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

211–220 of 264 posts

Re: Deepseek R1-0528

#211

Post by kouteiheika »

timschmidt wrote: post #205
kouteiheika wrote: post #201

Earlier quoted context omitted.

> The training process is fully deterministic. It's just an algorithm. Feed the same data in and you'll get the same weights out. No it is not. The training process is non-deterministic, and given exactly the same data, the same code and the same seeds you'll get different weights. Even the simplest operations like matrix multiplication will give you slightly different results depending on the hardware you're using (…

> Even the simplest operations like matrix multiplication will give you slightly different results depending on the hardware you're using That has everything to do with implementation, and nothing to do with algorithm. There is an important difference. Math is deterministic. The way [random chip] implements floating point operations may not be. Lots of scientific software has the ability to use IEEE-754 floats for sp…

> Math is deterministic.

The point is none of these models are trained with pure "math". It doesn't matter that you can describe a theoretical training process using a set of deterministic equations, because in practice it doesn't work that way. Your claim that "the training process is fully deterministic" is objectively wrong in this case because none of the non-toy models use (nor they practically can use) such a deterministic process. There is a training process which is deterministic, but no one uses it (for good reasons).

If you had infinite budget, exactly the same code, the same training data, and even the same hardware you would not be able to reproduce the weights of Deepseek R1, because it wasn't trained using a deterministic process.

Re: Deepseek R1-0528

#212

Post by marci »

kreijstal wrote: post #160
marci wrote: post #143

Earlier quoted context omitted.

When you're trully open source, you can make ethings like this: Today we introduce OLMoTrace , a one-of-a-kind feature in the Ai2 Playground that lets you trace the outputs of language models back to their full, multi-trillion-token training data in real time. OLMoTrace is a manifestation of Ai2’s commitment to an open ecosystem – open models, open data, and beyond. https://allenai.org/blog/olmotrace

you can do these same, except you would need to be a pirate website. It would even be better. except illegal. but it would be better.

That is why the others can't provide stuff like this. RAG/Hallucination check. I just wish Allen.AI models had bigger context, 4k is too small nowadays.

Re: Deepseek R1-0528

#213

Post by hbbio »

lvturner wrote: post #125
hbbio wrote: post #115

Earlier quoted context omitted.

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.

I know of enterprises in APAC now spending millions of dollars on Huawei GPUs, while they might not be as efficient, they are seen as geopolitically more stable (especially given the region). DeepSeek helped "prove" to a lot of execs that "Good" is "Good enough" and that there are viable alternatives with less perceived risk of supply chain disruption - even if facts differ may from this narrative.

Yes, I know them too, I live there!

The hardware is great, CANN is not CUDA yet.

Re: Deepseek R1-0528

#214

Post by therealpygon »

jazzyjackson wrote: post #75
jacob019 wrote: post #68

Well that didn't take long, available from 7 providers through openrouter. https://openrouter.ai/deepseek/deepseek-r1-0528/providers May 28th update to the original DeepSeek R1 Performance on par with OpenAI o1, but open-sourced and with fully open reasoning tokens. It's 671B parameters in size, with 37B active in an inference pass. Fully open-source model.

No sign of what source material it was trained on though right? So open weight rather than reproducible from source. I remember there's a project "Open R1" that last I checked was working on gathering their own list of training material, looks active but not sure how far along they've gotten: https://github.com/huggingface/open-r1

This was simply a mad scramble to prove/disprove the claims OpenAI was peddling that the model wasn’t actually performing as well as advertised and that they were lying about the training/compute resources. Open-R1 has since applied the training to a similar 7B model and got similar results. At the end of the day, no one really cares what the data was that it was trained on and most AI providers don’t always share this either when releasing open source models, and certainly not available for closed source models.

Re: Deepseek R1-0528

#215

Post by Wowfunhappy »

echoangle wrote: post #209
anonymoushn wrote: post #202

Earlier quoted context omitted.

providing a large list of bitrotted URLs and titles of books which the user should OCR themselves before attempting to reproduce the model doesn't seem very useful.

Aren't the datasets mostly shared in torrents? They probably won't bitrot for some time.

...no? They also use web crawlers.

Re: Deepseek R1-0528

#216

Post by mbac32768 »

lvturner wrote: post #124
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

Also worth it for the speed of AI autocomplete in coding tools, the round trip to my graphics card is much faster than going out over the network.

Anyone actually doing this? DeepSeek-R1 32b ollama can't run on an RTX 4090 and the 17b is nowhere near as good at coding as OpenAI or Claude models.

Re: Deepseek R1-0528

#218

Post by mxkopy »

sundarurfriend wrote: post #174
KeyBoardG wrote: post #46

Earlier quoted context omitted.

There will be a video from FireShip if its a big one. /s

Ah FireShip, I forgot that channel existed at all. I asked YouTube to not recommend that channel after every vaguely AI-related news was "BIG NEWS!!!", the videos were also thin on actual content, and there were repeated factual errors over multiple videos too. At that point, the only thing it's good for is to make yourself (falsely) feel like you're keeping up.

Fireship consistently makes some of the most entertaining tech content out there

Re: Deepseek R1-0528

#219

Post by EvgeniyZh »

chvid wrote: post #210
EvgeniyZh wrote: post #191

Earlier quoted context omitted.

> But those benchmarks are in general fairly narrow. They don't really measure the "broader" intelligence we are after. I think a general model that can - finish nethack, doom, zelda and civilization, - solve the hardest codeforces/atcoder problems, - formally prove putnam solution with high probability, not given the answer - write a PR to close a random issue on github is likely to have some broader intelligence. I…

A couple of things: I wasn't trying to invent anything. Just describing what you would obviously have to do if you were to take a "scientific" or "objective" approach: Sound experiments, reproducible, free of financial incentives. As far as I can tell, no one is doing that at a significant scale. Everything is buried in hype and marketing. Now for that broad set of benchmarks (PRs to GitHub, Putnam, Zelda). There is…

> I wasn't trying to invent anything. Just describing what you would obviously have to do if you were to take a "scientific" or "objective" approach: Sound experiments, reproducible, free of financial incentives.

But how is it different from what arena or matharena does?

> That sort of thing is not showing broad intelligence anymore than a person both knowing a chess player and a poet is having broad intelligence.

The claim is that these problems require somewhat broad intelligence by themselves, as opposed to specialization into specific task while unable to do anything else.

Re: Deepseek R1-0528

#220

Post by lvturner »

mbac32768 wrote: post #216
lvturner wrote: post #124

Earlier quoted context omitted.

Also worth it for the speed of AI autocomplete in coding tools, the round trip to my graphics card is much faster than going out over the network.

Anyone actually doing this? DeepSeek-R1 32b ollama can't run on an RTX 4090 and the 17b is nowhere near as good at coding as OpenAI or Claude models.

I specified autocomplete, I'm not running a whole model asking it to build something and await an output.

DeepSeek-coder-v2 is fine for this, I occasionally use a smaller Qwen3 (I forget exactly which at the moment... Set and forget) for some larger queries about code, given my fairly light used cases and pretty small contexts it works well enough for me

Post reply on HN