- cross-posted to:
- technology@beehaw.org
- technology@lemmy.zip
- technology@lemmy.ml
- cross-posted to:
- technology@beehaw.org
- technology@lemmy.zip
- technology@lemmy.ml
cross-posted from: https://lemm.ee/post/55428692
You must log in or register to comment.
I’m no expert but with musk likely doing a hostile takeover of OpenAI I’m down with this.
deleted by creator
True that. It’s the lesser evil. That would probably even be the case if it wasn’t just a deal for the Chinese market.
Since people don’t click through and read articles, it should be pointed out that this is specifically for the Chinese market in order for Apple to try and get back market share in China where they’ve been falling behind Huawei and others.
The bloomberg article was paywalled for me, and I didn’t see that in OP’s link, which is like 3 sentences long.
But honestly, they should’ve used Alibaba/Qwen for their other markets, too.
they should’ve used Alibaba/Qwen for their other markets, too.
Which would get them banned from all government use. Smart call.
Which is so stupid, as it’s more secure if Apple (or the government) just hosts it themselves, assuming its not run on-device.
For context, Alibaba is behind Qwen 2.5, which is the series of LLMs for desktop/self-hosting use. Most of the series is Apache licensed, free to use, and they’re what Deepseek based their smaller distillations on. Thier 32B/72B models, especially finetunes of them, can run circles around cheaper OpenAI models you’d need an $100,000+ fire-breathing server to run… if OpenAI actually released anything for public use.
I have Qwen 2.5 Coder or another derivative loaded on my desktop pretty much every day. And because its open, the whole community contributes back to it.
So… Yeah, if I were Apple, I would’ve picked Qwen/Alibaba too. They’re the undisputed king of iDevice-sized LLMs at the moment, and do it for a fraction of the cost/energy usage of US companies. They’d have to be stupid to pick OpenAI or Anthropic, and even stupider to reinvent the wheel themselves (which they already tried. Their local “small” LLMs are kinda crap for their size).
Deepseek R1 is currently the selfhosting model to use
Some of the distillations are trained on top of Qwen 2.5.
And for some cases, FuseAI (a special merge of several thinking models), Qwen Coder, EVA-Gutenberg Qwen, or some other specialized models do a better job than Deepseek 32B in certain niches.
How do you Apache license a LLM? Do they just treat the weights as code?
It’s software, so yeah, I suppose. See for yourself: https://huggingface.co/Qwen/QwQ-32B-Preview
Deepseek chose MIT: https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
For whatever reason, the Chinese companies tend to go with very permissive licensing, while Cohere, Mistral, Meta (Llama), and some others add really weird commercial restrictions (though this trend may be reversing). IBM Granite is actually Apache 2.0 and way more “open” and documented than even the Chinese tech companies, but unfortunately their models are “small” (3B) and not very cutting edge.
Another note: Huggingface Transformers (Also Apache 2.0, and from a US company) is the reference code to run open models, but there are many other implementations to choose from (exllama, mlc-llm, vllm, llama.cpp, internlm, lorax, Text Generation Inference, Apple MLX, just to name a few).
Reminder, Apple was always going to use a Chinese service for China. ChatGPT is banned in China.
Third party model integration is pretty dumb in Apple intelligence. When Apple’s private model hits a dead end, it asks if you want to throw the prompt into a bigger model.
Eventually, like with search, people will be able to select the default model that they integrate with.
A match made in Hell.
Apple’s outsourced just about every other aspect of its business to Chinese tech companies. I don’t see why this would be different.
Ohh wow… Will Sam altman ask President musk to sanction the Chinese terrorists?
Musk might do it himself, since it technically competes with X AI.
Grok is a laughing stock in LLM land, though; all those H100s he bought are completely wasted. And he straight up lies about their open source approach.
