Live data from Hacker News

Ferret: A Multimodal Large Language Model

github.com

151–160 of 332 posts

Re: Ferret: A Multimodal Large Language Model

#151
post #48

Earlier quoted context omitted.

Why would they dogfood Apple Silicon for training models? Seems like a waste of developer time to me.

Apple doesn’t even sell NVidia cards on their Mac Pros. Are they training it on Linux? I think Apple would strive to be great at all computing related tasks. “Oh, Macs are not good for that, you should get a PC” should make them sad and worried. AI/LLM is the new hot thing. If people are using Windows or Linux, you’re loosing momentum, hearts and minds… and sales, obviously.

> Are they training it on Linux

Yes. Apple has always run their servers on Linux e.g. App Store, iTunes.

And training isn't right now an end user activity but something reserved for server farms.

Re: Ferret: A Multimodal Large Language Model

#152

Presumable because this is Conda none of this can be run on any Apple hardware despite people managing to get M processors to do a bit of dabbling with AI?

> because this is Conda none of this can be run on any Apple hardware conda supports m1? https://www.anaconda.com/blog/new-release-anaconda-distribut...

Did not know that!

Re: Ferret: A Multimodal Large Language Model

#153
post #74

Earlier quoted context omitted.

Don't think that follows. I doubt all their cloud services run on Apple hardware. They make consumer devices.

AFAIK all their cloud services run on x86 hardware with Linux, including Xcode Cloud (which runs macOS in a QEMU VM, in a way that only Apple can technically and legally do).

It has always been x86 Linux going back to the NeXT days.

Today they have large scale Kubernetes ands some legacy Mesos clusters.

Re: Ferret: A Multimodal Large Language Model

#154
post #109

Earlier quoted context omitted.

They're plenty capable of delivering garbage. Certain year models of the MacBook pro were inherently faulty. I've had the displeasure of having two of them bought for me at work. All of Apple's windows software (iTunes, Safari, etc) has been, at best, a barely working port. I'm assuming they are putting a lot more thought and care into it than the touchbar, crappy keyboards and the rest, but I'm also not holding out…

Since they got rid of Jony, it's been great. That's all I'm saying.

Absolutely!

The hardware, that is.

With software, their Come to Jesus moment is still in the future.

(To me, Swift is sort of the Jony correlate on the software side. Doesn't fit perfectly, of course, but very similar "we are perfect who cares about evidence la la la I can't hear you" vibes and results)

Re: Ferret: A Multimodal Large Language Model

#155
post #96

Earlier quoted context omitted.

Siri is really quite dumb. I am confident that a 7B model would be able to provide better responses in over 90% of user queries. I can't even get Siri to reliably set a timer.

Yes, Siri is really dumb. But so is every 7B/13B model out there too.

Eh no, 7B Mistral / Deepseek would certainly almost already be able to function as a super Siri, but probably something closer to PHI-2 + the new MLX apple optimisations. Have you tried those? https://twitter.com/awnihannun/status/1735355067673526360

If trained on an iPhone API + documentation and given a little web access it would blow absolutely everything out of the water.

If they can already create -basic- Python/Swift/JS/rust apps that sets timers, save things, create lists, how's that too dumb for being a Siri replacement? They just have to give it access to an iPhone/Web Api like ChatGPT's code analysis tool.

So if you ask it "hey siri do this, this and this", it will create a script, then run it on the internal API, or fetch an article then work on that etc.

I know it's still logically "dumb" but i'm not trying to play game theoretical scenarios with my phone or do logic tests or advanced math (yet).

Re: Ferret: A Multimodal Large Language Model

#156

>Ferret: A Multimodal Large Language Model What I thought when reading the title: A new base model trained from the ground up on multimodal input, on hundreds to thousands of GPUS The reality: A finetune of Vicuna, trained on 8xA100, which already is a finetune of Llama 13b. Then it further goes on to re-use some parts of LLava, which is an existing multimodal project already built upon Vicuna. It's not really as exc…

Thanks for the summary.

Re: Ferret: A Multimodal Large Language Model

#158
post #80

Earlier quoted context omitted.

I think your expectations are wrong. They sit there in silence with a few leaks here and there and some github projects, people speculate and get all excited about extrapolating those things. Then Apple deliver what works which may or may not be related to it. What they don't do is sell you a lie a year before release then deliver shit (like every other fucking vendor).

They're plenty capable of delivering garbage. Certain year models of the MacBook pro were inherently faulty. I've had the displeasure of having two of them bought for me at work. All of Apple's windows software (iTunes, Safari, etc) has been, at best, a barely working port. I'm assuming they are putting a lot more thought and care into it than the touchbar, crappy keyboards and the rest, but I'm also not holding out…

Hello MacBook Pro 2016 !

Re: Ferret: A Multimodal Large Language Model

#159

Earlier quoted context omitted.

> Rumors suggest they’re gearing up to make iOS 18 an AI focused release. Don't underestimate Apple at disappointing enthusiasts like you and me. We've been hearing many awesome stories about the next thing Apple will do, only to realize their marketing team chose to keep it for future iOS/MBP/iPhone generations to keep the profits high.

My expectation of Apple is that they lurk in the shadows, looking at what others do while perfecting their own thing. Then on the day of release, they'll be a decade ahead of competition. They've done this a dozen times already.

Which dozen times has Apple released something decades ahead of the competition? I'm blanking on 4-12.

Re: Ferret: A Multimodal Large Language Model

#160
post #49

> Usage and License Notices: The data, and code is intended and licensed for research use only.

Boo. But what do we expect from these giants? They're not going to create fertile ground for new competition. The only businesses they foster are those living under thumb and paying tax. I guess I at least hoped for "commoditize the compliments" here. Make Google and OpenAI broadly less special.

it's more likely it's all "stolen" and this is CYA
Post reply on HN