It would also take a lot of money. The sticker price I've seen for the ad business is around $100B, but then you need to pay for data centers and engineers capable of delivering the code extracted from Google. I imagine the overlap in code base at the operational level is non-negligible, as is the data about people
The discovery feed is barely functional . Dosen't learn what i want to see nor does the search . Open protocol and api is bskys best feature that i love but a real algo isn't one git clone away for average user nor it is is financially viable.
You can purchase anti-theft purple die which is used in retail devices for not very much money. We had a problem in my previous building with someone getting through the locked vestibule and stealing packages. Had.
1. Do the Ukrainian get to be a self determining people? Are the free to join any alliance or economic group that will have them? Are they the ones who get to decide how much land they will trade for this?
2. What does the outcome tell other would be invaders? Does the Western alliance defend their ideals or cave to using force to take what you want?
One cannot evaluate the end game of Ukraine independent of the other global hot points in the Middle East and Asia. The autocrats have formed an alliance and seek to undermine democracy wherever and whenever they can.
I think number 1 is the essential point for anyone supporting democracy. Ukrainians, not Russians, decided who they elect as their leaders (1) and which organizations they want to join. It is depressing to hear people in democracies arguing otherwise.
As for point two: Yes it is also massively in the interest of democracies, that Ukraine wins and Europe is strengthened.
I think Stoltenberg summed up it up nicely in (2)
>So supporting Ukraine is not only the morally right thing to do.
It is also in our own security interest.
(1) Queue Russian non-sense about Maidan being a western coup.
Token costs are not zero when you’re running local models, because you paid for the hardware, and you can’t scale inference indefinitely without paying for more hardware.
Ok, but running a 11B model gets things 60% of the time right and consumes maximum of electricity of your machine. Not sure if that makes you product the best. Further video generation is very compute intensive. I guess price will decrease over time but the technical advance will allways be for the smarter model
reply