Either way, both tools simply generate the statistcially most likely response from their input data.
Incorrect.
The only statistical element is the reduction of the logits to a token.
The high-dimensional vector within the hidden state isn't anything approaching "most likely".
They are statistical parrots after all.
Stochastic. And no, they aren't.
LLMs are quite capable of understanding, as it is defined, as long as you remove the anthropocentric requirement.
Forget the startups an what they're trying to sell.
The science (papers) on the topic are quite real.
Oklahoma: the answer to the question "where is the worst place we can put the Indians?"
I'll give those fuckers credit- they nailed it.
Median wage in California is 10th place, not of the world, but of US states and territories. It's between Minnesota and Colorado.
False, because the omitted information is critical to the assertion.
It makes the following statement:
California's GDP has nothing to do with the people of California. A bunch of big tech companies and media conglomerates have their HQ there and that's where the income of all their US revenue is registered. Their foreign income is usually registered in some tax haven. This talk about agriculture is nice, but at $60 billion it is barely relevant for the so called 'GDP' number.
Fucking absolutely laughable.
As you point out- agriculture is only $60B.
And yet, every big tech company and media conglomerate combined are a small fraction of California's GDP.
Why do your argument apply for one and not the other?
Because the fact is- It has everything to do with the people of California, and their high median income.
Staff meeting in the conference room in 3 minutes.
