is not about not caring if the statement is true or false. The statement neither absolutely true nor absolutely false, because there is no absolute definition of when software engineering started, or what "a lifetime" is. It is about making a statement that communicates information. However, if I must prove that my statement can reasonably considered "true" in order to prove that it does communicate the short span of time for which software engineering has existed:

- Avg. life expectancy in USA: 77.5 years

- 2025 - 77.5 = 1947.5

- In 1945, Turing published "Proposed Electronic Calculator"

- The first stored-program computer was built in 1948

- The term "software engineering" wasn't used until the 1960s

If you want to define "software engineering" such that it is more than 77.5 years old, that's fine. But saying that software engineering is less than 77.5 years old is clearly a reasonable stance.

Please stop berating me for a perfectly harmless and reasonably accurate statement. If you're going to berate me for anything, it should be for its brevity and lack of discussion-worthy content. But those are posted all the time.

Re: Promising results from DeepSeek R1 for code

#745

Post by the_af »

boothby wrote: post #729
svara wrote: post #711

Earlier quoted context omitted.

> AI won't scrub a toilet, but it will cheaply and inexhaustibly do every job that humans find meaningful today That's the part I really don't believe. I'm open to being wrong about this, the risk is probably large enough to warrant considering it even if the probability of this happening is low, but I do think it's quite low. We don't actually have to build artificial humans. It's very difficult and very far away. I…

To be clear, I'm much more concerned about the rise of techo-authoritarianism than employment. And further ahead, where I said your original take might not age well; I'm also not worried about AI making humanoid bodies. I'd be worried about a future where mines, factories, and logistics are fully automated: an AI for whom we've constructed a body which is effectively the entire planet . And nobody needs to set out to…

> I'm also not worried about AI making humanoid bodies. I'd be worried about a future where mines, factories, and logistics are fully automated: an AI for whom we've constructed a body which is effectively the entire planet.

I know scifi is not authoritative, and no more than human fears made into fiction, but have you read Philip K. Dick's short story "Autofac"?

It's exactly what you describe. The AI he describes isn't evil, nor does it seek our extinction. It actually wants our well-being! It's just that it's taken over all of the planet's resources and insists in producing and making everything for us, so that humans have nothing left to do. And they cannot break the cycle, because the AI is programmed to only transition power back to humans "when they can replicate Autofac output", which of course they cannot, because all the raw resources are hoarded by the AI, which is vastly more efficient!

Re: Promising results from DeepSeek R1 for code

#746

Post by snapcaster »

Etherlord87 wrote: post #708
snapcaster wrote: post #699

Earlier quoted context omitted.

In some places censorship is done to make the space safe for advertisements. In other places it's to maintain social harmony. I wish people could get out of this reflexive "china bad and i must mention that every time the country is discussed" mindset it's so toxic and limiting

Criticizing malice is never toxic. I wish people could get out of this reflexive "you criticize my country? But your country is also bad because..." - it shouldn't even be treated as a counterargument, but as an admission of guilt.

I'm not bothered by criticism of China, it's the context and fixation. When an american company releases technology, i don't see the comments full of "Iraq war killed a million people" but i do see similar for China. It's just so exhausting

Re: Promising results from DeepSeek R1 for code

#747

Post by Etherlord87 »

snapcaster wrote: post #746
Etherlord87 wrote: post #708

Earlier quoted context omitted.

Criticizing malice is never toxic. I wish people could get out of this reflexive "you criticize my country? But your country is also bad because..." - it shouldn't even be treated as a counterargument, but as an admission of guilt.

I'm not bothered by criticism of China, it's the context and fixation. When an american company releases technology, i don't see the comments full of "Iraq war killed a million people" but i do see similar for China. It's just so exhausting

So if the America was criticized, then criticizing China wouldn't bother you? How come every time China or russia are criticized, the comment section is filled with "but America too!"? The way I see it, it's just an admittance of guilt by supporters of the empire of evil.

Re: Promising results from DeepSeek R1 for code

#748

Post by boothby »

the_af wrote: post #745
boothby wrote: post #729

Earlier quoted context omitted.

To be clear, I'm much more concerned about the rise of techo-authoritarianism than employment. And further ahead, where I said your original take might not age well; I'm also not worried about AI making humanoid bodies. I'd be worried about a future where mines, factories, and logistics are fully automated: an AI for whom we've constructed a body which is effectively the entire planet . And nobody needs to set out to…

> I'm also not worried about AI making humanoid bodies. I'd be worried about a future where mines, factories, and logistics are fully automated: an AI for whom we've constructed a body which is effectively the entire planet. I know scifi is not authoritative, and no more than human fears made into fiction, but have you read Philip K. Dick's short story "Autofac"? It's exactly what you describe. The AI he describes is…

I think that science fiction plays an important role in discourse. Science fiction authors dedicate years deeply contemplating potential future consequences of technology, and packaging such into compelling stories. This gives us a shorthand for talking about positive outcomes we want to see, and negative outcomes that we want to avoid. People who argue against scifi with a dismissal that "it's just fiction" aren't participating in good faith.

On the other hand, it's important not to pay too close attention to the details of scifi. I find myself writing a novel, and I'm definitely making decisions in support of a narrative arc. Having written the comment above... that planetary factory may very well become the third faction I need for a proper space opera. I'll have to avoid that PKD story for the moment, I don't want the influence.

Though to be clear, in this case, that potentiality arose from an examination of technological progress already underway. For example, I'd be very surprised if people aren't already training LLMs on troves of viruses, metasploit, etc. today.

Re: Promising results from DeepSeek R1 for code

#749

Post by snapcaster »

Etherlord87 wrote: post #747
snapcaster wrote: post #746

Earlier quoted context omitted.

I'm not bothered by criticism of China, it's the context and fixation. When an american company releases technology, i don't see the comments full of "Iraq war killed a million people" but i do see similar for China. It's just so exhausting

So if the America was criticized, then criticizing China wouldn't bother you? How come every time China or russia are criticized, the comment section is filled with "but America too!"? The way I see it, it's just an admittance of guilt by supporters of the empire of evil.

I would prefer if achievements in countries could be discussed without detailing a list of everything they've done wrong. I was using Iraq as an example, i would also be annoyed if someone brought it up every time OpenAI releases a new product
Post reply on HN