Z4rK

joined 1 year ago
[–] Z4rK@lemmy.world 0 points 3 months ago (1 children)

To be honest, I’m not sure what we’re arguing - we both seem to have a sound understanding of what LLM is and what it is not.

I’m not trying to defend or market LLM, I’m just describing the usability of the current capabilities of typical LLMs.

[–] Z4rK@lemmy.world 0 points 3 months ago (3 children)

It goes a tad bit beyond classical conditioning... LLM’a provides a much better semantic experience than any previous technology, and is great for relating input to meaningful content. Think of it as an improved search engine that gives you more relevant info / actions / tool-suggestions etc based on where and how you are using it.

Here’s a great article that gives some insight into the knowledge features embedded into a larger model: https://transformer-circuits.pub/2024/scaling-monosemanticity/

[–] Z4rK@lemmy.world 1 points 3 months ago (1 children)

To be fair, I think he is mostly endorsing the concept of the implementation, lined out in his seven points, not the actual implementation since it isn’t available yet.

[–] Z4rK@lemmy.world 0 points 3 months ago (5 children)

He sort of invented it, so you have to think he’s commenting on the concept here, not the implementation.

I have tried a lot of medium and small models, and there it just no good replacement for the larger ones for natural text output. And they won’t run on device.

Still, fine-tuning smaller models can do wonders, so my guess would be that Apple Intelligence is really 20+ small and fine tuned models that kick in based on which action you take.

 

cross-posted from: https://lemmy.world/post/16391311

Andrej Karpathy endorses Apple Intelligence

Actually, really liked the Apple Intelligence announcement. It must be a very exciting time at Apple as they layer AI on top of the entire OS. A few of the major themes.

Step 1 Multimodal I/O. Enable text/audio/image/video capability, both read and write. These are the native human APIs, so to speak.

Step 2 Agentic. Allow all parts of the OS and apps to inter-operate via "function calling"; kernel process LLM that can schedule and coordinate work across them given user queries.

Step 3 Frictionless. Fully integrate these features in a highly frictionless, fast, "always on", and contextual way. No going around copy pasting information, prompt engineering, or etc. Adapt the UI accordingly.

Step 4 Initiative. Don't perform a task given a prompt, anticipate the prompt, suggest, initiate.

Step 5 Delegation hierarchy. Move as much intelligence as you can on device (Apple Silicon very helpful and well-suited), but allow optional dispatch of work to cloud.

Step 6 Modularity. Allow the OS to access and support an entire and growing ecosystem of LLMs (e.g. ChatGPT announcement).

Step 7 Privacy. <3

We're quickly heading into a world where you can open up your phone and just say stuff. It talks back and it knows you. And it just works. Super exciting and as a user, quite looking forward to it.

https://x.com/karpathy/status/1800242310116262150?s=46

0
submitted 3 months ago* (last edited 3 months ago) by Z4rK@lemmy.world to c/technology@lemmy.world
 

Actually, really liked the Apple Intelligence announcement. It must be a very exciting time at Apple as they layer AI on top of the entire OS. A few of the major themes.

Step 1 Multimodal I/O. Enable text/audio/image/video capability, both read and write. These are the native human APIs, so to speak.

Step 2 Agentic. Allow all parts of the OS and apps to inter-operate via "function calling"; kernel process LLM that can schedule and coordinate work across them given user queries.

Step 3 Frictionless. Fully integrate these features in a highly frictionless, fast, "always on", and contextual way. No going around copy pasting information, prompt engineering, or etc. Adapt the UI accordingly.

Step 4 Initiative. Don't perform a task given a prompt, anticipate the prompt, suggest, initiate.

Step 5 Delegation hierarchy. Move as much intelligence as you can on device (Apple Silicon very helpful and well-suited), but allow optional dispatch of work to cloud.

Step 6 Modularity. Allow the OS to access and support an entire and growing ecosystem of LLMs (e.g. ChatGPT announcement).

Step 7 Privacy. <3

We're quickly heading into a world where you can open up your phone and just say stuff. It talks back and it knows you. And it just works. Super exciting and as a user, quite looking forward to it.

https://x.com/karpathy/status/1800242310116262150?s=46

[–] Z4rK@lemmy.world 1 points 5 months ago

All my devices and networks are constantly on VPN, it’s just basic obscurity, and it’s such a hassle posting on Lemmy now - my activity is probably down 90%.

Could you please consider allowing basic text posts and / or comments from behind a VPN? You could even disallow all markdown - that’s a decent trade off for my sake.

But please don’t just block all VPN usage if there are options you have not explored yet.

 

Proton announced some new optional way to swipe automatically to next email, that you could turn on in settings.

I can not find any setting to turn this on or off. However, since it was announced, the default swipe changed. When I open an email, I can no longer swipe back to the inbox. I can swipe to next message if it is not the last.

This brakes my primary way to navigate, and it was not announced, and I can’t change it back. It’s extremely annoying.

It’s also different from how Mail, Outlook and Gmail works, which all have the swipe action proton used to have, where you swipe back to the inbox.

[–] Z4rK@lemmy.world 0 points 6 months ago (1 children)

It seems that we’ll get up to 10x soeedups with v4: https://tailwindcss.com/blog/tailwindcss-v4-alpha

[–] Z4rK@lemmy.world 16 points 6 months ago

But that’s literally what you do, you build a starter factory to help you build a mid tier research factory which can then help you build a proper factory that can support megascale production

[–] Z4rK@lemmy.world 2 points 7 months ago

Cool, now do it over Ping!

 

Hi everyone! We're incredibly excited to announce that we're launching a beta of Finamp's redesign today. This is a major update to the app, and we're looking for feedback from anyone willing to try it out before we roll it out to everyone.

The beta is a work-in-progress, there are several new features already, but we will be adding more features over time.

Looks very nice!

[–] Z4rK@lemmy.world 1 points 7 months ago

Just try it? You’ll get a free trial, use a fake email, no payment details or use info needed as far as I can remember.

I believe also https://kagi.com/fastgpt is free to use.

[–] Z4rK@lemmy.world 3 points 7 months ago (1 children)

It’s his distinct style that he has kept since before most of us tried anything like ChatGPT etc, so personally I welcome his consistency.

[–] Z4rK@lemmy.world 1 points 7 months ago (2 children)

It’s really cool, but the example doesn’t produce any sensible output? If you have created something like this, why wouldn’t you have your demo output something sensible like Fibonacci or 1337 or whatever.

[–] Z4rK@lemmy.world 2 points 7 months ago

You have also implemented a ban on any contribution from a VPN without announcing it. It has been very annoying to figure out.

Please do announce these kind of changes and limitations you implement publicly unless you absolutely cannot.

 

I love Proton Privacy as a company, and most of their products.

However, I hate their current SoMe campaign of just ranting and bashing on every other company out there. It’s so negative.

Is more negativity really what we need? Can’t you just be positive and talk about all your good stuff - are you 100% sure the only way to grow is to do negative campaigns on everyone else?

I’d really love for you to be different ❤️

 

Is there really no way to view a week and / or work week on the Proton Calendar on mobile (iOS)? Who is only interested in either a day or month view? It’s just so weird, I feel I have missed a setting somewhere obvious.

Even the new beta desktop app provides a week view, even though a work week view is still missing there too.

 

I'm fine with the Proton Mail desktop client being an Electron app, but it still need to use desktop-based interactions. For example, when right clicking on the inbox, I expect to see options to mark all as read etc. - not an Inspect Element menu (that actually works and opens up devtools inside Proton Mail).

And to those that can't cope with 3981 unread emails - I've just imported from Gmail, and a lot of them appeared as unread, which is why I'm now looking for a way to mark all as read.

 

Has Proton added this yet? Most competitors I’ve used always had it. Just go to some url and it will tell you if you’re on Proton.

So far I’ve just been told to check my ip - this is cumbersome and often impractical. I don’t want to break my connection to compare, I’m on my mobile and VPN is on the WiFi router so if I just disconnects from WiFi that will send me to my carrier network so I can’t compare with the WiFi, I may have VPN also on mobile and want to verify that I stay on proton both on and off WiFi, etc.

I think first of all, there should be a clear message when visiting proton sites if I’m on their vpn or not, at least on ProtonVPN, and there should also be a separate url like test.proton.me that tells me just this and is easy to curl etc.

7
submitted 1 year ago* (last edited 1 year ago) by Z4rK@lemmy.world to c/protonprivacy@lemmy.world
 

On Reddit, someone was mentioning information and questions in a Drive for macOS channel somewhere, but i can’t find it.

Are there other channels / forums than Reddit / Lemmy where such discussion are taking place? X? - I’d hate to have to reinstall it.

Edit: Also, I guess, this could stay as a thread for the topic here on Lemmy.

 

I just released Pixel Pals 2! 🎉 With iOS 17 you now have a FULL virtual pets experience where you can add and battle friends, and play full games, like PixelQuest, 2048, and Eternal Stroll, all right on your literal home screen! (Plus fidget spinners, mech keyboards, and more!)

 

At Proton, we’re always working on new and innovative ways to protect the privacy and data of the Proton community.

Yeah, I guess that’s nice. I do like Proton.

I’d love for Proton to focus on completing some current services and make them actually usable though.

 

Just in case you wondered how many photos you really can have in your Apple Photos library, I can report that I have so far added 1 000 264 Photos and 10 242 Videos without any issues.

I’m fairly impressed and happy about it since all I could find was that it should support up to 100.000 photos, with a few reasoning about the limit being increased to 300.000 on modern hardware.

1 000 264 Photos, 10 242 Videos in Apple Photos

I’m running this on my MacBook Pro M2 Max with 64 GB ram.

Most formats gets converted to HEIC and HEVC on import, which are staggeringly effective compared to their original formats. The whole library file still only takes up 1.7TB, which is much less than expected. The original source on my NAS is around 5.6 TB.

Edit: Maybe I should add that I do not recommend this, and view it as an experiment for now. I’m still importing data. If it’s still stable and performant after a year and some OS updates then I can start recommending it.

view more: next ›