Tbh this just feels like the same playbook as OAI. Open start and then less so over time.
Mistral has been holding the line on that topic remarkable well.
DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
161–170 of 268 posts
Re: DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
#162Earlier quoted context omitted.
> DeepSeek is a small company that made a lot of money from other businesses, which makes their lack of focus on commercial interests feel more genuine. Google also made a lot of money from other businesses that aren't AI models, until they started selling AI models, just as DeepSeek now does. The reality is that DeepSeek is a full company, that was funded as a spin-off from the original business (a hedge fund that u…
> The reality is that DeepSeek is a full company, that was funded as a spin-off from the original business (a hedge fund that used its large GPU stockpile to pick stocks via ML). The company DeepSeek is owned by the hedge fund CEO not the hedge fund. It exists as a business aiming to make money, not as a pet project for another business. Of course they want money, lots of money, tons of money is required for hiring e…
Zuckerberg, who is also a developer, and countless other CEOs are listed on many patents from their companies. Doesn’t mean they actually had a strong input in the invention.
No business exists not to make money because that is a charity. It’s not a charity, because a charity is not a business, and DeepSeek is a business. I don’t care to quibble about how interesting they are in being a lucrative business, but simply that they’re a business.
My point wasn’t to question their motives about profit vs AGI (why would these be mutually exclusive btw), but to challenge the notion that it’s some side project from a random business. It’s a company with dedicated resources and staff.
Re: DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
#163Earlier quoted context omitted.
Those valuations are built on an imaginary future the founders made investors believe. The idea is: if we reach true AGI first, we are going to own ALL THE MONEY! Which erroneously assumes that models can't be siphoned off/recreated, as deepseek proved possible and even reasonably doable. Which in turn fundamentally shows that both openai and anthropic very likely have basically no moat. I can almost smell another AI…
I cant see a future where AGI exists and money in general isn't worthless within 6 months of it existing. Either it kills us all, or makes the creator so much money that it's essentially worthless because they're the only one with money, or creates a utopia where money isn't needed.
The laws of economics apply just as much to AI as they do to humans, if anything AI is an even better (more rational) homo economicus. Even if AI wiped out all humans, the AIs would still need a monetary system for trading among themselves.
Re: DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
#164“Pure garage-energy” is a great phrase. Most interested to see their inference stack, hope that’s one of the 5. I think most people are running R1 on a single H200 node but Deepseek had much lower RAM per GPU for their inference and so had some cluster based MoE deployment.
I don't think the RAM size of the H800 was nerfed (80GB), but rather the memory bandwidth between gpus. But yeah, would be interesting to see how they optimized for that.
Correct. There are 3 main ways to "gimp" high end GPUs meant for training - "cores", "on-chip memory speed" and "interconnects". IIUC the H800 had the first 2 unchanged but halved the interconnect speeds.
H20 is the next iteration of the "sanctions" that I believe also limited the "cores" but left the on-chip memory intact, or slightly higher (from the new generation).
Re: DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
#165The funding company holds the assets, and the news make the stock market blooming and they make money!
Re: DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
#166I really admire their mindset of striving for the betterment of humanity. There was a time when OpenAI, Anthropic, and even Musk used to talk with that same lofty vision. But now, they've all shifted to competing for national interests instead, which is honestly quite disappointing.
We are making the world a better place more than our competitors
you forgot to add "/sarcasm"
Re: DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
#167I really admire their mindset of striving for the betterment of humanity. There was a time when OpenAI, Anthropic, and even Musk used to talk with that same lofty vision. But now, they've all shifted to competing for national interests instead, which is honestly quite disappointing.
Saying that Musk "doesn't have the mindset" for betterment of humanity is just ignorant in a very short-sighted way. Sure, he currently has a side project of fixing the US government and ensuring US doesn't stray too far outside of its core interests, but SpaceX and Tesla are still his bread and butter he has spent most of his time on beside this scenic route. I've followed him closely since ~2016 so I can say this w…
I think you've been drinking the koolaid too much. He's only in it to enrich himself and his cronies. There's a reason he's on course to become a trillionaire and it ain't because of altruism.
Re: DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
#168DeepSeek is seeking deep to Open AI.
irony
Re: DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
#169Earlier quoted context omitted.
I have a more practical view: there's nothing wrong in making profit, the important thing is that they are also doing some good.
> As though you rendered the proletarians a service in first sucking out their very life-blood and then practicing your self-complacent, Pharisaic philanthropy upon them, placing yourselves before the world as mighty benefactors of humanity when you give back to the plundered victims the hundredth part of what belongs to them! Friedrich Engels: The Condition of the Working-class in England
this doesn't argue whether existence of profits necessarily implies exploitation of workers but asserts it and then proceeds to argue against philanthropy funded by profits. This line of reasoning only makes sense if one already accepts the initial assumption, whereas the original poster questions that very assumption, so it's a bit irrelevant quote.
Re: DeepSeek Open Infra: Open-Sourcing 5 AI Repos in 5 Days
#170Earlier quoted context omitted.
Semianalysis is a trusted and reputable source that has deep knowledge of the semiconductor industry and supply chains. So yes, they know. > love your joke! > Get a real life please. > Too bad that you didn't learn this back in school. Read the site guidelines, as you’re repeatedly resorting to personal attacks in your comments on discussions around DeepSeek or China. https://news.ycombinator.com/newsguidelines.html
SemiAnalysis has made up many things. They claim that a small Chinese hedge fund could acquire $1bln in GPUs, with no state support, including many sanctioned chips, then trained a model optimized for a far smaller server compute size, and that they have a source at this very small fund who is willing to admit to export violations. A 40bln param active model is exactly the size you would expect from a server of the s…
They had their A100s back in 2021 to early 2022, well before any GPU sanction kicked in.
For a few months H800 wasn't sanctioned and that's when they bought them.