Ancestors

@geerlingguy Let's pretend for a moment that the US and the CCP are playing a massive game of chess. Doesn't it seem like a hell of a move for China to come out of nowhere and steal the limelight from the US's new Big Tech darling? This disrupts the US player significantly as they were relying on their perceived technological advantage. It also lends CCP (indirectly, by way of DeepSeek) moral superiority by making it all open.

If I were in a Cold War, I'd be playing the player at least as much as I was playing the game.

Just saying: this may just as easily have been timed as an excellent political move in the larger game of thrones.

=> More informations about this toot | More toots from elight@tenforward.social

Written by Evan Light on 2025-01-28 at 04:12

@geerlingguy In the way, it's also a killer move as the Trump administration immediately took the US to a more publicly aggressive political stance with China.

=> More informations about this toot | More toots from elight@tenforward.social

Written by Suzanne Aldrich (she/her) on 2025-01-28 at 05:07

@elight @geerlingguy CCP playing joker poker with US: “I’ll see your [social media ban], and raise you an [AI industry].” ♥️ K♦️ K♣️ K♠️ K🃏

=> More informations about this toot | More toots from suzannealdrich@hachyderm.io

Written by Szipek on 2025-01-28 at 08:03

@elight

I think we are living in a new Cold War era (Cold War 2 is a nice term), but Xi does not even need to be a good strategist anymore, given that his opponent is a simpleton.

@geerlingguy

=> More informations about this toot | More toots from Szipek@mastodon.social

Written by Greg Brooks on 2025-01-28 at 04:42

@geerlingguy its open weights, not open source, right? The Chinese model is much more efficient, and cheaper to make if we believe them.

=> More informations about this toot | More toots from keyboardg@mastodon.social

Written by Alessandro on 2025-01-28 at 05:24

@geerlingguy it was open initially, they gave the world Whisper, to name one. That stuff is great. Unfortunately, the entire open thing was a strategy to attract talent, and nothing more. Giving the power of AI to everyone, to avoid it being only a few companies? That aged interestingly lol

=> More informations about this toot | More toots from AlexTheStampede@maly.io

Written by Shon :CB: on 2025-01-28 at 05:26

@geerlingguy Open-source/weight always wins

=> More informations about this toot | More toots from ShonP40@routing.center

Written by Stacy on 2025-01-28 at 05:49

@geerlingguy - lots of FUD today, not enough "this is cool". First thing that popped into my mind after hearing about it was "if it performs like latest ChatGPT on lesser hardware, what happens if you use it with awesome hardware?"

=> More informations about this toot | More toots from Sknashville@theatl.social

Written by Warriormaster on 2025-01-28 at 06:50

@geerlingguy Is it really open-source though? I went to their repository and mostly saw lfs files that were called ”models-x”. So it could be the same case as with Meta where the model is open-source but everything else isn’t.

=> More informations about this toot | More toots from warriormaster@mastodon.social

Written by CyberFrog on 2025-01-28 at 06:51

@warriormaster@mastodon.social @geerlingguy@mastodon.social that is commonly what 'Open Source' means in the context of AI, even given the training data you'd be unable to reproduce the same model, so the checkpoints and weights are the next best thing to modify and build on top of.

=> More informations about this toot | More toots from froge@social.glitched.systems

Written by dieTasse on 2025-01-28 at 07:26

@froge not really, I think this was finally defined a few weeks ago and in case of AI the source upon which it was trained needs to be also fully known for it to be really open source.

=> More informations about this toot | More toots from dieTasse@floss.social

Written by CyberFrog on 2025-01-28 at 08:02

@dieTasse@floss.social Defined where? the training of AI is non-deterministic, it uses random data at every single step for optimization reasons, the models can't be created without a bit of uncertainty with the final results.

Even if you literally had exactly the same data, and exactly the same software stack, and exactly the same servers - even if Deepseek literally re-trained their own model the exact same way, the result would be a slightly different model that doesn't behave the same still.

That's just the nature of AI training in general, especially for LLMs, so I'm not sure why access to their data would be that important... beyond just having another high quality public dataset I guess (which is broadly good for many reasons unrelated to AI).

=> More informations about this toot | More toots from froge@social.glitched.systems

Written by dieTasse on 2025-01-28 at 08:05

@froge I will try to find it, I heard it somewhere, can be bogus of course. The thing is about showing sources - so a collection of links/articles that were used for the learning of the ai. So in your case thats the part "exactly the same data" in both cases the sources would be that data (+ code of the learning algorithm + things I am forgetting for sure)

=> More informations about this toot | More toots from dieTasse@floss.social

Written by CyberFrog on 2025-01-28 at 08:07

@dieTasse@floss.social yeah, I guess it would be nice, but even with all that data I could never re-create a copy of deepseek tbh

So I think the model weights being open are the most useful thing for most people in general

=> More informations about this toot | More toots from froge@social.glitched.systems

Written by Aurélien Pupier on 2025-01-28 at 22:26

@froge @dieTasse there is the Open Source Initiative foundation which has proposed a definition for Open Source AI https://opensource.org/ai the 1.0 was released in late 2024 (and the definition looks very close to what @dieTasse described)

=> More informations about this toot | More toots from apupier@fosstodon.org

Written by Aurélien Pupier on 2025-01-28 at 22:29

@froge @dieTasse argh it was already mentioned (it was shown in another thread and missed it, sorry for the noise then)

=> More informations about this toot | More toots from apupier@fosstodon.org

Written by dieTasse on 2025-01-29 at 08:02

@apupier

No prob. Better twice than not at all 😊

@froge

=> More informations about this toot | More toots from dieTasse@floss.social

Written by shashankmc on 2025-01-28 at 12:30

@warriormaster @geerlingguy The definition of open source needs to change in the AI world. Releasing weights cannot be called open source IMO.

=> More informations about this toot | More toots from shashankmc@mastodon.social

Written by Jeff Geerling on 2025-01-28 at 13:45

@shashankmc @warriormaster I've just resorted to calling it open model, because that more accurately describes it. Open source means we would have all the code to completely replicate it. Open model means we can download, run, and tweak it.

=> More informations about this toot | More toots from geerlingguy@mastodon.social

Written by shashankmc on 2025-01-28 at 13:58

@geerlingguy @shashankmc @warriormaster On board with that! We gotta start calling companies out, again and again, before releasing weights becomes the new norm and tagged as “open source”.

=> More informations about this toot | More toots from shashankmc@mastodon.social

Written by Warriormaster on 2025-01-28 at 14:41

@geerlingguy @shashankmc I’ve might have missread your original post 😅, sorry for that. Still, it’s a good thing to discuss the fact that everyone called this open-source even though only model is open.

=> More informations about this toot | More toots from warriormaster@mastodon.social