You prope the trained model, delete/kill the weights and than you are done.

On federated learning, you just make sure to keep this mechanism in the right stage of your pipeline

You are just joking right?

From a skill and trust point of view, Google is doing a lot better than apple will ever.

Including ondevice AI

The data source is linked and is based on the ARM Datacenter Energy prediction.

But i don't think its too far fetched.

The compute needed for digital twins, simulating a whole army of robots than uploading it to the robots, who sitll need a ton of compute, is not unrealistic.

Cars like Tesla have A TON of compute build in too.

And we have seen what suddenly happens to an LLM when you switch the amount of parameters. We were in a investment hell were it was not clear in what to invest (crypto, blockchain and NFT bubble bursted) but AI opened up the sky again.

If we continue like this, it will not be far fetched that everyone has their own private agent running and paying for it (private / isolated for data security) + your work agent.

Thats actually a really good price.

The mac book air with the M chip was absolutly a steal already. I'm surprised by this.

Is that some thing to allow cheaper MX / Arm architecture in DCs? Is getting Apple affordable oO?!

I have to think economies of scale are coming into play for Apple. They can cut deals for chips and other components at a scale no one else is really capable of and they have the luxury of being able to pay up front in advance if they need to.

I mean yes for sure but apple always had a high margin. I'm pretty sure they did not suddenly become 'nice'.

Would be interesint to know though if the margin stayed the same and they literaly just save a lot of money.

It could make sense for you to consider an EV with bi-directional charging. For off-grid solar, you can charge your truck/car/EV for free or/and extend your off-grid battery by using your car to de/uncharge it for your home.

A 100kWh battery can help you out when you come back home if you know you can charge it the next day.

You don't have to, it makes it a lot easier that you can do it.

You can also charge supercapacitors and uncharge it very fast if you like.

You can use the whole capacity of a charging park for just one car (if not a lot of cars are there).

You could also do a more classical gas station setup instead of a charging park. Cooled cable, 1-2 proper ports on the bottom etc.

I like this 'my handwriting is write only'.

when i was 18, i looked at notes from my new coworker / manager and asked him how he learned to write that nice. He told me that he struggled writing cursive and just started writing letter by letter (print letter style?).

I changed my 'font' on that day and suddenly i was at least able to read what i wrote!

Interesting.

We grew up learning to write separate letters.

Cursive was taught only very briefly as a way of telling us tgis also exists but don't bother too much about it.

Nevertheless, cursive is where the speed is. University times I was writing at the speed of the professor chatting and still having it halfway readable even for others - even though the looks of it were tending to the Arabic (I'm Latin). Good times, now I can barely scratch a shopping list before I break my wrist.