I don't know if you mean that 3oz sounds high or low. It certainly sounds like a normal serving size to me. I normally aim for about 7oz per meal if I'm having pork, but I'm pretty sure that's on the high end. Even then, it's just a little over 100mg of sodium. I've seen the recommended minimum daily sodium for healthy adults to range anywhere between 500mg and 2000mg depending on the source.
And yes, I do mean it breaks up into sodium and chlorine. It makes sense to talk about "salt concentration" in the context of salt dissolved in distilled water, but less so when you have so many other things in solution because there's no straightforward way (to the best of my knowledge, but I'm also no chemist) to map from dissolved ions to the molecules they would've been a part of.
The sad reality of adulthood. It's so hard to get properly immersed in a good story unless you have several straight hours to spare, and those are hard to come by.
It wins in the sense that you still have access to the software and code, and you have the option to either hire someone new to maintain it or switch to something else. Closed source proprietary software only leaves you with the latter choice.
An important component of the cost to consider is how long we expect a company to support a piece of software, and how much it would cost to migrate everything when they drop support. FOSS wins in this regard, especially if you can get a support contact with the devs.
Yeah, salt normally refers to sodium content. Salts separate in solution, so you never have actual NaCl unless you have a solid hunk of salt.
But also, sodium isn't inherently bad for you. Just like most things we consume, too little is bad, and too much is also bad. 53mg per serving is still so far off from what a typical adult needs.
Could be. Classification is a type of problem. LLM is a type of model. You can use LLMs to solve classification problems. There's a good chance that's what's happening here.
development of this model over the years required X TWh of power
This part is kind of hard to measure. When do you start counting? From the first work that informed the research direction eventually leading to this model? From the point where the concept of this final model first came about? Do you split the energy usage between multiple models that came from the same work?
There is a new journal I know of (TMLR) that's becoming a bit more popular in these circles, but I believe they rely solely on volunteers to review rather than asking those who submit papers.
I think it's typical to get a 5 year contract and having to renegotiate a new mortgage at the end of said contract. At least, it is here in Canada. Rate goes up, monthly payment goes up.
Implying we have a way of determining whether an entity is conscious or not. That's the entire point of contention here.