It appears we will have four Chinese models with capabilities ranging between Fable and Opus 4.8 by the end of this month.
Artificial Intelligence thread
- Thread starter 9dashline
- Start date
Distillation is American cope talking point, Gemini and Grok can download open weight models in full to distill to their hearts content with unlimited compute to train, Elon even spent $60B buying Cusor which trains off K2.5, wheres their K3?Yup, agree completely.
People seem to overlook a basic point, especially those in the West.
Every single LLM that is competitive was provided to the public by a for profit corporation.
Simply put, AI in the form of LLM currently, is still a business.
The other important point that people do not talk enough about, is distillation, because everyone does it.
So, what does that mean? That clearly suggests the models will converge in terms of performance, because everyone is copying everyone else, to stay ahead.
So, therefore, if the AI current LLM product is always converging in capabilities, all that matters is price. Because it is still a business.
This reminds me of that expressions from the lawyers. This is what they say.
If the law is on your side, pound the law. If the facts are on your side, pound the facts. If neither the law or the facts and on your side, then pound the table.
Those American AI people, they are starting to pound the table, when they start to suggest is Chinese AI is really Chinese Yellow Peril AI.
That is not going to work, because this LLM is still a business, and too many mainstream and famous American companies are embracing Chinese open source AI due to the cost factor.
After all, it is still a business.
That is why I think we read too much bullshit all the time in the media.
If distillation is a normal practice in the development of AI models, then what is the outcome there? Convergence!
If frontier models are defined by compute power, then what is the outcome there? Most costs! Due to brute force methods!
Can LLM brute force its way to AGI? Notice how no one really talks about AGI anymore. It was a smokescreen all along.
Chinese labs converge most likely because new ideas from Chinese acedmia are deseminated at the same time, but not to the west, and being open weight theres not a much gate keeping so people from labs talk to other labs a lot more. This has always been the main befenit of a collaborative ecosystem.
That and theres a no zero chance model release policy and timing are cordinated by a higher power, after all Xi doesnt just give a speech on open sourve at an AI conference to give suggestions.
At end of the day America's problem is they have a severe lack of "human compute", wipipo cant do maf
You can’t complain about China stealing your tech and that Chinese students studying in the U.S. are all SeeSeePee spies while simultaneously lamenting that Chinese talents aren’t staying in the U.S.
I think that should be the question we should be asking ourselves in the first place. What makes the US so attractive to global talents(China included)?
It is wealthy, has innovative culture, is receptive to foreign capital until recently, and when you strike it rich you can pretty much do what ever the hell you want to with minimal consequences.
Hmmm…agree, but looking at some people view point here, you will think the US is in a worse shape than Russia.lol. I think nationalists from both sides tend to exaggerate the other side weak points while ignoring their many strong points .
The issue here is that the U.S. ruling class is not only taking the advantages for granted but doing its darndest to undermine most of them! Take Chinese EV/battery companies that want to invest in factories in America for instance. That’s literally opportunity for forced tech transfer handed over on a golden platter. What they do instead? Call them spy cars/batteries and ban the hell out of them. Same thing with Chinese students coming to study in the U.S.
I think the officials across the political aisle are not only entitled/stupid but non-technical and run more off vibes/cope than sound decision making. I think that if they strictly stuck to playing golf and engaging in insider trading, the nation will be better off because at least no one is actively sabotaging.
Super interesting
Not even HUGGING FACE itself can get access to un-nerfed Fable/5.6 Sol for defensive cybersecurity use. Had to use self-hosted GLM 5.2.
This incident show both that Chinese models are close, if not already caught up, to the best Western models for cyber-security, and that the proprietary US firms are going to be dealing with more and more issues (cost, nerfing, data privacy, govt interference) on all sides
Not even HUGGING FACE itself can get access to un-nerfed Fable/5.6 Sol for defensive cybersecurity use. Had to use self-hosted GLM 5.2.
This incident show both that Chinese models are close, if not already caught up, to the best Western models for cyber-security, and that the proprietary US firms are going to be dealing with more and more issues (cost, nerfing, data privacy, govt interference) on all sides
remember, the biggest fear a while back is US can use Mythos to launch cyber attacks on China & China has no responses to it. Kimi K3 consistently comes out to be very capable on cyber security.
well, keep in mind Mythos 5 was released in early June. That is Chinese labs target. At least in terms of cyber warfare. And Anthropic continues to improve it. Goal of Kimi K3.1 and GLM-5.5 has to be next Fable and GPT-6.0 rather than what's available right now.
distillation is a real thing. When Chinese labs didn't have enough data (because every dev in China were also using Claude Code), they had to distill, but now that people are buying their coding plans and also their models are good enough for distilling new models, this is not needed.
Can We please move on from this off topic stuff?