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

Deepseek R1-0528

huggingface.co

Post reply on HN
Print view View original

161–170 of 264 posts

Re: Deepseek R1-0528

#161

Post by kreijstal »

hbbio wrote: post #115
doctoboggan wrote: post #27

Earlier quoted context omitted.

Hard to say exactly how it will affect the market, but IIRC when deepseek was first released Nvidia stock took a big hit as people realized that you could develop high performing LLMs without access to Nvidia hardware.

Actually, the "narrative" crashed Nvidia for no reason. Not only DeepSeek uses a lot of Nvidia hardware for the training. But even more so, by releasing an open weight frontier model, people around the world need more Nvidia chips than ever for inference.

someone has not heard about huawei GPU

Re: Deepseek R1-0528

#162

Post by nprateem »

codedokode wrote: post #134
karencarits wrote: post #119

What use cases are people using local LLMs for? Have you created any practical tools that actually increase your efficiency? I've been experimenting a bit but find it hard to get inspiration for useful applications

Anyone who does not want to leak their data? I am actually surprised that people are ok with trusting their secrets to a random foreign company.

No one cares about your 'secrets' as much as you think. They're only potentially valuable if you're doing unpatented research or they can tie them back to you as an individual. The rest is paranoia.

Having said that, I'm paranoid too. But if I wasn't they'd have got me by now.

Re: Deepseek R1-0528

#163

Post by zackangelo »

omneity wrote: post #66
lodovic wrote: post #42

Earlier quoted context omitted.

If you use the excess memory for AI only it's cheaper to rent . A single H100 costs less than $2 per hour. (incl power)

Worth mentioning that a single H100 (80-96GB) is not enough to run R1. You're looking at 6-8 GPUs on the lower end, and factor in the setup and download time. An alternative is to use serverless GPU or LLM providers which abstract some of this for you, albeit at a higher cost and slow starts when you first use your model for some time.

Yeah, to run the full precision model you need either two 8xH100 nodes connected via Infiniband or one 8xH200 node or one 8xB200 node.

Not for the GPU poor, to be sure.

Re: Deepseek R1-0528

#165

Post by yetihehe »

otabdeveloper4 wrote: post #152
piperswe wrote: post #102

Earlier quoted context omitted.

But where's the source? I just see a binary blob, what makes it open source?

You can fine-tune their weights and release your own take. E.g. see all the specialized third-party models out there based on Qwen. "Open-source" is the wrong word here, what they mean is "you can modify and redistribute these weights".

You can also reverse engineer and modify closed source programs (see mods for games). Weights are like compiled version of source data.

Re: Deepseek R1-0528

#166

Post by pdimitar »

ribelo wrote: post #92
lossolo wrote: post #57

Earlier quoted context omitted.

The only benchmarks that match my experience with different models are here https://livebench.ai/#/

livebench was good, but now it's a joke. Gemini flash is better in coding than pro and sonnet 3.7. And this is only the beginning of weird results.

Flash is better than Pro in coding? Whoa... [makes a note to try a few things later this day]

Out of curiosity, how did you gauge that?

Re: Deepseek R1-0528

#167

Post by dyauspitr »

jsemrau wrote: post #121
karencarits wrote: post #119

What use cases are people using local LLMs for? Have you created any practical tools that actually increase your efficiency? I've been experimenting a bit but find it hard to get inspiration for useful applications

I have a signal tracer that evaluates unusual trading volumes. Given those signals, my local agent receives news items through API to make an assessment what happens. This helps me tremendously. If I would do this through a remote app, I'd have to spend a several dollars per day. So I have this on existing hardware.

Do you want to share it?

Re: Deepseek R1-0528

#168

Post by jumski »

timschmidt wrote: post #137
jacob019 wrote: post #126

Earlier quoted context omitted.

The weights are the source. It isn't as though something was compiled into weights. They're trained directly. But I know what you mean, it would be more open to have the training pipeline and souce dataset available.

The weights seem much more like a binary to me, the training pipeline the compiler, and the training dataset the source.

Come here to write this - perfect analogy!

Re: Deepseek R1-0528

#169

Post by reedciccio »

echelon wrote: post #104
fragmede wrote: post #77

Earlier quoted context omitted.

It's. not. open. source! https://www.downloadableisnotopensource.org/

Open source is a crazy new beast in the AI/ML world. We have numerous artifacts to reason about: - The model code - The training code - The fine tuning code - The inference code - The raw training data - The processed training data (which might vary across various stages of pre-training and potentially fine-tuning!) - The resultant weights - The inference outputs (which also need a license) - The research papers (hop…

Https://opensource.org/ai ... Lots of reasoning has been done on those artifacts

Re: Deepseek R1-0528

#170

Post by macrolime »

yetihehe wrote: post #165
otabdeveloper4 wrote: post #152

Earlier quoted context omitted.

You can fine-tune their weights and release your own take. E.g. see all the specialized third-party models out there based on Qwen. "Open-source" is the wrong word here, what they mean is "you can modify and redistribute these weights".

You can also reverse engineer and modify closed source programs (see mods for games). Weights are like compiled version of source data.

Not legally. That's the difference.
Post reply on HN